@@BloodyMobile it really is. slows progress down. but to a certain extent, if its perfect you dont have to go back and revise it in the future, but then again you never know what problems you will run into anyway that you didnt foresee. but getting something that works now, VS is perfect later is sometimes good enough. of course design choices make it more difficult because you might not be able to come back to 'make it perfect later' if its a fundamental design choice. I guess its best to aim for 'roughly perfect'. good enough, and modular so parts can be adjusted as needed.
in the audio example i could hear it even at 1.2 MS off. But i suppose what matters is 'when listening to music, if the notes are off by 1.2ms, would anyone notice?' and i suppose the answer is no. How many musician play that tight to be 0ms off?
@@jessiejanson1528 I think the majority of people can not though. For me it simply sounded like "one" sound until 20 ms, and only because I was looking for a difference. Without that I might have noticed it even later. And I /do/ have an above average good hearing, but apparently not for this kind of stuff.
Yeah that was weird, "3.9ms isn't good enough for precise timing" - wait what, you barely could hear anything Haas-effecty in the beginning, the very moment this kind of timing issue presents itself to the listener, you're already surpassing perceptual thresholds of 99.999 % of all listeners out there. Easy to prove too, but yeah, this went against everything he established in the beginning. I mean, whatever floats your boat and all that, but that's truly overkill.
Keep in mind that according to a journal article testing musical "tightness", pro drummers have a standard deviation of 6.6ms for their dominant hand and 9.5ms for their off-hand. Being a drummer for Martin would be stressful lol.
Also the concept of a machine that makes music by dropping marbles is only interesting because of how dumb it is. Does he think this will ever become an actual instrument? It is and always will be a novelty that's enjoyed because it's dumb, its not the new drum kit and if thats what he wants then he should just use MIDIs
It's the stackup, not an individual feature, you have to hit. As he observed around 3ms or so of delay starts to be noticable and 1/2 of that error is in just dropping the ball you have very little slop allowable for the rest of the system. Things like the timing of the programming wheel, linkage delay, the angle of the instrument, where it hits on the instrument, etc adds to this error greatly and if you want it to be a good sounding instrument it needs to be tight tolerance across a number of components to get the sort of drop accuracy it needs.
@@MikeLikesMaking It won't really matter what the stackup is if the marble machine is the source of the MIDI clock, then all the other digital instruments will depend on it for their timing.
The thing that Martin doesn't explain that clearly is that Standard Deviation isn't the same as the gap between notes. 1/300 things is more than 3 standard deviations out from the mean, and that goes in both directions. So if you have a bunch of marble gates going, notes can routinely be 6x that deviation from each other. Not to mention that this is only one of potentially dozens of sources of deviation (at least in the final machine). It's possible he's overdoing it, but reducing variability in each individual piece is what you need, because it can add up, and it'll be a lot harder to diagnose in a fully-built machine than in component parts.
The missile knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation.
Yeah in many ways this project has turned into Martin doing a PhD without a supervisor, which is kind of a terrible idea. One of the important roles such a person fills is telling you when not to go down the rabbit hole.
Remember that the Achilles heel of the MMX gates was marble pressure! Timing is important, but your number one priority when designing should be reliability, and your testing setup should replicate the stress the gate will be under on the finished machine.
Yes. It might be a good idea to estimate the pressure the marbles upstream is putting on the gate by making some mass that can push against the marbles/gate being tested to simulate a full track
on thing at a time if you want to mash everything together from day 1, you will never be able to engineer anything good. one obvious reason: cannot test this. some problems will emmerge only on final stages of production and some of them will need you to disassemble (not literally if design is good) the whole thing to debug. and the lack of testing opportunities will get you to build the whole thing from anew. look at marble machine X if you have doubts
As a drummer I can safely say you don't need to get 0ms accuracy of timing. You're not making the world's most over-engineered midi roll, you're making an instrument.
@@jameswatson4429 As far as reliability, I feel like the second machine was pretty much within range of reasonable expectations, he was dropping what? like a marble or 2 per 10k drops? totally reasonable. His only real issues were the limitations from the bottlenecking that was occurring with the dividing gate. I would liken that to the limitations you’d find on any instrument though. A Violin has 4 strings and no frets. A Guitar has 6 strings and imperfect intonation, the second marble machine was a bit limited as to the number of simultaneous inputs per second. perfectly reasonable when you think about it.
As an acoustic engineer designing audio systems and loudspeakers (and also a musician), I totally agree with this as well. A rule of thumb is about 20ms as worst case. Everything within plus or minus 5-10ms is more than good enough, and probably much better than most drummers. At this point this feels heavily like creating a problem, a bit like pixel peeping on a photo. A photo that has beautiful scenery, composition, story and feeling.
Please note that this is only the marble gate and that inaccuracy in the machine will add up. 4 ms isn't noticeable, but if you lose that amount a couple of times you end up with audible delays. So he might seem crazy in designing a gate that is super accurate, but good design now may lead to better results in the end.
@@EaglePicking humans are very inaccurate. I don't think most musicians can keep withing 10 ms of timing for entire songs. it's just insignificant. Also the original marble machine sounded great but I'm pretty sure it isn't within 10ms
@@Anonymous41726 that is in the most important part in the machine. And deviations add up, and esp. if you play many notes together and fast. And we STILL have not counted the deviation in other parts, like programming wheels, which could need this extra space within what we can have before it can be detectable. So yes, it is still important to save space to other parts.
@@AndersJackson I see your point with adding up with other parts but the deviation of it hitting the ground does not effect the timing of the release switch.
I’m torn between celebrating Martin doing what I had deemed impossible, and crying because I see reflections of my own perfectionism ocd in him I wish him all the best!!
I couldn’t help thinking the exact same thing, but even during machine 2 livestream era lol… just didn’t comment because thought I’d sound crazy. So easy to get stuck on details, but at least he’s picking his details this time lol.
i would focus on modularity, reparability, and the challenges of live performances / logistically moving the machine if you plan to tour with it. or, if it's to be a studio instrument, i guess continue down the perfectionist's path either way, thanks for showing us your progress and your process. it truly is inspirational on many levels. i love your test setup, and the way you've turned your DAW into precision timing testing + the graphs you're making, etc. i'm just excited to see you making music, however it winds up. love that you can chase what makes you happy, and seeing you explore your curiosities, and passions, and sharing your learnings about learning is worth more than any world tour in my eyes. you're already on a world tour with every video!
+1 for Modularity. Each instrument could be designed as self-contained, with its own marbles. Portability, massive reduction in the probability of terminal failure for the whole project, and much easier iteration as designs evolve.
@@revcofe That... well the idea of /not/ mixing them together and /not/ having to separate them entirely avoids the issue of the marble divider altogether. That said its main issue is that a perfect one needs too many dividers to split them up and its too slow. If you were to combine both ideas 'dont mix up all the marbles, but mix up some of them' and then 'use several small marble dividers, then you only need to split the stream for each one 1-3 times and it takes up much less space while retaining some of the idea of the art aspect.
Something hurts in a weird way seeing a video of MMX at the end playing perfectly good drum beats after a video chasing an impossibly small and unimportant margin.
The tighter he can get the marble drop the more complex the music he will be able to make. Hopefully chacing these little details will lead to some amazing music though it may be a long way down the road!
Music complexity has FAR more to do with gate speed and the programming wheel. As usual he has the wrong priorities, hence why he has literally given up on the number one priority, making sure the machine has personality.
Not to mention that actually using the MMX would also have brought lot of learnings (on top of really good music). For instance, what about the vibrations on stage coming from the accoustic / the returns ? Maybe the stdev due to this phenomena is already far bigger than 3ms 😢😢😢 but I will still follow and support Martin… even if I think it was a mistake to drop the MMX like this…
Martin, when you combine programming wheel inaccuracies, linkage to gate inaccuracies, and mounting of the whole assembly inaccuracies + vibrations and everything.. you'll be lucky if you get below 20ms across different gates. And 20ms is very good. Just keep that in mind when you invest 3 times the effort for 10% improvement. You'll need to compromise a bit if you're ever to play anything. Other than that, keep up the good work.
Sound travels at a speed of 330m/s. 10ms are equivalent to a distance of 3.3 meters. A band can play tight music well beyond those distances. Psychoacoustics mean that a human is absolutely incapable of even hearing a difference of 10ms around a beat. You're on a goose chase.
This!! the millisecond chase is completely insane. the numbers he is upset about is equivalent to moving a bit away from a speaker... Has he tried to measure the standard deviation of him self playing to a beat... might end up chopping his arms off.... he is upsett about a _deviation_ of 2 ms!? that is completely insane to be upset about....
He´s still chasing his tail. As long as the goal is "perfect" this project is, again, going nowhere. Sometimes a solution is not just good but good enough. Until Martin accepts this first principle there won´t be a working marble machine.
Well, let's be honest, if you're talking Standard Deviations, it's accuracy not precision. Martin wants all of his notes to land predictably and reliably; @imrahil is correct that there is a finite amount of precision that can be gained - here Martin goes for 1ms precision, which seems a little much to me, but thank goodness he's not after nano-second precision! However, as evidenced by the clock drift in his arduino, if he has many arduinos drifting apart at different rates without some ability to monitor, control or adjust for their *inaccuracy* then different harmonies and rhythms will quickly become out of sync.
Some suggestions for more accuracy: - cooling the system with liquid nitrogen to reduce thermal noise - dropping the marbles in vaccuum only, as air density, pressure and movement may vary the drop time - with a vaccum acoustic instruments will be out of question, so replacing the instruments with triggers will help - triggers have a built in delay due to processing time, so it would be best to use software which knows what sound the marble wants to play at a given time and pretime the sound before the marble hits the trigger - also cooling the triggers with liquid nitrogen might help with thermal noise and delay - using optical fibre instead of copper wire would help reduce delay in signal processing - liguid nitrogen only gets you down to like 100K which if far from perfect, so for more accuracy you could try liquid helium - friction with the marble gate definitely causes random variance, so it would help to remove mechanical friction by suspending them in strong magnetic field - again liquid helium would come in handy - speed of sound varies across a room due to density fluctuation, so it would also be good to not play the sound via speakers. Maybe all listeners should wear headphones with optical fibre connection (all cables same length, helium cooled ofc) but you’d still get at least a few picoseconds of variance Anyone else have any ideas? Human hearing resolution is 5ms at best, so why stop at 0ms stdev? I would keep going until femtosecond accuracy.
I know this is satirical, but I think the suspension in magnetic field would be pretty bad for precision. Magnetic couplings tend to have a decaying resonant wobble when sharp transitions are involved. In the spirit of your post, I suggest not suspending the marbles with magnets, but keeping them in an acceleration loop which spins the marbles around and just lets go when the marble's velocity is parallel to gravity.
I think the idea about the trigger delay compensation is good but truthfully if you look into it you realize that even with the best estimations the programming wheel may still be off. I have an idea to solve this: -replace the programming wheel with a digital file, which is only limited by the clock accuracy of a computer. -replace the marbles with some kind of programmable sequence to put into the digital file. -play the resulting file from the computer. I think this will solve all the potential problems Martin has been experiencing!
I love you so much you have no idea!!!! I am literally DYING OF LAUGHTER 😭🤣😭😂 like 64% of what you said went over my head - scientifically speaking, but if my guess is correct you’re having a slight troll moment?? Honestly either way; if your comment was meant in 100% earnest, or if you were trolling, you TOTALLY MADE MY DAY WITH THIS!! Im going to re-read your comment later when I’m having a bad day 👍😂 That’s brilliant stuff 👌
I love checking in on this project once in a while. It's completely impractical, utterly insane, there's absolutely no reason for this to happen - except that one man has an incredible vision and the drive to learn whatever is necessary to make it happen. I am so proud of you. Thank you for sharing all of this with us and taking us along with you on your journey.
This concept of as tight as possible martin has been chasing is an interesting one - one at the core of his engineering challenges. I wonder if he ever did a study of popular music to analyze what natural standard deviation is to get comfortable with inaccuracy. My guess - after listening to Christian Hand’s The Twitch Sessions (you should listen if you haven’t) is that even the best musicians probably have significantly more standard deviation than what he is struggling to engineer away.
Yeah I'm no musician but I think slight inaccuracies just makes music sound natural. Like even when producing music digitally people will shift things slightly
If he's planning on dropping the marbles with a piston and an arduino, why not just position the pistons to play the vibraphone directly? We're losing the mechanical aspect and I feel Martin is chasing a perfection that will be unattainable. Perfect is the enemy of good
Didn’t he kinda did that in the last video, when he showed different delays from the right and left channel and demonstrated at which point it gets noticeable, in so showing the „acceptable“ range
@@bazahaza Simone Gertz made a musical bubble wrap machine in a pretty short time. Sure it's not as complex as the MMX. But there is something to be said for settling for "good enough".
To be fair Sam has the same reliability issues with the gig gear. He is remaking modules using industrially printed PCBs. And it's electronics without moving parts!
@@doBobro yeah, I know that MMX is complex af but in order to use it on world tour it needs to be assembled and disassembled twice a week, I'm not a mechanical engineer but I think some clearances and tightness will drift pretty fast. I think the best way is to make it "good enough" and travel with enough spare parts to build another 2-3 machines. And maybe emulate world tour shenanigans to iron out the weakest links that will break first.
In 3 more years, he gives up on a 90% finished MM3. Four years afer that, MM4 bites the dust. MM5 on the other hand, gets finished surprisingly quickly as it will just a MIDI player.
Wobbling + inaccurate gates will result in worse accuracy than wobbling alone The purpose is to minimize the total deviation after all the systems add up
this is exactly the reason to get accurate first, otherwise you're just stacking inaccuracies together. "Don't worry about inaccuracies because it'll get worse anyway"
@@jslavertu and if one part is late and the next is late, you're 10ms late. It's +-10ms, you can't expect them to cancel out all the time. Sometimes both will stack early and other times late
There's something about when you dig deep enough to get "above the problem". Don't let the accusation of perfectionism get you down. It's clear to me that you've transcended endlessly perfecting escapements and now you are seeing the true nature of the problem. Enlightenment can be found anywhere, thank you for showing how you found it by loosing your marbles.
This type of insane obsession with detail is starting to remind me of Miura-sensei, the mangaka of Berserk. His release schedule slowed a lot towards the end because he felt the need to go back and edit *individual pixels* of his scanned manga pages even though his editor constantly told him that no one could see the difference. Really hope Martin snaps out of it so it doesn't end the same way either..
Honestly, I felt a little bit of excitement seeing you work on this again, until you said at 1:44 that 3.9 MS is not good enough. It made me physically eye roll because then I realized that this project will never ever get finished, as you will always be chasing perfection. Seems like you are losing the majority of your followers interest because of this. I've followed you from the beginning and want nothing more than for you to succeed, but please Martin, follow the advice from others here, and ease up on your expectations.
