Does Into The Breach Really Have Perfect Information? Why RNG Matters.
HTML-код
- Опубликовано: 20 июн 2024
- Follow me on Twitter- / thefearalcarrot
Support The Architect of Games on Patreon- / architectofgames
Into the breach is a great game, but something has been bothering The Architect, everyone's been praising the game for all the wrong reasons.
Into The Breach is neither a game with perfect information, nor a game with zero randomness. What it is, is a game that uses uncertainty to great effect. Find out how in the video.
You Saw:
Fire Emblem Path of Radiance: 2005
Heathstone: 2014
XCOM 2: 2016
Into The Breach: 2018
Opus Magnum: 2017
Starcraft 2: 2012
Dishonored 2- Death of the Outsider: 2017
A Hat in Time: 2017
Pokemon Emerald: 2004
Sid Meirs Civilisation 6: 2016
FTL: Faster than Light: 2012
Eternal card game: 2016
Interesting links:
That article I was talking about: www.rockpapershotgun.com/2018...
BUY BEN PRUNTY'S SOUNTRACKS: benprunty.bandcamp.com/
Cool polygon article on ITB's difficulty: www.polygon.com/2018/2/26/170... 
Игры
-Plays FTL
-Doesn't upgrade doors
-Complains about boarders
Those who control the doors, control space itself
shouda sent em into medbay and open the airlocks lol
best strat against boarding
You can easily control boarders without upgrading doors
Or you could man doors for the bonus, vent & retreat towards the medbay if doors are boarded or the boarders are too challenging, and spend the 35 scrap on actual game saving upgrades because that's what's important.
The big concern in that moment is the engines were upgraded before cloak. Getting that +5 seconds of cloaking is critical for Stealth
🅱️.
The RNG Stealth 🅱️ truly struggles with is drone spawns and getting hit in weapons, sometimes through cloak.
I had a building resist a hit on what was quite literally the last phase of the last turn of the entire campaign, with only one energy left. I was utterly amazing. I knew I was fucked. No matter how I tried, there was no way to prevent that last hit. And then bam, resisted.
I'm still not convinced the game didn't cheat to save my run, but it felt great all the same.
Shrike I had that happen as welk, but I didn't feel good that it did. Like, my brain said that I should've lost
It was the only resist I had the entire campaign, which put it squarely within the % chance for a resist. Honestly, if it had happened on a previous turn of the final mission, I still would have won, it just would have felt less epic. I get what you mean though. I would have preferred to have been able to prevent that last hit, but there just was no way to do so :(
Yep, same here
Happened to me as well! I'm 99% certain that it's a hidden mechanic to make battles feel closer and give players a final chance.
Deus Ex Machina.
Every video. "How can I talk about xcom?'
In ftl, if you are boarded, just open all the doors
And you do not run low health crew through airless systems...
Indeed, I blast them out of the airlock every time. I mean they don't float off into space. But just upgrade the doors.
I guess you could say that the commander is a Door Monster.
President of Whiteistan I get that reference.
"He who controls the doors controls space"
Not knowing what move the opponent is going to make does not make the game not a game of perfect information. Chess is a game of perfect information, but you don't know what your opponent will do.
I wouldnt call Chess a game of perfect information (when seen from one side), as you do not know which move your opponent will do (though you do know what he can do). It is however a deterministic game, as the game will always go the same way given the same input from the players.
By definition, chess is a game of perfect information. All it means is that both players are instantly aware of the other player's move.
I had forgotten the detail, that the above destinction I made is for games with concurrent turns (such as stone-paper-siccors) and does not apply to purely consecutive games, such as chess.
It is possible that I wrote it while being fairly tired.
sorcdk ‘stone’-paper-scissors...
the mistakes continue, maybe I even made a misyake with this one...
10d6 is actually easier to predict the results of than 1d6 because the distribution is a steep bell curve.
1d6 has equal chance of any outcome. Meaning you have no capacity to predict anything.
But 10d6 will strongly favor a result of 35, or very close to it. There are many combinations that result close to this center number. And only one result for each 10 and 60 by comparison.
This results in more 'entropy' being more ordered.
That's true, however entropy can also be affected by the total number of possible states for the system to be in. and 6^10 possibilities is much greater than 6^1 possibilities. This is why dissolving solutes in water increases entropy, technically it is more ordered but the increase in possible states out weighs that by far.
This depends on whether you mean "sum of 10d6" or "a set of 10 independent d6 rolls", the former of which is your example, and the latter of which is described in the video.
He clearly displays the result of 10d6 as a total of all die results and not a string of each dies independent result.
entropy is bigger (there are more possible outcomes), but variance is "smaller" (average of the sum being 35 with small standard deviation).
He clearly shows the results on all 10 rolls. And the description clearly references that and not the sum (which yes, is also shown).