Chasing perfection is a fool's errand, but Martin is no fool so we appear to have a paradox. However he doesn't seem to want to listen to anyone in the comments when they contradict his drive for perfectionism so I'm afraid he's just going to have to learn for himself. Some people won't be told, they can only learn from experience ......... usually.
I watched the entire MMX project. 99.992% perfection is why the last one failed. I'm not sure why that isn't good enough for it to never have played a song full song. by adding the finale parts. I assume personally, Finishing the project would end the journey. My logical speculation base of what I seen is, The closer the MMx was to being done the more depressed he became. most likely subcutaneously finishing that project making him depressed, so the idea of starting over made him happy and comfortable. This project is about the journey and not the finish line. I spent years checking in. And once I heard the MM 3.0 idea I unsubscribed I was so invested into seeing a man slay his dragons! And threw will power defeat a huge project.. I felt hurt he gave up... I'm not impressed by the new 100% standard because 99.992% perfection ended the last one.. There is beauty in imperfections. Peer pressure from engineers 100% perfection has gotten into his head. I add +/- 5ms to my medi for the humanize effect. I feel little hope watching the part from-to 1:13 -16:10...
@@SlartiMarvinbartfast Martin reads-and shares with us-some excellent books! I’m sure he knows the name Douglas Hofstadter, for example. Also definitely a paradox.
@@mormacil It is, however, important to keep one's expectations realistic and not chase perfection as hard as Martin does. Being sad about 4 ms instead of 2 ms in music precision is like discarding an otherwise acceptable cup of tea because it had 1 gram of sugar too much added into it.
@@pizzarella985 Depends if your goal is setting tea for a visiting friend or perfecting your skills in brewing tea or making the best possible tea. One could in the latter case go all out, trying to remove imperfections in the water and air that could in some way impact the flavor of the tea. It would be insane overkill for daily use but daily use might not be the goal.
More than anything, I think the marble machine projects have all been fantastic sorta outsider art. I hope you make a third one, and I hope it ends up as something you can be proud of. I don't think you really owe anyone anything and only you really know what this machine needs to be. I can totally vibe with being a perfectionist even when other people think it's completely unnecessary - it's just part of the artistic process!
Just watching the makings of something new and moving towards something that can actually work is fun. maybe it wont be soon, but watching these videos is both fun and informational.
I admire your tenacity on this project, I used to love to wait every week for your episodes and even bought merch of the MMX, but now I'm failing to find awe on this level of perfectionism, this feels like seeking for something you won't find on a professional musician or in an engineering project, perfection will always feel like trying to chase infinity.
Exactly! I used to follow the mmx project closely and watched every episode with great joy. As soon as the whole livestream stuff started happening, I could just not get myself to watch. In the end this sympathetic project of a big marble machine playing music turned into a lame hunt for accuracy which was never the point in my opinion. Of course it's Martin who decides the direction, but it's exactly the direction that makes me wholly uninterested.
It was always a joy when martin uploaded. Now, since the last 2 videos, i am not happy as i was. I understand, that the purpose/goal is no more music, no more exciting spinning, rotating devices, but an impossible mission to 0ms delays, to perfection in any way. NO HUMAN can EVER be perfect, no one WILL hear these miliseconds. I hope we will go back to the roots, to nice music and exciting for the eyes too. Love and hope from germany
If you can't see the beauty in surpassing normal human musical ability through the application of mechanical ingenuity, then I think you are missing most of the point of these marble machines.
It hurts how close he got to a working MMX before throwing his goals completely out of reach. He was finally going to get back into music-making and now we’re stuck chasing milliseconds like a robot chasing the end of a rainbow
Yeah… but, at least from my understanding, there really isn’t much that can be done with the MMX (can’t handle marble throughput in actual music- the magnetic lift just doesn’t have the capacity) unless he rebuilds a few major parts- which would present more problems. I can really relate the the feeling of hopelessness when you discover a major flaw in your project which is almost done..tho the best choice would be to try your best to save it. I really hope he does that for the MMX, it’s a great machine! I hope he doesn’t lose the art that started this all
Is it just me or does it sound like he lost the magical spark he had when he was chasing the crazy dream music machine? It now he just sounds like an engineering robot filling out spreadsheets and checking code. 🥺
The MMX couldn't play a full song - let alone a concert - or a full tour! Martin could have recorded an album using samples and major repairs during the process. (That is how the original Marble Machine "played" after all) But the frustration was hampering his creativity. The video series of him building includes a LOT of original music. But notice how much other instruments featured in each song? The MMX was never a solo instrument in those creations. Many times it was barely a sample or a backing track. This project is all about getting a machine that CAN play like whole band by dropping marbles.
all he had to do was write one bloody song and play it on the MMX and put it on youtube as a payoff to all the people who supported him with time and money. I can't even get myself to care about these videos anymore.
Martin is personification of Morty when he discovered true level. Martin needs a bit of memory wipe so we can get this going again. Great to see more videos and I really look forward to the final product.
While your goal might be designing a perfect marble instrument, I think the true blessing has been the art you've created along the way, and the fact you've been able to share it with us
Don’t ever give up on perfecting your marble machine, Martin! Even if it just becomes a side project hobby, please don’t ever stop on your build and design. You are creating a beautiful machine that will last centuries! You are creating a mechanical wonder for thousands of generations to come! ❤
During my time in the engineering profession, one of the more important lessons i've learned is to prioritize: 1. Finish in time. 2. Make it work as good as possible 3. Make it look nice. I believe Martin started with nr 3 on top. He now has moved up to number two, but my prediction is that nothing will come out of this endeavor before he learns to include point nr one aswell.
I love watching Martin's videos exactly because he doesn't sacrifice quality for arbitrary deadlines. "There's never time to do it right, but there's always time to do it again."
@@AndrewCooks the thing is you can’t know what right is without a preliminary version of the rest of the marble machine. How does this gate react under the vibrations of the rest of the machine? How does it react when the marbles are jostling? Sometimes good enough is good enough, especially when the problem isn’t well defined yet.
Lmfao look at this guy. The Marble gates are a central mechanism and he's solved them to perfection in 2 weeks. That's a far cry from the months and thousands he's previously spent. Get over yourselves lmao he's doing this right.
This isn't an engineering project however. If this were an engineering project he would give up on the marbles completely because it violates the various principles about efficiency.
@@gideonrabson4600 This guy knows what he's talking about. Right now Marton is taking the 'make everything as precise as possible, and the end result will also be precise' approach. This can work, but is very laborious and there's a cost to quality. More than what's requires is actually detrimental, as it takes time, energy and resources away from other parts of the design, or from moving on to other designs. The goal here isn't to make the most precise marble gate, it appears, but rather to make a better marble instrument. That prpbably entails a more precise gate than previously, but there can be emergent properties that arise when you begin coupling things together in a larger system that need to be controlled for that could very well make this extra precision unneccessary, or worse, actually detrimental. Imagine he gets a perfectly dialed in gate but it somehow induces a resonance with another part where a bit more scatter would've avoided it!
If this crazy precision is something you are really passionate about, I'm glad to see you going after it. Just as long as you're doing this because it's something you want to do, not just because you feel like you _have_ to in order to make the ultimate marble machine, lest it all be in vain. Sometimes, imperfections are okay... but if you really think you can do this, and it's what you want to do, then I'm glad to get to watch your journey. After all, it's not our dream you're chasing. It's yours. I can only hope we'll one day get to see it in all its glory.
@@MoritzvonSchweinitz that's really jawdroppingly accurate from an outsider's perspective, sheesh. (i play rhythm games and my standard deviation always sits around ±20ms :dead:)
Well if we look beyond the numbers, its just martin completely removing one factor from the inevitable future of him debugging his machine when something performs suboptimally. The gate wasnt that big of an endeavor, and it also gave him the peace of mind. I dont think you creatives get how many sources of errors there really were in his previous attempt, and how he led himself to a deadlock by the end.
@@r3ll282 HOLY JESUS CHRIST, HOW MANY TIMES DO I HAVE TO SCREAM INTO THE VOID THAT IS THE INTERNET BEFORE PEOPLE LEARN THAT THE SUBCONSCIOUS IS STUPID PSEUDOSCIENCE?? THERE IS NO GODDAMN EVIDENCE THAT THERE IS A SUBCONSCIOUS! IT'S A FOOLISH IDEA CREATED BY GODDAMN SIGMUND FREUD! THERE IS NO GODDAMN REASON TO BELIEVE IT EXISTS! THE ONLY REASON PEOPLE BELIEVE IT EXISTS IS BECAUSE PEOPLE MADE BAD DECISIONS! LOOK, THAT GUY MADE A BAD DECISION AGAINST HIS BETTER JUDGEMENT! IT COULDN'T POSSIBLY BE THAT HES IMPULSIVE, OR DIDNT REALISE THE TRUE SCALE OF HIS ACTIONS, OR THAT HE MIGHT JUST VALUE SHORT TERM PLEASURE OVER LONG TERM HEALTH, NO, IT'S SOME GODDAMN BOOGEYMAN THAT CONTROLS HIM! ALSO, THIS BOOGEYMAN ALWAYS CHOOSES THE WORST GODDAMN OPTIONS, LIKE STAYING WITH AN ABUSER! AND IT'S ALSO SUPER-INTELLIGENT AND CAN PERFECTLY READ ANOTHER PERSON BECAUSE IT'S THE BOOGEYMAN! IT CAN DO ANYTHING IT NEEDS TO ASLONG AS IT BOOGEYS! I AM SO GODDAMN SICK AND TIRED OF PEOPLE ACCEPTING THE SUBCONSCIOUS AS FACT! ESPECIALLY SINCE **SIGMUND GODDAMN FREUD CAME UP WITH IT!**
@@r3ll282 ALSO, AMAZING HOW FROM A FEW CURATED VIDEOS WHERE HE DOESNT EXPRESS ANY DISTRESS, OR ANGER TOWARDS THE MACHINE, YOU CAN MAKE A PERFECT JUDGEMENT ABOUT HIS PSYCHOLOGY! MAKES ME WONDER WHY THOSE PSYCHIATRISTS CAN TAKE WEEKS TO DIAGNOSE SOMEONE AND NOT ALWAYS GET IT RIGHT! GUESS YOURE JUST BUILT DIFFERENT HUH? AND WHY IS IT SO BAD HE IS PURSUING PERFECTION? BECAUSE YOU WANT TO HEAR THE MUSIC ALREADY? BECAUSE IT'S "UnHeAlThY" TO BE GREATLY INVESTED IN SOMETHING? DO YOU KNOW HOW MANY THINGS WOULD BE WORSE IF WE ALL FOLLOWED YOUR STUPID GODDAMN DEFEATIST MINDSET? MASTERPIECES HAVE SOMETIMES TAKEN DECADES TO CREATE! CAN YOU NAME A SINGLE CLASSICAL SCULPTURE THAT HASN'T BEEN MADE IN MORE THAN A YEAR? MARTIN DOESNT WANT TO BE KNOWN AS A MAN WHO CREATED *A* MARBLE MACHINE, HE WANTS TO BE KNOW AS THE MAN WHO BUILT *THE* MARBLE MACHINE, A PIECE OF ART SO AMAZING IN IT'S ATTENTION TO DETAIL AND THE QUALITY IN ITS CRAFT IT'S NOTABLE FOR MILLENIA TO COME! THAT IS AN ASPIRATION, A DETERMINATION, TO BE THE BEST YOU CAN BE, THE GREATEST, THAT CAN ONLY BE ADMIRED! AND YET YOU SPIT ON HIS YEARS OF EFFORT, SEE HIS HARD WORK AS A CURSE? SO DECADENT YOU ARE THAT YOU SEE A MASTER AND NAME HIM A FOOL? BECAUSE IT'S "GOOD ENOUGH" AND DOESN'T NEED TO BE IMPROVED? BECAUSE YOUR STANDARDS ARE SO LOW YOU THINK STRIVING FOR EXCELLENCE IS A WASTE OF TIME? WELL SCREW YOU! AND YES, I AM FULLY AWARE OF HOW RUDE THIS COMMENT HAS BEEN, AND I STAND BY EVERY SINGLE GODDAMN INSULT, BECAUSE YOU ARE AN HONORLESS COCKROACH WHO SEES NO VALUE IN IMPROVEMENT!
These investigations go far away from common music and I SO love it. Your videos are pure engineering, mathematician and statistics experience, and as for me, non-musician, this content is much more interesting than just listening to your (of course amazing too) music. So go on, thank you so much for keeping in course and hello from Russia!
I'm happy that you're back Martin, but don't lose your marbles with the timing thing, I personally think that music isn't about being in perfect sync always, it's important, don't get me wrong but, music it's about having fun, expressing yourself, delivering a message. We don't care if the message reaches 1ms, 2ms or even 5ms late, we care that the message it's meaningful. Make that message have a meaning. We'll be here to listen to it.
In all seriousness, I 100% agree. What I love about the original marble machine (as well as most other arts) are the "imperfections". It feels natural, emotional, human. If it's 100% perfect, it just feels stagnant. In the end, it's soul & intention that drives the whole narrative.
@@gomichow The thing is that at the end of the day, the marble machine will rattle and vibrate. Wind can and most likely will be a factor during a play, and so many other possible cases that I can't think of. So imperfections are still very much a thing with this project. What I see is that same "it feels inhuman" argument every single time, and not once do you guys think about what he wants to convey with this project but rather what you want from him to convey to you. Kinda selfish if you ask me. Don't take this as an attack at you specifically, but I just dislike that take.
@@kyucumbear I agree that it's kind of selfish to think that he's making this for us, well, in a way, the viewers are one of the main pillars that support this project, but we can't force him to change the way he thinks or the vision that Martin has with this machine. My take it's more focused around the fact that, if, only if the way his vision is focused around making a machine to show to the world, to impress people and to make a instrument that will be travelling around the planet, I think that focusing and making all this effort, tiny changes, all the measurements, redesign's and all the things that he thinks that are essential, will lose the battle again like happened with mmx, because even if you can control 50,100,200 variables inside the machine to be absolutely perfect, there will be hundreds, thousands of external variables that he just can't control and it's possible that because he just can't control everything, he will get very frustrated, because he wants to make the best machine, for us, for him or for the instrument, either way, it's not a healthy approach to make a project. It's important to have fun, to be creative, to make it your own, break a few rules, upset some people (us included) because of the decisions he took. The optimal way isn't always the best way in life, in my humble opinion, the best way in your life, proyects, plans, it's your own way
Cheerful Martin makes me cheerful. As always, I want to emphasize that the value of your videos is your videos, not some end result. It is the evolution of design and discussion of basic principles that hold me enthralled.
1:02 I couldn't tell the difference in the sound but I could definably tell that the sound was coming from the left first even at 1.46ms. I'm using a 30y+ old Sony MDR-V600 headphone. That surely helped a lot. If it was on normal speakers I'm 100% sure that I wouldn't not be able to notice the side that was coming first.
I got really confused for a second when you said "hold onto your papers scholars". Your deep dives into design is amazing to watch and is incredibly inspiring. Heres to MM3.
3:10 Not in this case. It's Martin's third approach to this and he already knows what issues are the most important to solve. Optimizing later like on the MMX was and would be harder
@@maticz3923 true, and at the same time, one would not have to put all the pieces together before optimization. But by working on a broader range of parts before optimizing one might make connections that will help anticipate later problems. But as a disclaimer, I am talking out my ass.
I disagree in this case. Early optimization may be the road to succes, because he knows what he wants now and can test and optimize the design before losing time on building a machine that won't work because the design isn't optimized (like the MMX).
Wow this is still going? Good. Now for the important part...NO ONE IS GOING TO NOTICE UNDER 4 ms DEVIATION! Our tvs don't come close to that for gaming, and no band is ever playing that tight, it's humanly impossible to do that for a whole song anyway.
this fails to take into account tolerance stacking. there's multiple parts of the machine that have timing variance. So whilst you're correct that 4ms deviation doesn't matter, 4+4+4+4+4 ms does
While I agree that most people won't notice anything below 5ms jitter there is a difference in the game console to TV scenario because that is a constant delay so it won't affect things internal timing between notes. Jitter is worse than constant delay for music. Also, standard deviation might be all fine and so but for measuring over all performance the 95th+ percentile can be just as important to keep track of because you want to know that you have acceptable performance at the edges as well. Contemporary games are typically designed to allow for some input and output latencies. An example would be allowing a character to start a jump in the air 50+ ms after they have left a platform. You can also notice this when you try to play some pixel perfect platform games from the 80's on some modern TVs, the timing is so tight that the games get almost impossible to play.
@@DeanLawrence_ftw I am sure this is why Martin is doing this knowing his past attempts. Even if each part has no deviation in isolated testing, when he puts them together into a whole system on a newly designed Marble Machine, there will be deviation introduced through the sheer complexity of the system. Eliminating as much as possible up front in the design process is important for that reason.
The second marble machine didn’t suck, you just kept moving the goal posts back more and more until you managed to set goals that were totally unreasonable for a complex mechanical machine. Think of watches... The finest Swiss watch on earth has allowances for beat errors... The finest horological pieces in the world will gain or lose seconds per day... perfect accuracy isn’t the point. They aren’t nor ever will be perfect... If you want digital accuracy, make digital music on a laptop and be done with this nonsense.
Right? The marble machine X is freaking BEAUTIFUL! Its a work of art! Its freaking breathtaking! You're a musician and an artist Marvin, do you really want to spend your days making sure marbles drop with 0.0005 ms accuracy? F*CK THAT! we don't give a sh*t about that, and neither should you!
It's important to not see the failures as such; as failure. Rather, they should be seen with pride, of something you attempted and didn't work to build better from it. Be proud of your attempts! Be proud of your work! Learn from your mistakes and one day, materialize this dream that felt so far away before! Bon courage, Martin! You can do it! I love to see you're never giving up!
First, I'm really glad that you're working on it again, and I hope that you have a lot of success. Second, I know you didn't like your first version, but several hundred of those many views you got were from my family, so it's fine that it wouldn't be great for a concert, but it was also special and beautiful, so don't focus on the imperfections, because no one else did. Last, lots of other people have said it, "it doesn't have to be perfect, blah blah" and they are correct, but I understand that it's as much art as it is music, and so you for sure want it to be as close to the perfect as you can. But imperfect and finished is better than never finished because it can never be perfect. Congrats on figuring out the timing solution.
Just a friendly reminder to not forget the pressure that gets built up by all the marbles pushing down on each other as it could affect the timing of the marbles. HOWEVER that being said, anything under 5 ms is not going to be noticeable to anyone Martin. You will ALWAYS be able to find some flaws with the machine. So don’t stress to make it absolutely perfect going forward. I think this kind of deep dive at the start of the project is the way to go. Just don’t get too caught up in the details when you do build another machine. Even if you don’t, I’m still here to watch and support whatever you do :)
The pressure does not change anything as everything accelerates downwards at the same rate. I can stack fifteen tanks on a marble and it will fall at the same speed
5 ms is so insanely tight. That's the timing on many drum transient attack sounds before they reach their peak amplitude. It's stupidly tight. I make electronic music and deliberately give my drum programming more loose timing than that to give it a bit of a "human" feel, and it is STILL very robotic. 5-20 ms is fine most of the time. At this point he should just sample the sounds and make his music in Ableton. This misses the entire point of creating a physical mechanical machine that makes music imo.
The musician in me says DAWs have functions to remove perfection from MIDI-Tracks to make them sound more realistic. The engineer in me, as i build measurement devices says: if you have no dispersion in your data, your resolution is not high enough 😉 As someone who believes I say: I missed you Martin! Thanks for coming back! 😀
We've heard some songs on it. The MMX had a few fundamental flaws that prevented it from playing a song utilizing all instruments. The marble divider on top let the marbles traverse too slowly, leading to lower gates to be starved of marbles. The marble lift gears had the wrong gear ratio, with the lower one lifting more marbles than the upper could lift, leading a full machine to spill many marbles. He redesigned the upper part and when he wanted to connect the marble divider to the marble gates, he noticed he didn't have enough space for the connections. Originally Martin wanted to finish it to play a last song for RUclips but at the marble divider to gates space issue, he quit. It really suffered more issues than it seemed from the videos.
@@Treksh the first one never worked. The famous RUclips clip was composed of probably 60 different takes and fixes on the fly and then edited together afterwards. That video shows off a concept, an idea. Not a working machine. It could not play a single tune.
@@tjalve1 ... and yet he made a massively popular music video with it. As an absolute minimum he could have programmed MMX to play the same song, and edit it together if necessary.
sure "no timetables" while literally hundred of people keep giving him MONEY each month. yeah you could stretch this out for YEARS... oh wait, he already has. And just as he got close, decided to scrap it and start again, for even more years of free money while playing with toys.
@@potatosmasher1072 Yes. I know you may think this is serious engineering, but my problem with this is as I stated. If Martin was actually serious about getting this done he would be doing the same thing as he did the first time with the wooden one, get a WORKING version completed - doesn't have to be PERFECT, it just needs to WORK - after that he can work on the iterative process of streamlining efficiency. That is what a true engineer does. The problem is Martin has brought his "creative" brain to a LOGIC problem - he wasted so much time trying to make it look "cool" that he was building problems into the systems - and now he's paying the price, but he's just diverted from one "cool" situation to another. It's all "cool" and not "what do I need to do to get this working, I can make it look cool later". Every single engineer on Earth will tell you a simple fact of engineering, you never get it right the 1st time or the 3rd, or 5th, and the more complex and moving parts it has, the odds against you go up exponentially; if you try to get it 100% "cool" + 100% "efficient" the first time, YOU WILL FAIL. Brutally. In the tens of thousands of years of humanity, it has never been done - the simple lightbulb took something like 10,000 attempts before a working version was found. A bloody basic lightbulb, 10,000 times, think about that for a moment. Every single thing ever made in human history from the humble fork to a rocketship has been through dozens or hundreds or thousands of iterative changes, and Martin thinks he can get it 100% right and 100% cool on only his 3rd try after his 2nd try didn't even reach completion? Give me a break, he's deluding himself, and all of you. I'm probably the only person among the rest of his simpering fans that will actually say the brutal truth, the truth that might; just might, get him to stop following the "perfection" rabbit down the hole and instead FINISH IT. Martin didn't even succeed at finishing the Machine X, so he only got PARTIAL DATA from it, and engineering with only partial data will mean only one thing - the next iteration still has flaws, and if Martin follows the same path as last time, this next machine will not be completed either, because more flaws will show up, and so...he will start again.... and again.... and again. He could spend a lifetime chasing it, and some people have. IN ENGINEERING IT'S IMPOSSIBLE TO BUILD SOMETHING BETTER, WHEN YOU NEVER "FINISH" THE PREVIOUS VERSION. So yes, emphatically yes, Martin is "playing with toys", fancy expensive and technologically more advanced toys, but toys all the same. The only advice I would give to Martin is make the new system modular, so improvements can be made without having to restart from scratch, again, as he is now. Martin owes it to the people who gave him money or thier professional time to get a completed version, otherwise it's what we call "vapourware" - all idea's but no solid outcome, in which case he is a fraud.
@@superkaboose1066 People thought they were going to see a project from start to finish not, start, get 3/5 (many years later) then abandon it to start again.
One thing that he not mentioning in the video is that total variation on timing has to be considered, but he is mesuring just the variation due to the gate.
Please note that this is only the marble gate and that inaccuracy in the machine will add up. 4 ms isn't noticeable, but if you lose that amount a couple of times you end up with audible delays. So he might seem crazy in designing a gate that is super accurate, but good design now may lead to better results in the end.
@@EaglePicking the inaccuracy is only within 1 beat, they don't compound(its all tied to a master beat of sorts in the machine, the mechanism that plays) and the gates tied to the release. So whatever it is it can only add up between the the program wheel/device and the gate. The program wheels the real problem anyway in portability and keeping it as an 'instrument' instead of a music box. The keeping it as an instrument part kinda adds up something anyway that has variation. Its just tinkering with a gate and spending manual time measuring an automable thing to measure, to pass the time. Edit: also the 'good design now for better result later' has already been on meme level for like 3 years(not functioning as intended for having a result)
I can appreciate the dedication to excellence. True that all the inaccuracies will be additive in nature. Also bonus Kudos for realizing part of the deviation was in the measuring.
He is too far down the rabbit hole at this point. He hears the faint murmur of those on the surface telling him to end this quest for the impossible but it is over powered by the perfectionist within. I wish you the best Martin and hope you don't lose yourself along the way.
As soon as I saw design 8 with the extra bit holding the marbles after the escapement wheel, my mind instantly went to the old design with the kinetic fingers lmao
This video has given me so much hope and really showcased the growth in Martins ability. Seeing Martin blast through something that caused so much trouble prevoously and seeing how deliberate he was with each of the design iterations was rad. So much of his earlier design work was "because it looks good" but that lead to so much rework later on. Even tho Martin is taking baby steps now, I wouldn't be surprised if he has a new machine designed and built within a year.
As far as the comments here go, I would just like to say that I'm loving the engineering aspect of this. It's very fun to see, and the enthusiasm is a pleasure. I believe that the "it's good enough, don't waste your time" are both sensible, but also besides the point. When all things are put together, errors will compound. Having each individual designed well, will make it motivating to put together. I'm sure you're mindful of the adage that "perfect is the enemy of good".
It's pretty bad engineering though. Focusing on single aspect and trying to optimise them to perfection before you even know what it's going to connect to is a waste of time. Part of the reasoning he gave for abandoning the old machine was that he kept having to rework old parts when adding new ones. He clearly leant nothing from that experience. You don't go straight to making a perfect machine piece by piece, you first make a prototype, but he refuses to make a prototype of the whole machine, instead he wants it to be perfect as soon as all the pieces are put together.
@@emmastrange5557 I don't disagree with you. If you consider engineering as "get the job done sufficiently well", as to move on and solve the bigger picture. Then you need a clear set of constraints as to what is "good enough". And, I believe Martin is way off here. Even 10 ms would arguably be sufficient. However, that aside, one can appreciate the engineering in improving a system. And, I do think that he is up front about it. He isn't solving one part of a whole. He is solving each part separately, where the goal itself is a "let's see how good we can make them". The MMX thing was a bit of a sad story. But this looks like having fun designing contraptions. The goal is the design of the contraption itself, and not "build a machine to play music". If you disregard the previous history and mistakes. And just see this video on "how to perfect the timing for dropping marbles", I most certainly enjoyed the engineering aspect of it. And, I wonder if its not unfair to force a different context onto this work.
My suggestion, if you end up making a third marble machine, is to make the small parts (like the gates) which could break snap onto the larger mechanism, so that if one ever breaks during a performance you can just quickly replace it and figure out why it broke later
Im so damn happy that this channel is back with the regular uploads as it is one of the best engineering channels out there. It doesn't even matter that you are starting a new project Martin, showing these perfect examples of the benefits of designing to first principles, robust design, useful testing and data analysis, all while still keeping it super interesting for us viewers is just a joy to watch. Keep it up.
This is the correct comment. Ignore the haters. They think the purpose of the marble machine is to be a musical instrument, while in reality, it's true purpose is to generate a collection of design videos for us to enjoy, and cool merch for us to buy. Silly haters.
@@iamiggie its understandable though. Martin is a musician, and his fame came from releasing music on a marble machine, and the goal of the next one was to create music and tour with it. There are two large segments of his audience - the musicians and the engineers - and there isn't much overlap. The former category are somewhat disenfranchised by these pursuits of perfection and obsession over minutia since they would prefer to simply see art being made, with all its quirks and flaws. Meanwhile, the latter category is perfectly happy to come along for the ride and are more amenable to the deep dives into design and desire to achieve incredibly high levels of precision.
Also he needs to measure, calculate and account for differences in strength in local gravity (gravity is not a constant on Earth, it has ever so tiny changes, thus it's value changes every yoctometer), and put the machine in absolute vacuum to prevent air molecules from causing uneven resistance and drag (molecules in gasses aren't spread evenly), causing a deviation as big as femtosecond. Can't wait for him to resolve all the issues like this in design phase and prototyping and move on to build the machine in the next few hundred eons when science and technology finally allow him to live in virtual reality where all of that is finally possible. /s
awesome idea ! just a little and very accurate scale before each marble dropper, and a simple mechanic to move the whole dropper up and down to compensate for the differences.
Nice video! It is very informative! I started to watch your videos as soon as first video arrives, since then, just the first video I watched dozens of times, and using as an background music to meditate. Im am a soviet engender. I assumebyou couldnuse all your 1 trough 8 gates, with even better performed results, by adjusting un-delay or in another word, drop your ball earlier, that way you can use potentiometers to adjust all from - to +
I found your original marble machine video last night and remembered I'd seen it years ago, I've since spent the last day watching as much of your progress and recent videos as possible. You have such amazing energy and passion for this project, I hope the 3rd machine does end up being viable and turns out to be all you hope it to be in the end. Keep at it, many more small steps and revelations to go, thank you for inviting us all along for the journey :D
Honestly, I can’t hear a noticeable delay until you played the 50ms example. Not sure how vital this is going forward- maybe it somehow becomes more noticeable with many notes? Might be a goose chase if not. Haven’t finished the video yet, so maybe Martin addresses this, but I think he should define the acceptable “good enough” deviation. He uses comparatives without having a defined requirement/goal.
Get your ears adjusted! ;) A 10ms delay is clearly audible as a double hit and a smaller delay is audible as a “wider”/longer hit - with my ears, at least.
@@peterfireflylund Got to remember there will also be huge variation when dealing with that level of fussiness in the quality of the sound delivered by YT and how the viewers audio system processes it. If your PC is running at CD bit rate but YT isn't even more than 50ms could be lost quite easily to the conversion processes, and if the room has some echo again the difference can easily be lost, we don't all live in sound deadening music studio boxes...
It is an "optimal settings" situation. There will be many factors adding to the delay and the tighter it is initially, the better the end result will be. And in the best case, everything holds together well enough for the programming to control the output to a degree of inaudible delay. Imagine a drummer how does not play a drum but nudges another drummer who plays the drum. Each of them has a delay of 10 ms, the precision of a pro studio musician drummer. But together, they end up with 20 ms. The machine will have a lot of drummers nudging each other.
I keep watching these videos in the hope they go back to how they used to be and they keep being less interesting. He used to be a crazy, genius musician creating an amazing machine. Now it is like watching an engineering student who has just read the latest book and thinks it is everything.
I know I’m late to the party on this video but, I’ve been watching you on and off since you started the cad one, and honestly it’s incredible how far you’ve come. I mainly keep up-to-date with it because of your passion for the project it’s fun to see you succeed.
Idea: analyze the accuracy of timing in the tracks Wintergatan has recorded previously to see what kind of deviation is there. I’ll bet you $100 it’s over 10ms!
yeah but this is just the gate, he didn't even want 0ms accuracy, it just happened. What he wants is to be tight enough so the rest of the machine's parts have margin of error, because if 20 parts have 3ms delay, you end up with 60ms. The more accurate each part is, the less it will add up in the end, that's why he made a 5 point list he wants to get accurate before building a whole machine that isn't :(
Standard deviation itself is not the most intuitive metric to report. I would recommend 95% confidence interval (which is calculated from standard deviation). That way you can say “95% of marbles fall within X ms”
I can't wait for the moment when Martin realizes that too tight music is sterile and uninteresting and then he goes on for another 3 years to make a mechanical randomizer to make it sound more human.
Yeah especially because those inaccuracies are what make music sound good. Take a jazz swing for example. The reason it sounds good is because of it's lack of precision. It can be best described as a triplet based rhythm but it's not quite.
Eh, I don't agree. I think the real substance of music and what makes it interesting is in the composition. The making of something creative from nothing but the artist's imagination; that's where the magic happens. I've listened to many deeply beautiful and immersive pieces of music that didn't have any of those tiny imperfections, and they didn't feel boring or lifeless at all. A composer doesn't just crank out material for a musician to play, they are an artist in their own right. There's definitely artistry to playing an instrument, and sometimes you lose a lot by taking that away (violin soundfonts and sample libraries always sound terrible, for example, because it's an instrument that involves so much subtlety and variation), but I'd say there are also cases where that rigid timing is a better fit. It can help keep rhythms clearer, and in some styles too much expression can even be distracting. In a lot of really fast and rhythmically dense electronic genres, that clockwork precision is absolutely essential for making it work at all.
this whole never ending pursuit of "tightness" forces me to back up and ask "Why do you need this instrument to be so precise" as I would very much doubt any human being playing any instrument has a "tightness" of less than 2 milliseconds. Why not just keep the tightness to the "human scale"?
I guess that he is spending so much time on this because any other deviation from the true beat that other let’s of the machine will introduced will be multiplied by the gate.
@@maticz3923 okay. Lets carry that onwards. 2 in the gates + the 2 in the programming wheel = 4ms. Lets be generous and say vibration is 3ms, so 7ms total. Marbel decay? sure, 1ms gives 8ms. Different air movement on stage? + 1ms = 9ms. Instrument rattling, 3ms = 12ms. Friction warming the machine and changing the paths, 1ms = 13ms. I could keep doing this, make as many bogus claims as I want, and we are _still_ under half the delay of a professional drummer. There is no precision here, this is just pure madness. This is putting carpenter nails under an electron microscope to make sure they are all "perfectly" equal, just to hammer them into the shed roof.
I'm torn - as much as I enjoy these videos and the data analysis and constant design improvement, I also just want to hear some music. That 5 second clip of marble machine two at the very end of the video is incredible. Its more so about the journey than it is the end goal, but we also want some pay off in terms of that sweet sweet marble machine music!!
Watching that line hit a different spot for each measurement at 13:30 while Martin says its "zero" made me realize I still have no idea what he's talking about.
You are a huge inspiration, Martin. The way, you look at the meta level every single video, amazing. Going for the best results and celebrating perfection. I love it.
Dude at this point your going to have to put the marble machine in a vaccum chamber to eliminate air currents. Control for temperature to eliminate expansion of material. Put the machine on shock absorbers to remove vibrations, and have every marble rounded to a perfect sphere with 0 tolerance, to achieve the perfection you are striving for. It doesn't have to be THAT perfect, and if you look at any instrument, there will always be little imperfections. It just needs to be good enough, and your MM2 was just that if not more so. I would say it's impossible to make a perfect machine. Once you have all the moving parts running there will be vibrations that will cause varience and parts interacting in ways you may not expect. I was with you in the beginning, but now I feel you are chasing an impossible dream.... Best of luck on your journey!
I think Martyn's point is that if a gate connected to a computer controlled solenoid is off by a few milliseconds, then it's going to add to all the other inconsistencies. If you could design a gate that adds no inaccuracy, like the one in this video, then it's something that you don't have to worry about. There are things that he can't do anything about, which will add inconsistencies. Saying that, I still don't think removing a couple of milliseconds will have any impact on the way the machine sounds.
I've been watching this channel for a couple months shy of 10 years and as much as I love Martins work, this obsessing and scrapping perfectly workable components has become a looking glass into the world of a mad genius who is steadily losing his mind and he doesn't even know it. If you ever read this Martin, you need to look after your mind, this obsession is not healthy.
I am very impressed by your skill to build perfection. But remember, that perfect is the enemy of good. Keep on with your amazing journey! It is one of the things, that fullfills all my weekly needs in one single video.
I feel like I must have missed or forgotten about the video where you explain what went wrong with the previous marble machine. It was over-engineered, yes, but it was awesome. Why did that project fail, why did it suck, what happened? I'm confused...
Martin was dead-set on being able to take the machine on tour and do live performances with it. This placed very high demands on reliability, to ensure it could be transported and perform accurately and without failing during a performance. Ultimately, he decided MMX had too many features that were too complicated to ever be as reliable as he wanted it to be. Whether Martin is throwing away "good enough" in pursuit of unrealistic perfection is up to debate.
With regards to live tour performances i think the straw that broke the camels back was that after designing the same system 20 times he couldn't get the marble drop system to do perfect 1 million drop loops without losing marbled Eventually parts broke down
Martin has a point. Yes, people wont notice if gate itself has small deviation. However, gates are not the whole machine. Mechanism that goes from the programming wheel to the gate has its own deviation, and marble dropping on an angled surface has its own. So deviation gets compounded over the mechanism. He is going backwards, minimizing deviation from the end to the beggining of the mechanism, which is really smart.
I mean...if you want to build some high-precision thingamob (instead of "only" a very cool musical instrument), more power to you. But in that case, I don't think you should be using contact microphones and an audio program for measurements (software will introduces more variables - that's what osciloscopes are for!), and use precision build parts (instead of, it looks like, FDM 3D printed parts. Resin printed parts would be way more precise). Also, you'd have to bear in mind that sound only travels aprox. 30cm in one millisecond. And that the solenoid may have different performance metrics after heating up a bit. Also, ALL other parts of an eventual machine will have to work to the same spec. What if some other vibration of the machine affects the marble gates? And so on, and so on. Chasing precision is of course a great and sometimes fun rabbit hole to go down. But it's a very deep and frustrating one, and reeks of depth-first search.
I hope the overwhelmingly negative and concerned reaction to this video reaches Martin in some way. We really care about this machine and the art you set out to create and doing this to yourself isn't healthy.
I actually hope for the opposite. He wants the journey, not the destination. He likes to challenge himself so he can learn more and more. And the engineering part seem to be just as important as the music part for him, I imagine because both are pleasurable for him. There not being a finished Marble Machine (EDIT: a Marble Machine that can achieve what he set as his goals for it) is not bad as long as he feels it's not possible. Right now he's trying to make sure of that, which is really the important part.
@@mateusbmedeiros Makes me wonder if he's just milking a paycheck at this point. He's redesigned everything like 20 times now and it seems like he plans on doing that forever. Discord really lives up to its name.
@@BaronCreel I think if he was he would've kept the youtube membership in which people gave him money directly and maybe some filler videos to net more ads, I'm guessing. But of course not being Martin himself I can't really say for sure.
I often end up struggling in spirals of destructive perfectionism, which is an obsessive drive to endlessly improve details that are irrelevant to the quality of the final product. I suffered a huge amount of frustration in my professional life because of that. I see some worrying patterns of that attitude in Martin's work as well, but maybe I'm just biased due to some personal failures I have suffered and attributed to that.
The first marble machine was tight enough to inspire millions of people. I don't think it needs to be any tighter, just modular for transport and durable for the world tour
The speed of sound in air is 343 m/s. It takes 2.9ms for sound to travel one meter. You are optimizing for error which is smaller than the error introduced by the fact that the instruments on the machine are different distances away from the listener depending on where they're standing. With love, from an engineer who has heard you say you want to balance engineering and art: this is not engineering. Engineering determines what error is acceptable and then optimizes for that. If the ear can't hear the error, and the error doesn't accumulate into subsequent notes (which it doesn't due to the drum design), it's irrelevant.
It's hilarious that Gate 8 resembles the very gate that was worked on years ago - but that the release finger is part of the very escapement arm and much much shorter instead of a long lever.
Ahhhhh, such a delight to see Martin work! And I must say, the infrequent/randomness has me so much more excited to tune in than the high-stress weekly vids! KEEP UP THE GREAT WORK IN WHATEVER PROJECTS YOU PURSUE MARTIN!!
Martin:
- Step 1: Look, we can't hear any difference up to ~20ms of delay
- Step 2: Proceeds to optimize a 2ms delay down to 0ms
Perfectionism is a blessing and a curse in one package.
@@BloodyMobile it really is. slows progress down. but to a certain extent, if its perfect you dont have to go back and revise it in the future, but then again you never know what problems you will run into anyway that you didnt foresee. but getting something that works now, VS is perfect later is sometimes good enough. of course design choices make it more difficult because you might not be able to come back to 'make it perfect later' if its a fundamental design choice. I guess its best to aim for 'roughly perfect'. good enough, and modular so parts can be adjusted as needed.
in the audio example i could hear it even at 1.2 MS off. But i suppose what matters is 'when listening to music, if the notes are off by 1.2ms, would anyone notice?' and i suppose the answer is no. How many musician play that tight to be 0ms off?
@@jessiejanson1528 I think the majority of people can not though.
For me it simply sounded like "one" sound until 20 ms, and only because I was looking for a difference.
Without that I might have noticed it even later.
And I /do/ have an above average good hearing, but apparently not for this kind of stuff.
Yeah that was weird, "3.9ms isn't good enough for precise timing" - wait what, you barely could hear anything Haas-effecty in the beginning, the very moment this kind of timing issue presents itself to the listener, you're already surpassing perceptual thresholds of 99.999 % of all listeners out there.
Easy to prove too, but yeah, this went against everything he established in the beginning. I mean, whatever floats your boat and all that, but that's truly overkill.
Martin being upset about 3.91ms is giving me visions of the Whiplash movie where Fletcher just keeps screaming "NOT MY TEMPO"
Keep in mind that according to a journal article testing musical "tightness", pro drummers have a standard deviation of 6.6ms for their dominant hand and 9.5ms for their off-hand. Being a drummer for Martin would be stressful lol.
@@Bbeaucha88 Agreed, I don't get the obsession with such a small deviation that is imperceptible, unless I missed something.
Rushing or dragging?
Honestly could really use more cowbell.
@@Bbeaucha88 I was wondering about exactly this, it just seems a bit silly after a certain point
Martin:"I need to make my design requirements less dumb."
Also Martin:"1.4ms deviation is not good enough "
Also the concept of a machine that makes music by dropping marbles is only interesting because of how dumb it is. Does he think this will ever become an actual instrument? It is and always will be a novelty that's enjoyed because it's dumb, its not the new drum kit and if thats what he wants then he should just use MIDIs
It's the stackup, not an individual feature, you have to hit. As he observed around 3ms or so of delay starts to be noticable and 1/2 of that error is in just dropping the ball you have very little slop allowable for the rest of the system. Things like the timing of the programming wheel, linkage delay, the angle of the instrument, where it hits on the instrument, etc adds to this error greatly and if you want it to be a good sounding instrument it needs to be tight tolerance across a number of components to get the sort of drop accuracy it needs.
@@MikeLikesMaking It won't really matter what the stackup is if the marble machine is the source of the MIDI clock, then all the other digital instruments will depend on it for their timing.
The thing that Martin doesn't explain that clearly is that Standard Deviation isn't the same as the gap between notes. 1/300 things is more than 3 standard deviations out from the mean, and that goes in both directions. So if you have a bunch of marble gates going, notes can routinely be 6x that deviation from each other. Not to mention that this is only one of potentially dozens of sources of deviation (at least in the final machine). It's possible he's overdoing it, but reducing variability in each individual piece is what you need, because it can add up, and it'll be a lot harder to diagnose in a fully-built machine than in component parts.
The missile knows where it is at all times. It knows this because it knows where it isn't. By subtracting where it is from where it isn't, or where it isn't from where it is (whichever is greater), it obtains a difference, or deviation.
The process in this series is becoming more and more like my PhD.
I hear that. Started mine about when the MMX started. The rabbit hole has no end.
When he was optimizing a little thing just because it bothered him, I felt a strange camaraderie.
Learning scientific method on trial and error instead just going to uni lol
Frick that! ;-)
Yeah in many ways this project has turned into Martin doing a PhD without a supervisor, which is kind of a terrible idea. One of the important roles such a person fills is telling you when not to go down the rabbit hole.
Remember that the Achilles heel of the MMX gates was marble pressure! Timing is important, but your number one priority when designing should be reliability, and your testing setup should replicate the stress the gate will be under on the finished machine.
Seconding this
Yes. It might be a good idea to estimate the pressure the marbles upstream is putting on the gate by making some mass that can push against the marbles/gate being tested to simulate a full track
on thing at a time
if you want to mash everything together from day 1, you will never be able to engineer anything good. one obvious reason: cannot test this.
some problems will emmerge only on final stages of production and some of them will need you to disassemble (not literally if design is good) the whole thing to debug. and the lack of testing opportunities will get you to build the whole thing from anew.
look at marble machine X if you have doubts
As a drummer I can safely say you don't need to get 0ms accuracy of timing.
You're not making the world's most over-engineered midi roll, you're making an instrument.
This machine will have no sense of feel if he keeps chasing this..
@@sonikboom007 Exactly. The design priority needs to be reliability before timing.
@@jameswatson4429 As far as reliability, I feel like the second machine was pretty much within range of reasonable expectations, he was dropping what? like a marble or 2 per 10k drops? totally reasonable. His only real issues were the limitations from the bottlenecking that was occurring with the dividing gate. I would liken that to the limitations you’d find on any instrument though. A Violin has 4 strings and no frets. A Guitar has 6 strings and imperfect intonation, the second marble machine was a bit limited as to the number of simultaneous inputs per second. perfectly reasonable when you think about it.
All the delay deviations will add up eventually and minimizing each components as much as possible is key or you might end up with a bad deviation
As an acoustic engineer designing audio systems and loudspeakers (and also a musician), I totally agree with this as well.
A rule of thumb is about 20ms as worst case. Everything within plus or minus 5-10ms is more than good enough, and probably much better than most drummers.
At this point this feels heavily like creating a problem, a bit like pixel peeping on a photo.
A photo that has beautiful scenery, composition, story and feeling.
This is literally beyond human hearing capability. Don't get lost in the details Martin, it's a dangerous rabbit hole
Please note that this is only the marble gate and that inaccuracy in the machine will add up.
4 ms isn't noticeable, but if you lose that amount a couple of times you end up with audible delays.
So he might seem crazy in designing a gate that is super accurate, but good design now may lead to better results in the end.
@@EaglePicking that’s standard deviation. He can still release it’s just each might be off by 4 ms individually
@@EaglePicking humans are very inaccurate. I don't think most musicians can keep withing 10 ms of timing for entire songs. it's just insignificant.
Also the original marble machine sounded great but I'm pretty sure it isn't within 10ms
@@Anonymous41726 that is in the most important part in the machine. And deviations add up, and esp. if you play many notes together and fast.
And we STILL have not counted the deviation in other parts, like programming wheels, which could need this extra space within what we can have before it can be detectable.
So yes, it is still important to save space to other parts.
@@AndersJackson I see your point with adding up with other parts but the deviation of it hitting the ground does not effect the timing of the release switch.
I’m torn between celebrating Martin doing what I had deemed impossible, and crying because I see reflections of my own perfectionism ocd in him
I wish him all the best!!
I couldn’t help thinking the exact same thing, but even during machine 2 livestream era lol… just didn’t comment because thought I’d sound crazy. So easy to get stuck on details, but at least he’s picking his details this time lol.
i would focus on modularity, reparability, and the challenges of live performances / logistically moving the machine if you plan to tour with it.
or, if it's to be a studio instrument, i guess continue down the perfectionist's path
either way, thanks for showing us your progress and your process. it truly is inspirational on many levels.
i love your test setup, and the way you've turned your DAW into precision timing testing + the graphs you're making, etc.
i'm just excited to see you making music, however it winds up.
love that you can chase what makes you happy, and seeing you explore your curiosities, and passions, and sharing your learnings about learning is worth more than any world tour in my eyes. you're already on a world tour with every video!
+1 for Modularity. Each instrument could be designed as self-contained, with its own marbles. Portability, massive reduction in the probability of terminal failure for the whole project, and much easier iteration as designs evolve.
@@revcofe That... well the idea of /not/ mixing them together and /not/ having to separate them entirely avoids the issue of the marble divider altogether. That said its main issue is that a perfect one needs too many dividers to split them up and its too slow. If you were to combine both ideas 'dont mix up all the marbles, but mix up some of them' and then 'use several small marble dividers, then you only need to split the stream for each one 1-3 times and it takes up much less space while retaining some of the idea of the art aspect.
"hold onto your papers dear scholars" instant upvote. Fascinating, love your projects.
for a sec I thought i was watching Two Minute Papers
This was a major niche reference!
I really loved this damn reference man! WHAT A TIME TO BE ALIVE!
What will we see two more machines down the way?
Redditor detected
Something hurts in a weird way seeing a video of MMX at the end playing perfectly good drum beats after a video chasing an impossibly small and unimportant margin.
We could have had a song from that machine but we never will
@@WillHirsch yeah, it's just sad...
The tighter he can get the marble drop the more complex the music he will be able to make. Hopefully chacing these little details will lead to some amazing music though it may be a long way down the road!
Music complexity has FAR more to do with gate speed and the programming wheel. As usual he has the wrong priorities, hence why he has literally given up on the number one priority, making sure the machine has personality.
Not to mention that actually using the MMX would also have brought lot of learnings (on top of really good music). For instance, what about the vibrations on stage coming from the accoustic / the returns ? Maybe the stdev due to this phenomena is already far bigger than 3ms 😢😢😢 but I will still follow and support Martin… even if I think it was a mistake to drop the MMX like this…
Martin, when you combine programming wheel inaccuracies, linkage to gate inaccuracies, and mounting of the whole assembly inaccuracies + vibrations and everything.. you'll be lucky if you get below 20ms across different gates. And 20ms is very good. Just keep that in mind when you invest 3 times the effort for 10% improvement. You'll need to compromise a bit if you're ever to play anything.
Other than that, keep up the good work.
Sound travels at a speed of 330m/s. 10ms are equivalent to a distance of 3.3 meters. A band can play tight music well beyond those distances.
Psychoacoustics mean that a human is absolutely incapable of even hearing a difference of 10ms around a beat.
You're on a goose chase.
He knows that, he’s chasing precision for precisions sake and for a machine level accuracy in music creation
@@johanbogg9158 And as a result, he's spent *how* many years building a machine that sits in a workshop and does nothing instead of producing music?
This!! the millisecond chase is completely insane.
the numbers he is upset about is equivalent to moving a bit away from a speaker...
Has he tried to measure the standard deviation of him self playing to a beat... might end up chopping his arms off....
he is upsett about a _deviation_ of 2 ms!? that is completely insane to be upset about....
He´s still chasing his tail. As long as the goal is "perfect" this project is, again, going nowhere. Sometimes a solution is not just good but good enough. Until Martin accepts this first principle there won´t be a working marble machine.
Well, let's be honest, if you're talking Standard Deviations, it's accuracy not precision.
Martin wants all of his notes to land predictably and reliably; @imrahil is correct that there is a finite amount of precision that can be gained - here Martin goes for 1ms precision, which seems a little much to me, but thank goodness he's not after nano-second precision!
However, as evidenced by the clock drift in his arduino, if he has many arduinos drifting apart at different rates without some ability to monitor, control or adjust for their *inaccuracy* then different harmonies and rhythms will quickly become out of sync.
Some suggestions for more accuracy:
- cooling the system with liquid nitrogen to reduce thermal noise
- dropping the marbles in vaccuum only, as air density, pressure and movement may vary the drop time
- with a vaccum acoustic instruments will be out of question, so replacing the instruments with triggers will help
- triggers have a built in delay due to processing time, so it would be best to use software which knows what sound the marble wants to play at a given time and pretime the sound before the marble hits the trigger
- also cooling the triggers with liquid nitrogen might help with thermal noise and delay
- using optical fibre instead of copper wire would help reduce delay in signal processing
- liguid nitrogen only gets you down to like 100K which if far from perfect, so for more accuracy you could try liquid helium
- friction with the marble gate definitely causes random variance, so it would help to remove mechanical friction by suspending them in strong magnetic field - again liquid helium would come in handy
- speed of sound varies across a room due to density fluctuation, so it would also be good to not play the sound via speakers. Maybe all listeners should wear headphones with optical fibre connection (all cables same length, helium cooled ofc) but you’d still get at least a few picoseconds of variance
Anyone else have any ideas?
Human hearing resolution is 5ms at best, so why stop at 0ms stdev? I would keep going until femtosecond accuracy.
I know this is satirical, but I think the suspension in magnetic field would be pretty bad for precision. Magnetic couplings tend to have a decaying resonant wobble when sharp transitions are involved.
In the spirit of your post, I suggest not suspending the marbles with magnets, but keeping them in an acceleration loop which spins the marbles around and just lets go when the marble's velocity is parallel to gravity.
I think the idea about the trigger delay compensation is good but truthfully if you look into it you realize that even with the best estimations the programming wheel may still be off. I have an idea to solve this:
-replace the programming wheel with a digital file, which is only limited by the clock accuracy of a computer.
-replace the marbles with some kind of programmable sequence to put into the digital file.
-play the resulting file from the computer.
I think this will solve all the potential problems Martin has been experiencing!
This idea is chaos
I love you so much you have no idea!!!!
I am literally DYING OF LAUGHTER 😭🤣😭😂 like 64% of what you said went over my head - scientifically speaking, but if my guess is correct you’re having a slight troll moment??
Honestly either way; if your comment was meant in 100% earnest, or if you were trolling, you TOTALLY MADE MY DAY WITH THIS!!
Im going to re-read your comment later when I’m having a bad day 👍😂 That’s brilliant stuff 👌
I love checking in on this project once in a while. It's completely impractical, utterly insane, there's absolutely no reason for this to happen - except that one man has an incredible vision and the drive to learn whatever is necessary to make it happen. I am so proud of you. Thank you for sharing all of this with us and taking us along with you on your journey.
This concept of as tight as possible martin has been chasing is an interesting one - one at the core of his engineering challenges. I wonder if he ever did a study of popular music to analyze what natural standard deviation is to get comfortable with inaccuracy. My guess - after listening to Christian Hand’s The Twitch Sessions (you should listen if you haven’t) is that even the best musicians probably have significantly more standard deviation than what he is struggling to engineer away.
Yeah I'm no musician but I think slight inaccuracies just makes music sound natural. Like even when producing music digitally people will shift things slightly
If he's planning on dropping the marbles with a piston and an arduino, why not just position the pistons to play the vibraphone directly? We're losing the mechanical aspect and I feel Martin is chasing a perfection that will be unattainable. Perfect is the enemy of good
Didn’t he kinda did that in the last video, when he showed different delays from the right and left channel and demonstrated at which point it gets noticeable, in so showing the „acceptable“ range
@@stokeseta7107 noticable and acceptable is not necessarily the same
@@kameronpeterson3601 The arduino and the eletromagnet linkage is just for testing purposes. The machine will be entirely mechanical.
Going back and forth between Wintergatan and LOOK MUM NO COMPUTER is like looking at two extremes of the same world of music DIY.
Hahahaha oh fs! And notice how many projects lmnc has finished in the same amount of time, there's something to be said for settling for 'good enough'
Sam would have this machine up and working in a month.
@@bazahaza Simone Gertz made a musical bubble wrap machine in a pretty short time. Sure it's not as complex as the MMX. But there is something to be said for settling for "good enough".
To be fair Sam has the same reliability issues with the gig gear. He is remaking modules using industrially printed PCBs. And it's electronics without moving parts!
@@doBobro yeah, I know that MMX is complex af but in order to use it on world tour it needs to be assembled and disassembled twice a week, I'm not a mechanical engineer but I think some clearances and tightness will drift pretty fast. I think the best way is to make it "good enough" and travel with enough spare parts to build another 2-3 machines. And maybe emulate world tour shenanigans to iron out the weakest links that will break first.
and when the whole marble machine 3 is wobbling, vibrating and shaking it doesn't matter.
martin is too perfectionist to ever finish this.
In 3 more years, he gives up on a 90% finished MM3. Four years afer that, MM4 bites the dust. MM5 on the other hand, gets finished surprisingly quickly as it will just a MIDI player.
Meaning we have a lifetime of content to look forward to!
Wobbling + inaccurate gates will result in worse accuracy than wobbling alone
The purpose is to minimize the total deviation after all the systems add up
this is exactly the reason to get accurate first, otherwise you're just stacking inaccuracies together.
"Don't worry about inaccuracies because it'll get worse anyway"
@@jslavertu and if one part is late and the next is late, you're 10ms late. It's +-10ms, you can't expect them to cancel out all the time. Sometimes both will stack early and other times late
There's something about when you dig deep enough to get "above the problem". Don't let the accusation of perfectionism get you down. It's clear to me that you've transcended endlessly perfecting escapements and now you are seeing the true nature of the problem. Enlightenment can be found anywhere, thank you for showing how you found it by loosing your marbles.
This type of insane obsession with detail is starting to remind me of Miura-sensei, the mangaka of Berserk. His release schedule slowed a lot towards the end because he felt the need to go back and edit *individual pixels* of his scanned manga pages even though his editor constantly told him that no one could see the difference. Really hope Martin snaps out of it so it doesn't end the same way either..
Rip Miura-sensei.
Honestly, I felt a little bit of excitement seeing you work on this again, until you said at 1:44 that 3.9 MS is not good enough. It made me physically eye roll because then I realized that this project will never ever get finished, as you will always be chasing perfection. Seems like you are losing the majority of your followers interest because of this. I've followed you from the beginning and want nothing more than for you to succeed, but please Martin, follow the advice from others here, and ease up on your expectations.
Chasing perfection is a fool's errand, but Martin is no fool so we appear to have a paradox.
However he doesn't seem to want to listen to anyone in the comments when they contradict his drive for perfectionism so I'm afraid he's just going to have to learn for himself. Some people won't be told, they can only learn from experience ......... usually.
Pain is temporary, glory is forever!
1 ms for the entire system or scrap it and start over. Best way to get them RUclips moneys.
I watched the entire MMX project. 99.992% perfection is why the last one failed. I'm not sure why that isn't good enough for it to never have played a song full song. by adding the finale parts. I assume personally, Finishing the project would end the journey. My logical speculation base of what I seen is, The closer the MMx was to being done the more depressed he became. most likely subcutaneously finishing that project making him depressed, so the idea of starting over made him happy and comfortable. This project is about the journey and not the finish line. I spent years checking in. And once I heard the MM 3.0 idea I unsubscribed I was so invested into seeing a man slay his dragons! And threw will power defeat a huge project.. I felt hurt he gave up... I'm not impressed by the new 100% standard because 99.992% perfection ended the last one.. There is beauty in imperfections. Peer pressure from engineers 100% perfection has gotten into his head. I add +/- 5ms to my medi for the humanize effect. I feel little hope watching the part from-to 1:13 -16:10...
@@SlartiMarvinbartfast Martin reads-and shares with us-some excellent books! I’m sure he knows the name Douglas Hofstadter, for example. Also definitely a paradox.
this machine is both amazing and also seems like it will never be finished.
I pay to watch a man's journey, getting a working machine seems like a secondary concern.
@@mormacil It is, however, important to keep one's expectations realistic and not chase perfection as hard as Martin does. Being sad about 4 ms instead of 2 ms in music precision is like discarding an otherwise acceptable cup of tea because it had 1 gram of sugar too much added into it.
@@pizzarella985 Depends if your goal is setting tea for a visiting friend or perfecting your skills in brewing tea or making the best possible tea.
One could in the latter case go all out, trying to remove imperfections in the water and air that could in some way impact the flavor of the tea. It would be insane overkill for daily use but daily use might not be the goal.
@@mormacil enabler!
Well he deconstructed the other one. The MMX is dead. It's honestly depressing
More than anything, I think the marble machine projects have all been fantastic sorta outsider art. I hope you make a third one, and I hope it ends up as something you can be proud of. I don't think you really owe anyone anything and only you really know what this machine needs to be. I can totally vibe with being a perfectionist even when other people think it's completely unnecessary - it's just part of the artistic process!
That’s a fresh perspective! Martin’s fanbase generally prefers to shit on him nowadays 😂😢
Personally, I see both sides of the story.
Martin was the art all along!
I'm not sure I can get invested in this again. I'm still healing from the last one 😂
Just watching the makings of something new and moving towards something that can actually work is fun. maybe it wont be soon, but watching these videos is both fun and informational.
I thought the same thing at first, but then it just happened. Boom! I'm invested again :D
Pain is temporary…
tbh i'm here for the process. doesn't matter if it ever gets finished. it's nice to indulge in perfectionism, even vicariously.
I used to sit down every Wednesday during lunch and middle school and watch the weekly update video
I admire your tenacity on this project, I used to love to wait every week for your episodes and even bought merch of the MMX, but now I'm failing to find awe on this level of perfectionism, this feels like seeking for something you won't find on a professional musician or in an engineering project, perfection will always feel like trying to chase infinity.
Exactly! I used to follow the mmx project closely and watched every episode with great joy. As soon as the whole livestream stuff started happening, I could just not get myself to watch. In the end this sympathetic project of a big marble machine playing music turned into a lame hunt for accuracy which was never the point in my opinion. Of course it's Martin who decides the direction, but it's exactly the direction that makes me wholly uninterested.
Same concerns here… losing interest in the project
It was always a joy when martin uploaded. Now, since the last 2 videos, i am not happy as i was. I understand, that the purpose/goal is no more music, no more exciting spinning, rotating devices, but an impossible mission to 0ms delays, to perfection in any way. NO HUMAN can EVER be perfect, no one WILL hear these miliseconds. I hope we will go back to the roots, to nice music and exciting for the eyes too. Love and hope from germany
If you can't see the beauty in surpassing normal human musical ability through the application of mechanical ingenuity, then I think you are missing most of the point of these marble machines.
It hurts how close he got to a working MMX before throwing his goals completely out of reach. He was finally going to get back into music-making and now we’re stuck chasing milliseconds like a robot chasing the end of a rainbow
Yeah… but, at least from my understanding, there really isn’t much that can be done with the MMX (can’t handle marble throughput in actual music- the magnetic lift just doesn’t have the capacity) unless he rebuilds a few major parts- which would present more problems. I can really relate the the feeling of hopelessness when you discover a major flaw in your project which is almost done..tho the best choice would be to try your best to save it. I really hope he does that for the MMX, it’s a great machine! I hope he doesn’t lose the art that started this all
Is it just me or does it sound like he lost the magical spark he had when he was chasing the crazy dream music machine? It now he just sounds like an engineering robot filling out spreadsheets and checking code. 🥺
The MMX couldn't play a full song - let alone a concert - or a full tour! Martin could have recorded an album using samples and major repairs during the process. (That is how the original Marble Machine "played" after all) But the frustration was hampering his creativity.
The video series of him building includes a LOT of original music. But notice how much other instruments featured in each song? The MMX was never a solo instrument in those creations. Many times it was barely a sample or a backing track. This project is all about getting a machine that CAN play like whole band by dropping marbles.
all he had to do was write one bloody song and play it on the MMX and put it on youtube as a payoff to all the people who supported him with time and money. I can't even get myself to care about these videos anymore.
@@ssl3546 then you're the problem
Martin is personification of Morty when he discovered true level.
Martin needs a bit of memory wipe so we can get this going again.
Great to see more videos and I really look forward to the final product.
While your goal might be designing a perfect marble instrument, I think the true blessing has been the art you've created along the way, and the fact you've been able to share it with us
Don’t ever give up on perfecting your marble machine, Martin! Even if it just becomes a side project hobby, please don’t ever stop on your build and design. You are creating a beautiful machine that will last centuries! You are creating a mechanical wonder for thousands of generations to come! ❤
During my time in the engineering profession, one of the more important lessons i've learned is to prioritize:
1. Finish in time.
2. Make it work as good as possible
3. Make it look nice.
I believe Martin started with nr 3 on top. He now has moved up to number two, but my prediction is that nothing will come out of this endeavor before he learns to include point nr one aswell.
I love watching Martin's videos exactly because he doesn't sacrifice quality for arbitrary deadlines.
"There's never time to do it right, but there's always time to do it again."
@@AndrewCooks the thing is you can’t know what right is without a preliminary version of the rest of the marble machine. How does this gate react under the vibrations of the rest of the machine? How does it react when the marbles are jostling? Sometimes good enough is good enough, especially when the problem isn’t well defined yet.
Lmfao look at this guy. The Marble gates are a central mechanism and he's solved them to perfection in 2 weeks. That's a far cry from the months and thousands he's previously spent. Get over yourselves lmao he's doing this right.
This isn't an engineering project however. If this were an engineering project he would give up on the marbles completely because it violates the various principles about efficiency.
@@gideonrabson4600 This guy knows what he's talking about.
Right now Marton is taking the 'make everything as precise as possible, and the end result will also be precise' approach. This can work, but is very laborious and there's a cost to quality. More than what's requires is actually detrimental, as it takes time, energy and resources away from other parts of the design, or from moving on to other designs.
The goal here isn't to make the most precise marble gate, it appears, but rather to make a better marble instrument. That prpbably entails a more precise gate than previously, but there can be emergent properties that arise when you begin coupling things together in a larger system that need to be controlled for that could very well make this extra precision unneccessary, or worse, actually detrimental.
Imagine he gets a perfectly dialed in gate but it somehow induces a resonance with another part where a bit more scatter would've avoided it!
If this crazy precision is something you are really passionate about, I'm glad to see you going after it. Just as long as you're doing this because it's something you want to do, not just because you feel like you _have_ to in order to make the ultimate marble machine, lest it all be in vain.
Sometimes, imperfections are okay... but if you really think you can do this, and it's what you want to do, then I'm glad to get to watch your journey.
After all, it's not our dream you're chasing. It's yours. I can only hope we'll one day get to see it in all its glory.
Martin, I don’t even know that human musicians can be that accurate.
They're not even *close*.
I just looked it up, and really got drummers seem to be able to hit +/- 5 milliseconds, it seems.
@@MoritzvonSchweinitz which I’d think would be way more than enough for having tight music.
@@MoritzvonSchweinitz that's really jawdroppingly accurate from an outsider's perspective, sheesh.
(i play rhythm games and my standard deviation always sits around ±20ms :dead:)
Well if we look beyond the numbers, its just martin completely removing one factor from the inevitable future of him debugging his machine when something performs suboptimally. The gate wasnt that big of an endeavor, and it also gave him the peace of mind. I dont think you creatives get how many sources of errors there really were in his previous attempt, and how he led himself to a deadlock by the end.
This is gonna take forever if he wants the entire band this tight
Don't you understand yet? That's the point! Chasing a rabbit and getting views.
@@ZbyszekJot If that was really his intent I'd be much happier about this, but it isn't.
@@ZbyszekJot martin is getting addicted to the process that he is slowly and subconsciously denying the thought of finishing the machine
@@r3ll282 HOLY JESUS CHRIST, HOW MANY TIMES DO I HAVE TO SCREAM INTO THE VOID THAT IS THE INTERNET BEFORE PEOPLE LEARN THAT THE SUBCONSCIOUS IS STUPID PSEUDOSCIENCE?? THERE IS NO GODDAMN EVIDENCE THAT THERE IS A SUBCONSCIOUS! IT'S A FOOLISH IDEA CREATED BY GODDAMN SIGMUND FREUD! THERE IS NO GODDAMN REASON TO BELIEVE IT EXISTS! THE ONLY REASON PEOPLE BELIEVE IT EXISTS IS BECAUSE PEOPLE MADE BAD DECISIONS! LOOK, THAT GUY MADE A BAD DECISION AGAINST HIS BETTER JUDGEMENT! IT COULDN'T POSSIBLY BE THAT HES IMPULSIVE, OR DIDNT REALISE THE TRUE SCALE OF HIS ACTIONS, OR THAT HE MIGHT JUST VALUE SHORT TERM PLEASURE OVER LONG TERM HEALTH, NO, IT'S SOME GODDAMN BOOGEYMAN THAT CONTROLS HIM! ALSO, THIS BOOGEYMAN ALWAYS CHOOSES THE WORST GODDAMN OPTIONS, LIKE STAYING WITH AN ABUSER! AND IT'S ALSO SUPER-INTELLIGENT AND CAN PERFECTLY READ ANOTHER PERSON BECAUSE IT'S THE BOOGEYMAN! IT CAN DO ANYTHING IT NEEDS TO ASLONG AS IT BOOGEYS!
I AM SO GODDAMN SICK AND TIRED OF PEOPLE ACCEPTING THE SUBCONSCIOUS AS FACT! ESPECIALLY SINCE **SIGMUND GODDAMN FREUD CAME UP WITH IT!**
@@r3ll282 ALSO, AMAZING HOW FROM A FEW CURATED VIDEOS WHERE HE DOESNT EXPRESS ANY DISTRESS, OR ANGER TOWARDS THE MACHINE, YOU CAN MAKE A PERFECT JUDGEMENT ABOUT HIS PSYCHOLOGY! MAKES ME WONDER WHY THOSE PSYCHIATRISTS CAN TAKE WEEKS TO DIAGNOSE SOMEONE AND NOT ALWAYS GET IT RIGHT! GUESS YOURE JUST BUILT DIFFERENT HUH?
AND WHY IS IT SO BAD HE IS PURSUING PERFECTION? BECAUSE YOU WANT TO HEAR THE MUSIC ALREADY? BECAUSE IT'S "UnHeAlThY" TO BE GREATLY INVESTED IN SOMETHING? DO YOU KNOW HOW MANY THINGS WOULD BE WORSE IF WE ALL FOLLOWED YOUR STUPID GODDAMN DEFEATIST MINDSET?
MASTERPIECES HAVE SOMETIMES TAKEN DECADES TO CREATE! CAN YOU NAME A SINGLE CLASSICAL SCULPTURE THAT HASN'T BEEN MADE IN MORE THAN A YEAR? MARTIN DOESNT WANT TO BE KNOWN AS A MAN WHO CREATED *A* MARBLE MACHINE, HE WANTS TO BE KNOW AS THE MAN WHO BUILT *THE* MARBLE MACHINE, A PIECE OF ART SO AMAZING IN IT'S ATTENTION TO DETAIL AND THE QUALITY IN ITS CRAFT IT'S NOTABLE FOR MILLENIA TO COME! THAT IS AN ASPIRATION, A DETERMINATION, TO BE THE BEST YOU CAN BE, THE GREATEST, THAT CAN ONLY BE ADMIRED! AND YET YOU SPIT ON HIS YEARS OF EFFORT, SEE HIS HARD WORK AS A CURSE? SO DECADENT YOU ARE THAT YOU SEE A MASTER AND NAME HIM A FOOL? BECAUSE IT'S "GOOD ENOUGH" AND DOESN'T NEED TO BE IMPROVED? BECAUSE YOUR STANDARDS ARE SO LOW YOU THINK STRIVING FOR EXCELLENCE IS A WASTE OF TIME? WELL SCREW YOU!
AND YES, I AM FULLY AWARE OF HOW RUDE THIS COMMENT HAS BEEN, AND I STAND BY EVERY SINGLE GODDAMN INSULT, BECAUSE YOU ARE AN HONORLESS COCKROACH WHO SEES NO VALUE IN IMPROVEMENT!
These investigations go far away from common music and I SO love it. Your videos are pure engineering, mathematician and statistics experience, and as for me, non-musician, this content is much more interesting than just listening to your (of course amazing too) music. So go on, thank you so much for keeping in course and hello from Russia!
Love the subtle reference "hold on to your paper scholars" 🥰. Love your work Martin /Martin
What a time to be alive
Just wait for two more videos down the line.
Wow!
took me two minutes to realize.,. . it was tooooo subtle
@@mm-hl7gh This got me. xD
I'm happy that you're back Martin, but don't lose your marbles with the timing thing, I personally think that music isn't about being in perfect sync always, it's important, don't get me wrong but, music it's about having fun, expressing yourself, delivering a message.
We don't care if the message reaches 1ms, 2ms or even 5ms late, we care that the message it's meaningful.
Make that message have a meaning. We'll be here to listen to it.
I noticed that pun :)
In all seriousness, I 100% agree. What I love about the original marble machine (as well as most other arts) are the "imperfections".
It feels natural, emotional, human. If it's 100% perfect, it just feels stagnant. In the end, it's soul & intention that drives the whole narrative.
@@gomichow Exactly! The beauty it's in the things that makes it unique
@@gomichow The thing is that at the end of the day, the marble machine will rattle and vibrate. Wind can and most likely will be a factor during a play, and so many other possible cases that I can't think of. So imperfections are still very much a thing with this project.
What I see is that same "it feels inhuman" argument every single time, and not once do you guys think about what he wants to convey with this project but rather what you want from him to convey to you. Kinda selfish if you ask me. Don't take this as an attack at you specifically, but I just dislike that take.
@@kyucumbear I agree that it's kind of selfish to think that he's making this for us, well, in a way, the viewers are one of the main pillars that support this project, but we can't force him to change the way he thinks or the vision that Martin has with this machine.
My take it's more focused around the fact that, if, only if the way his vision is focused around making a machine to show to the world, to impress people and to make a instrument that will be travelling around the planet, I think that focusing and making all this effort, tiny changes, all the measurements, redesign's and all the things that he thinks that are essential, will lose the battle again like happened with mmx, because even if you can control 50,100,200 variables inside the machine to be absolutely perfect, there will be hundreds, thousands of external variables that he just can't control and it's possible that because he just can't control everything, he will get very frustrated, because he wants to make the best machine, for us, for him or for the instrument, either way, it's not a healthy approach to make a project.
It's important to have fun, to be creative, to make it your own, break a few rules, upset some people (us included) because of the decisions he took. The optimal way isn't always the best way in life, in my humble opinion, the best way in your life, proyects, plans, it's your own way
Cheerful Martin makes me cheerful. As always, I want to emphasize that the value of your videos is your videos, not some end result. It is the evolution of design and discussion of basic principles that hold me enthralled.
1:02 I couldn't tell the difference in the sound but I could definably tell that the sound was coming from the left first even at 1.46ms.
I'm using a 30y+ old Sony MDR-V600 headphone. That surely helped a lot. If it was on normal speakers I'm 100% sure that I wouldn't not be able to notice the side that was coming first.
I got really confused for a second when you said "hold onto your papers scholars". Your deep dives into design is amazing to watch and is incredibly inspiring. Heres to MM3.
Awesome work, and yet premature optimization is the root of all evil
Agreed. Sometimes it's better to work on all aspects to find out where the bottleneck is
3:10 Not in this case. It's Martin's third approach to this and he already knows what issues are the most important to solve.
Optimizing later like on the MMX was and would be harder
@@maticz3923 true, and at the same time, one would not have to put all the pieces together before optimization. But by working on a broader range of parts before optimizing one might make connections that will help anticipate later problems. But as a disclaimer, I am talking out my ass.
@@maticz3923 He doesn't know the most important issues, if this is what he's currently concerned about.
I disagree in this case. Early optimization may be the road to succes, because he knows what he wants now and can test and optimize the design before losing time on building a machine that won't work because the design isn't optimized (like the MMX).
Wow this is still going? Good. Now for the important part...NO ONE IS GOING TO NOTICE UNDER 4 ms DEVIATION! Our tvs don't come close to that for gaming, and no band is ever playing that tight, it's humanly impossible to do that for a whole song anyway.
this fails to take into account tolerance stacking. there's multiple parts of the machine that have timing variance. So whilst you're correct that 4ms deviation doesn't matter, 4+4+4+4+4 ms does
Yeah this is kinda what I'm feeling haha
While I agree that most people won't notice anything below 5ms jitter there is a difference in the game console to TV scenario because that is a constant delay so it won't affect things internal timing between notes. Jitter is worse than constant delay for music. Also, standard deviation might be all fine and so but for measuring over all performance the 95th+ percentile can be just as important to keep track of because you want to know that you have acceptable performance at the edges as well.
Contemporary games are typically designed to allow for some input and output latencies. An example would be allowing a character to start a jump in the air 50+ ms after they have left a platform. You can also notice this when you try to play some pixel perfect platform games from the 80's on some modern TVs, the timing is so tight that the games get almost impossible to play.
@@DeanLawrence_ftw While this is true, Martin doesn't seem to address this clearly in recent videos.
@@DeanLawrence_ftw I am sure this is why Martin is doing this knowing his past attempts. Even if each part has no deviation in isolated testing, when he puts them together into a whole system on a newly designed Marble Machine, there will be deviation introduced through the sheer complexity of the system. Eliminating as much as possible up front in the design process is important for that reason.
The second marble machine didn’t suck, you just kept moving the goal posts back more and more until you managed to set goals that were totally unreasonable for a complex mechanical machine. Think of watches... The finest Swiss watch on earth has allowances for beat errors... The finest horological pieces in the world will gain or lose seconds per day... perfect accuracy isn’t the point. They aren’t nor ever will be perfect... If you want digital accuracy, make digital music on a laptop and be done with this nonsense.
Well said
Right? The marble machine X is freaking BEAUTIFUL! Its a work of art! Its freaking breathtaking!
You're a musician and an artist Marvin, do you really want to spend your days making sure marbles drop with 0.0005 ms accuracy?
F*CK THAT! we don't give a sh*t about that, and neither should you!
He should at least perform one track with MMX
Word
He wanted to perform live music with it... It absolutely did suck for that purpose.
It's important to not see the failures as such; as failure. Rather, they should be seen with pride, of something you attempted and didn't work to build better from it. Be proud of your attempts! Be proud of your work! Learn from your mistakes and one day, materialize this dream that felt so far away before! Bon courage, Martin! You can do it! I love to see you're never giving up!
First, I'm really glad that you're working on it again, and I hope that you have a lot of success.
Second, I know you didn't like your first version, but several hundred of those many views you got were from my family, so it's fine that it wouldn't be great for a concert, but it was also special and beautiful, so don't focus on the imperfections, because no one else did.
Last, lots of other people have said it, "it doesn't have to be perfect, blah blah" and they are correct, but I understand that it's as much art as it is music, and so you for sure want it to be as close to the perfect as you can. But imperfect and finished is better than never finished because it can never be perfect.
Congrats on figuring out the timing solution.
Just a friendly reminder to not forget the pressure that gets built up by all the marbles pushing down on each other as it could affect the timing of the marbles. HOWEVER that being said, anything under 5 ms is not going to be noticeable to anyone Martin. You will ALWAYS be able to find some flaws with the machine. So don’t stress to make it absolutely perfect going forward. I think this kind of deep dive at the start of the project is the way to go. Just don’t get too caught up in the details when you do build another machine. Even if you don’t, I’m still here to watch and support whatever you do :)
The pressure does not change anything as everything accelerates downwards at the same rate. I can stack fifteen tanks on a marble and it will fall at the same speed
5 ms is so insanely tight. That's the timing on many drum transient attack sounds before they reach their peak amplitude. It's stupidly tight.
I make electronic music and deliberately give my drum programming more loose timing than that to give it a bit of a "human" feel, and it is STILL very robotic. 5-20 ms is fine most of the time.
At this point he should just sample the sounds and make his music in Ableton. This misses the entire point of creating a physical mechanical machine that makes music imo.
The musician in me says DAWs have functions to remove perfection from MIDI-Tracks to make them sound more realistic.
The engineer in me, as i build measurement devices says: if you have no dispersion in your data, your resolution is not high enough 😉
As someone who believes I say: I missed you Martin! Thanks for coming back! 😀
yeah zero milliseconds is not zero microseconds... but let's not encourage Martin in the wrong direction! lol.
It's fine, actually playing on a full machine will surely have enough going on to remove absolute perfection on its own. :P
Marble machine 2 was already almost done, it didn't need to be perfect or be able to be moved, we should have heard some songs from it.
We've heard some songs on it.
The MMX had a few fundamental flaws that prevented it from playing a song utilizing all instruments.
The marble divider on top let the marbles traverse too slowly, leading to lower gates to be starved of marbles. The marble lift gears had the wrong gear ratio, with the lower one lifting more marbles than the upper could lift, leading a full machine to spill many marbles. He redesigned the upper part and when he wanted to connect the marble divider to the marble gates, he noticed he didn't have enough space for the connections.
Originally Martin wanted to finish it to play a last song for RUclips but at the marble divider to gates space issue, he quit.
It really suffered more issues than it seemed from the videos.
@@Caesim9 If he could make the first one work even if it took splitting clips he sure as hell could make the second one work to make some songs on it.
@@Treksh the first one never worked. The famous RUclips clip was composed of probably 60 different takes and fixes on the fly and then edited together afterwards. That video shows off a concept, an idea. Not a working machine. It could not play a single tune.
@@tjalve1 I've never heard this, though I've often suspected that was the case. Do you have a source I could share?
@@tjalve1 ... and yet he made a massively popular music video with it. As an absolute minimum he could have programmed MMX to play the same song, and edit it together if necessary.
Love the Two Minute Papers "Hold onto your papers" reference :)
Yes!!!! It made me smile so hard
I just love that your enjoying yourself here, no pressure, no time tables, just you experimenting!
sure "no timetables" while literally hundred of people keep giving him MONEY each month. yeah you could stretch this out for YEARS... oh wait, he already has. And just as he got close, decided to scrap it and start again, for even more years of free money while playing with toys.
@@rafezetter8003 he stopped taking patreon ages ago, I didn't give any money, and that's a risk everyone who did took knowing that, it's their call.
@@rafezetter8003 You’re calling this “playing with toys”? Seriously?
@@potatosmasher1072 Yes. I know you may think this is serious engineering, but my problem with this is as I stated. If Martin was actually serious about getting this done he would be doing the same thing as he did the first time with the wooden one, get a WORKING version completed - doesn't have to be PERFECT, it just needs to WORK - after that he can work on the iterative process of streamlining efficiency. That is what a true engineer does.
The problem is Martin has brought his "creative" brain to a LOGIC problem - he wasted so much time trying to make it look "cool" that he was building problems into the systems - and now he's paying the price, but he's just diverted from one "cool" situation to another.
It's all "cool" and not "what do I need to do to get this working, I can make it look cool later".
Every single engineer on Earth will tell you a simple fact of engineering, you never get it right the 1st time or the 3rd, or 5th, and the more complex and moving parts it has, the odds against you go up exponentially; if you try to get it 100% "cool" + 100% "efficient" the first time, YOU WILL FAIL. Brutally.
In the tens of thousands of years of humanity, it has never been done - the simple lightbulb took something like 10,000 attempts before a working version was found. A bloody basic lightbulb, 10,000 times, think about that for a moment.
Every single thing ever made in human history from the humble fork to a rocketship has been through dozens or hundreds or thousands of iterative changes, and Martin thinks he can get it 100% right and 100% cool on only his 3rd try after his 2nd try didn't even reach completion?
Give me a break, he's deluding himself, and all of you. I'm probably the only person among the rest of his simpering fans that will actually say the brutal truth, the truth that might; just might, get him to stop following the "perfection" rabbit down the hole and instead FINISH IT.
Martin didn't even succeed at finishing the Machine X, so he only got PARTIAL DATA from it, and engineering with only partial data will mean only one thing - the next iteration still has flaws, and if Martin follows the same path as last time, this next machine will not be completed either, because more flaws will show up, and so...he will start again.... and again.... and again.
He could spend a lifetime chasing it, and some people have.
IN ENGINEERING IT'S IMPOSSIBLE TO BUILD SOMETHING BETTER, WHEN YOU NEVER "FINISH" THE PREVIOUS VERSION.
So yes, emphatically yes, Martin is "playing with toys", fancy expensive and technologically more advanced toys, but toys all the same.
The only advice I would give to Martin is make the new system modular, so improvements can be made without having to restart from scratch, again, as he is now.
Martin owes it to the people who gave him money or thier professional time to get a completed version, otherwise it's what we call "vapourware" - all idea's but no solid outcome, in which case he is a fraud.
@@superkaboose1066 People thought they were going to see a project from start to finish not, start, get 3/5 (many years later) then abandon it to start again.
Isn't 7ms deviation considered excellent for even high end audio systems?
One thing that he not mentioning in the video is that total variation on timing has to be considered, but he is mesuring just the variation due to the gate.
@@rafaelschipiura9865 That's probably why he's chasing perfection here, so it will be less of an issue further down the road.
Please note that this is only the marble gate and that inaccuracy in the machine will add up.
4 ms isn't noticeable, but if you lose that amount a couple of times you end up with audible delays.
So he might seem crazy in designing a gate that is super accurate, but good design now may lead to better results in the end.
@@EaglePicking the inaccuracy is only within 1 beat, they don't compound(its all tied to a master beat of sorts in the machine, the mechanism that plays) and the gates tied to the release. So whatever it is it can only add up between the the program wheel/device and the gate. The program wheels the real problem anyway in portability and keeping it as an 'instrument' instead of a music box.
The keeping it as an instrument part kinda adds up something anyway that has variation.
Its just tinkering with a gate and spending manual time measuring an automable thing to measure, to pass the time.
Edit: also the 'good design now for better result later' has already been on meme level for like 3 years(not functioning as intended for having a result)
I can appreciate the dedication to excellence. True that all the inaccuracies will be additive in nature.
Also bonus Kudos for realizing part of the deviation was in the measuring.
He is too far down the rabbit hole at this point. He hears the faint murmur of those on the surface telling him to end this quest for the impossible but it is over powered by the perfectionist within. I wish you the best Martin and hope you don't lose yourself along the way.
As soon as I saw design 8 with the extra bit holding the marbles after the escapement wheel, my mind instantly went to the old design with the kinetic fingers lmao
This video has given me so much hope and really showcased the growth in Martins ability. Seeing Martin blast through something that caused so much trouble prevoously and seeing how deliberate he was with each of the design iterations was rad. So much of his earlier design work was "because it looks good" but that lead to so much rework later on. Even tho Martin is taking baby steps now, I wouldn't be surprised if he has a new machine designed and built within a year.
As far as the comments here go, I would just like to say that I'm loving the engineering aspect of this. It's very fun to see, and the enthusiasm is a pleasure. I believe that the "it's good enough, don't waste your time" are both sensible, but also besides the point. When all things are put together, errors will compound. Having each individual designed well, will make it motivating to put together. I'm sure you're mindful of the adage that "perfect is the enemy of good".
It's pretty bad engineering though. Focusing on single aspect and trying to optimise them to perfection before you even know what it's going to connect to is a waste of time. Part of the reasoning he gave for abandoning the old machine was that he kept having to rework old parts when adding new ones. He clearly leant nothing from that experience.
You don't go straight to making a perfect machine piece by piece, you first make a prototype, but he refuses to make a prototype of the whole machine, instead he wants it to be perfect as soon as all the pieces are put together.
@@emmastrange5557 I don't disagree with you. If you consider engineering as "get the job done sufficiently well", as to move on and solve the bigger picture. Then you need a clear set of constraints as to what is "good enough". And, I believe Martin is way off here. Even 10 ms would arguably be sufficient. However, that aside, one can appreciate the engineering in improving a system.
And, I do think that he is up front about it. He isn't solving one part of a whole. He is solving each part separately, where the goal itself is a "let's see how good we can make them".
The MMX thing was a bit of a sad story. But this looks like having fun designing contraptions. The goal is the design of the contraption itself, and not "build a machine to play music".
If you disregard the previous history and mistakes. And just see this video on "how to perfect the timing for dropping marbles", I most certainly enjoyed the engineering aspect of it. And, I wonder if its not unfair to force a different context onto this work.
My suggestion, if you end up making a third marble machine, is to make the small parts (like the gates) which could break snap onto the larger mechanism, so that if one ever breaks during a performance you can just quickly replace it and figure out why it broke later
So glad to see your creative spark is back Martin!!
Im so damn happy that this channel is back with the regular uploads as it is one of the best engineering channels out there. It doesn't even matter that you are starting a new project Martin, showing these perfect examples of the benefits of designing to first principles, robust design, useful testing and data analysis, all while still keeping it super interesting for us viewers is just a joy to watch. Keep it up.
Second that:)
Regular? Hold your horses after 2 videos in 2 weeks.
@@reneeschke i guess Ben is talking about the live streams. They were more regular, but less regular ;-)
This is the correct comment. Ignore the haters.
They think the purpose of the marble machine is to be a musical instrument, while in reality, it's true purpose is to generate a collection of design videos for us to enjoy, and cool merch for us to buy.
Silly haters.
@@iamiggie its understandable though. Martin is a musician, and his fame came from releasing music on a marble machine, and the goal of the next one was to create music and tour with it. There are two large segments of his audience - the musicians and the engineers - and there isn't much overlap. The former category are somewhat disenfranchised by these pursuits of perfection and obsession over minutia since they would prefer to simply see art being made, with all its quirks and flaws. Meanwhile, the latter category is perfectly happy to come along for the ride and are more amenable to the deep dives into design and desire to achieve incredibly high levels of precision.
Nice improvements- a standard deviation of zero just means the error is below the resolution of your measurement system. It can never be zero.
Martins perfectionism has been making me think he is slowly going mad and then he talked to the Wilson's at the end and confirmed it.
If you are willing to reach this accurracy you must check the weight of your marbles individually
pls don't give him more ideas
Marble drop speed is not strongly affected by marble weight
@@Pidrittel I agree but it's interaction with the machine is. Levers and motors care a great deal about weight deviation.
Also he needs to measure, calculate and account for differences in strength in local gravity (gravity is not a constant on Earth, it has ever so tiny changes, thus it's value changes every yoctometer), and put the machine in absolute vacuum to prevent air molecules from causing uneven resistance and drag (molecules in gasses aren't spread evenly), causing a deviation as big as femtosecond. Can't wait for him to resolve all the issues like this in design phase and prototyping and move on to build the machine in the next few hundred eons when science and technology finally allow him to live in virtual reality where all of that is finally possible. /s
awesome idea ! just a little and very accurate scale before each marble dropper, and a simple mechanic to move the whole dropper up and down to compensate for the differences.
"hold on to your papers dear fellow scholars"
Awsome little reference there.... i love 2MinuetPapers
Károly Zsolnai-Fehér always brightens my day. It's nice to see other's appreciating him too, even if he is a "big youtuber" so to speak
I'm glad Martin is back to making normal videos again! :-)
"normal"
@@mattcy6591 "to"
Nice video! It is very informative! I started to watch your videos as soon as first video arrives, since then, just the first video I watched dozens of times, and using as an background music to meditate.
Im am a soviet engender.
I assumebyou couldnuse all your 1 trough 8 gates, with even better performed results, by adjusting un-delay or in another word, drop your ball earlier, that way you can use potentiometers to adjust all from - to +
I found your original marble machine video last night and remembered I'd seen it years ago, I've since spent the last day watching as much of your progress and recent videos as possible. You have such amazing energy and passion for this project, I hope the 3rd machine does end up being viable and turns out to be all you hope it to be in the end. Keep at it, many more small steps and revelations to go, thank you for inviting us all along for the journey :D
Awesome work Martin! As an engineer and a musician these videos stimulate both halves of my brain. 😁
I love how math and music work together flawlessly
You only have to adjust your measuring until everything fits perfectly… 😉
music = math
Honestly, I can’t hear a noticeable delay until you played the 50ms example. Not sure how vital this is going forward- maybe it somehow becomes more noticeable with many notes? Might be a goose chase if not.
Haven’t finished the video yet, so maybe Martin addresses this, but I think he should define the acceptable “good enough” deviation. He uses comparatives without having a defined requirement/goal.
He hasn't made a smart goal for each aspect of the machine, he needs to start there
@@kameronpeterson3601 The purpose of the marble machine is a sonic orgasm.)
My english is really bad, sorry.
Get your ears adjusted! ;)
A 10ms delay is clearly audible as a double hit and a smaller delay is audible as a “wider”/longer hit - with my ears, at least.
@@peterfireflylund Got to remember there will also be huge variation when dealing with that level of fussiness in the quality of the sound delivered by YT and how the viewers audio system processes it. If your PC is running at CD bit rate but YT isn't even more than 50ms could be lost quite easily to the conversion processes, and if the room has some echo again the difference can easily be lost, we don't all live in sound deadening music studio boxes...
It is an "optimal settings" situation. There will be many factors adding to the delay and the tighter it is initially, the better the end result will be. And in the best case, everything holds together well enough for the programming to control the output to a degree of inaudible delay. Imagine a drummer how does not play a drum but nudges another drummer who plays the drum. Each of them has a delay of 10 ms, the precision of a pro studio musician drummer. But together, they end up with 20 ms. The machine will have a lot of drummers nudging each other.
I keep watching these videos in the hope they go back to how they used to be and they keep being less interesting. He used to be a crazy, genius musician creating an amazing machine. Now it is like watching an engineering student who has just read the latest book and thinks it is everything.
I know I’m late to the party on this video but, I’ve been watching you on and off since you started the cad one, and honestly it’s incredible how far you’ve come. I mainly keep up-to-date with it because of your passion for the project it’s fun to see you succeed.
I know you probably won’t see this but I just wanted to let you know how amazing your project is
Martin, I love your dedication. Rooting for your second challenge!
Idea: analyze the accuracy of timing in the tracks Wintergatan has recorded previously to see what kind of deviation is there. I’ll bet you $100 it’s over 10ms!
and nobody batted an eye. We all loved it! Martin is just chasing too much perfection here
yeah but this is just the gate, he didn't even want 0ms accuracy, it just happened. What he wants is to be tight enough so the rest of the machine's parts have margin of error, because if 20 parts have 3ms delay, you end up with 60ms. The more accurate each part is, the less it will add up in the end, that's why he made a 5 point list he wants to get accurate before building a whole machine that isn't :(
I applaud your commitment, its been really fun watching this journey!
Standard deviation itself is not the most intuitive metric to report. I would recommend 95% confidence interval (which is calculated from standard deviation). That way you can say “95% of marbles fall within X ms”
Love the "hold on to your papers fellow scholars" I am a big fan of two minute papers XD
I can't wait for the moment when Martin realizes that too tight music is sterile and uninteresting and then he goes on for another 3 years to make a mechanical randomizer to make it sound more human.
Not 3 minutes into the video I’m having the same thought😂
Yeah especially because those inaccuracies are what make music sound good. Take a jazz swing for example. The reason it sounds good is because of it's lack of precision. It can be best described as a triplet based rhythm but it's not quite.
Eh, I don't agree. I think the real substance of music and what makes it interesting is in the composition. The making of something creative from nothing but the artist's imagination; that's where the magic happens. I've listened to many deeply beautiful and immersive pieces of music that didn't have any of those tiny imperfections, and they didn't feel boring or lifeless at all. A composer doesn't just crank out material for a musician to play, they are an artist in their own right. There's definitely artistry to playing an instrument, and sometimes you lose a lot by taking that away (violin soundfonts and sample libraries always sound terrible, for example, because it's an instrument that involves so much subtlety and variation), but I'd say there are also cases where that rigid timing is a better fit. It can help keep rhythms clearer, and in some styles too much expression can even be distracting. In a lot of really fast and rhythmically dense electronic genres, that clockwork precision is absolutely essential for making it work at all.
this whole never ending pursuit of "tightness" forces me to back up and ask "Why do you need this instrument to be so precise" as I would very much doubt any human being playing any instrument has a "tightness" of less than 2 milliseconds. Why not just keep the tightness to the "human scale"?
Imagine the disappointment when the Marble Machine finally goes on a world tour, and his fellow musicians have a 30 millisecond standard deviation.
@@markgriz Right?! If my piano tutor graded me on deviation I'd have been banned from playing piano by now haha
I guess that he is spending so much time on this because any other deviation from the true beat that other let’s of the machine will introduced will be multiplied by the gate.
2ms in the gates + 2ms in the programming wheel + vibrations + many more will equal to a lot more than 2ms
@@maticz3923 okay. Lets carry that onwards. 2 in the gates + the 2 in the programming wheel = 4ms. Lets be generous and say vibration is 3ms, so 7ms total. Marbel decay? sure, 1ms gives 8ms. Different air movement on stage? + 1ms = 9ms. Instrument rattling, 3ms = 12ms. Friction warming the machine and changing the paths, 1ms = 13ms.
I could keep doing this, make as many bogus claims as I want, and we are _still_ under half the delay of a professional drummer. There is no precision here, this is just pure madness. This is putting carpenter nails under an electron microscope to make sure they are all "perfectly" equal, just to hammer them into the shed roof.
After many years of watching your videos, I still can't understand how much happiness i feel watching your problems being solved
I'm torn - as much as I enjoy these videos and the data analysis and constant design improvement, I also just want to hear some music. That 5 second clip of marble machine two at the very end of the video is incredible. Its more so about the journey than it is the end goal, but we also want some pay off in terms of that sweet sweet marble machine music!!
Watching that line hit a different spot for each measurement at 13:30 while Martin says its "zero" made me realize I still have no idea what he's talking about.
Your marble drop design is a piece of art in its own right. Congratulations! 😀
You are a huge inspiration, Martin. The way, you look at the meta level every single video, amazing.
Going for the best results and celebrating perfection. I love it.
Dude at this point your going to have to put the marble machine in a vaccum chamber to eliminate air currents. Control for temperature to eliminate expansion of material. Put the machine on shock absorbers to remove vibrations, and have every marble rounded to a perfect sphere with 0 tolerance, to achieve the perfection you are striving for.
It doesn't have to be THAT perfect, and if you look at any instrument, there will always be little imperfections. It just needs to be good enough, and your MM2 was just that if not more so.
I would say it's impossible to make a perfect machine. Once you have all the moving parts running there will be vibrations that will cause varience and parts interacting in ways you may not expect.
I was with you in the beginning, but now I feel you are chasing an impossible dream.... Best of luck on your journey!
I think Martyn's point is that if a gate connected to a computer controlled solenoid is off by a few milliseconds, then it's going to add to all the other inconsistencies.
If you could design a gate that adds no inaccuracy, like the one in this video, then it's something that you don't have to worry about.
There are things that he can't do anything about, which will add inconsistencies.
Saying that, I still don't think removing a couple of milliseconds will have any impact on the way the machine sounds.
I've been watching this channel for a couple months shy of 10 years and as much as I love Martins work, this obsessing and scrapping perfectly workable components has become a looking glass into the world of a mad genius who is steadily losing his mind and he doesn't even know it. If you ever read this Martin, you need to look after your mind, this obsession is not healthy.
Thats it - gravity rules. Because with springs and droppers you always will have deviations in the applied force, but gravity is always the same.
Ok now martin pls dont make your 3rd marble machine bigger than your door i warned you!!!!
I am very impressed by your skill to build perfection. But remember, that perfect is the enemy of good.
Keep on with your amazing journey! It is one of the things, that fullfills all my weekly needs in one single video.
I feel like I must have missed or forgotten about the video where you explain what went wrong with the previous marble machine. It was over-engineered, yes, but it was awesome. Why did that project fail, why did it suck, what happened? I'm confused...
Martin was dead-set on being able to take the machine on tour and do live performances with it. This placed very high demands on reliability, to ensure it could be transported and perform accurately and without failing during a performance. Ultimately, he decided MMX had too many features that were too complicated to ever be as reliable as he wanted it to be.
Whether Martin is throwing away "good enough" in pursuit of unrealistic perfection is up to debate.
With regards to live tour performances i think the straw that broke the camels back was that after designing the same system 20 times he couldn't get the marble drop system to do perfect 1 million drop loops without losing marbled
Eventually parts broke down
Martin has a point. Yes, people wont notice if gate itself has small deviation.
However, gates are not the whole machine.
Mechanism that goes from the programming wheel to the gate has its own deviation, and marble dropping on an angled surface has its own. So deviation gets compounded over the mechanism.
He is going backwards, minimizing deviation from the end to the beggining of the mechanism, which is really smart.
I mean...if you want to build some high-precision thingamob (instead of "only" a very cool musical instrument), more power to you. But in that case, I don't think you should be using contact microphones and an audio program for measurements (software will introduces more variables - that's what osciloscopes are for!), and use precision build parts (instead of, it looks like, FDM 3D printed parts. Resin printed parts would be way more precise).
Also, you'd have to bear in mind that sound only travels aprox. 30cm in one millisecond. And that the solenoid may have different performance metrics after heating up a bit.
Also, ALL other parts of an eventual machine will have to work to the same spec. What if some other vibration of the machine affects the marble gates?
And so on, and so on.
Chasing precision is of course a great and sometimes fun rabbit hole to go down. But it's a very deep and frustrating one, and reeks of depth-first search.
I hope the overwhelmingly negative and concerned reaction to this video reaches Martin in some way. We really care about this machine and the art you set out to create and doing this to yourself isn't healthy.
I actually hope for the opposite. He wants the journey, not the destination. He likes to challenge himself so he can learn more and more. And the engineering part seem to be just as important as the music part for him, I imagine because both are pleasurable for him.
There not being a finished Marble Machine (EDIT: a Marble Machine that can achieve what he set as his goals for it) is not bad as long as he feels it's not possible. Right now he's trying to make sure of that, which is really the important part.
@@mateusbmedeiros Makes me wonder if he's just milking a paycheck at this point. He's redesigned everything like 20 times now and it seems like he plans on doing that forever. Discord really lives up to its name.
@@BaronCreel I think if he was he would've kept the youtube membership in which people gave him money directly and maybe some filler videos to net more ads, I'm guessing. But of course not being Martin himself I can't really say for sure.
@@mateusbmedeiros Yeah I can't say for sure yet either. I do know his discord is living up to its name and is making sure he never gets anywhere.
For added context it would be nice to see how the gates from the MMX do in this test, side by side.
I often end up struggling in spirals of destructive perfectionism, which is an obsessive drive to endlessly improve details that are irrelevant to the quality of the final product. I suffered a huge amount of frustration in my professional life because of that. I see some worrying patterns of that attitude in Martin's work as well, but maybe I'm just biased due to some personal failures I have suffered and attributed to that.
The first marble machine was tight enough to inspire millions of people. I don't think it needs to be any tighter, just modular for transport and durable for the world tour
Love the two minute papers reference. I'm so glad that you are back though. I believe in machine #3
I love this project and how youre doing your best
The speed of sound in air is 343 m/s. It takes 2.9ms for sound to travel one meter. You are optimizing for error which is smaller than the error introduced by the fact that the instruments on the machine are different distances away from the listener depending on where they're standing. With love, from an engineer who has heard you say you want to balance engineering and art: this is not engineering. Engineering determines what error is acceptable and then optimizes for that. If the ear can't hear the error, and the error doesn't accumulate into subsequent notes (which it doesn't due to the drum design), it's irrelevant.
I am glad Martin is back to making videos that make me sigh at how long this has-to-be-perfect-for-some-reason-project is going to take.
It's hilarious that Gate 8 resembles the very gate that was worked on years ago - but that the release finger is part of the very escapement arm and much much shorter instead of a long lever.
Ahhhhh, such a delight to see Martin work! And I must say, the infrequent/randomness has me so much more excited to tune in than the high-stress weekly vids! KEEP UP THE GREAT WORK IN WHATEVER PROJECTS YOU PURSUE MARTIN!!