The "perfect information" claim about Into The Breach was never about players having 100% perfect information all game - only about you having perfect information regarding the current turn. With your actions, the game makes it clear what each move you make will result in, and what every enemy will do if you don't interact with them, and how your actions will change their behaviour. Also, this video makes me very curious about your perspective on Frozen Synapse and Frozen Cortex, which are equally deterministic, but give a player much more potential to be the RNG for their opponent.
This is where we need to make use of the difference between known and unknown information, and how unknown information results in psudo-randomness. In that case, Frozen Synapse and such are not perfect information games, but they are deterministic games, when considering the full amount of input. Chess and rock, paper, scissors also falls into this catagory. Note that they have their own special brand of probability theory, and there are several different levels of information you can have for such games (such as know your opponents move or know strategy).
Technically, there is still some randomness on your turn because buildings have a chance to resist damage, including on your turn.
@@caltheuntitled8021 Yeah, that's a good point.
I don't think you really understand what a "perfect information game" is. It doesn't mean that you know what your opponent will do. In chess you also don't know how your opponent will play, yet it is a perfect information game, because you know everything on the board and there are no random actions possible. But the point of the video is still legit.
He also doesn't understand what information is in the context of information theory. He basically got every single thing wrong that you can get wrong. Maybe he should have sought some better information.
@@2Cerealbox what is positive and negative variance?
Into The Breach has more layers besides the chessboard combat mechanics. The combat missions are a tactical; but the strategy game between the missions is more poker than chess; meaning that you have to take an educated guess at which mission is going to be the best (or rather, least worst) choice at the given situation. Not all missions are winnable; it's possible that the mission generator gives you something that you barely survive no matter how good you are. If you have several FUBAR missions in a row; you need to weight to take a big risk to turn the tables against the certainty of losing. Other times you have a lucky start and you're better off playing it safe.
@@2Cerealbox the last sentence made me giggle
Well said. I don't know much about information theory, but I know that complexity is fun in games like Chess for the challenge of trying to navigate it to an optimal endgame.
Thing with FTL is that it is less about influencing the variance and more about being able to react to to those random events. If you could influence the results then it would be less about reacting and more about planning.
That and having memorized what the outcomes of each event are. For example, NEVER help colonists with space spiders. EVER!
Tom Sko Unless you have a boarding drone, then light that place up!
Correct, but you learn very early on that all blue/special options ALWAYS yield good/great results. It was talked about in Joseph Anderson's review of Darkest Dungeon: to win at certain games, "skill" is based on how much knowledge you have about the games mechanics and how well you can recall and apply it.
I don't think you have to learn that, the game actually tells you outright in the tutorial.
Bio beam or clone bay make that event worth taking. This is in contrast to the Zoltan nebula event which can just erase someone, clone bay or not.
Re: FTL and boarding crews
From the footage you showed, the bit you said about losing 50% of your ships to boarding parties, and the fact that it's a "pause-anytime-and-issue-commands" game, I can see that you never learned how to handle boarding parties in FTL.
1. Send a crew member to the door room. The free upgrade on doors will slow the boarding party down immensely, this helps a ton when combo'ed with step 2.
2. Vent the enemy crew to space. Yes, I see that your O2 is down, but they don't have a way to heal and you do.
3. This is the biggest one. Abused the hell out of your medbay! You can have your guys sit in/swap in and out of the medbay to constantly get healed over time, which makes the enemy boarding party stand no chance.
Works every time.
don't forget to but at least the first door upgrade when you can, it nets you enogh time to get a crew member to the door module to further reinforce.
wait does this disprove his points?
@@weirdo3116 Not really, FTL still has a bunch of bullshit moments in it, but having good knowledge of game mechanics allows you to get fucked a bit less.
Basically, git gud
@@weirdo3116 it's actually really hard to lose the game by getting your crew killed off by boarding parties
In that ending scenario, limiting yourself to relevant information is a huge help. You've got 3 building threats, and 3 mechs to maneuver. Negating 2 threats with one mech is ideal.
There are multiple ways you can go about bodyblocking to attempt to achieve this. Only some of which are favorable. The best way I see is to have your Lightning Mech step in front of the Firefly Leader and attack it, dealing 2 damage to the adjacent Fireflies and negating one building threat.
From there, the Grapple Mech can tank the remaining Firefly's attack, though it can't use its grapple that way. Moving the Boulder Mech into the position you highlighted AFTER moving your Grapple Mech means you can hit the Goo in charge of the last building threat.
And that's just one way you can resolve that turn in an advantageous way.
Just saw this video, and just posted this, I'll repost here:
11:51
1) Use Hook Mech to pull the blob into the path of the firefly, both destroying the blob before it can damage anything and blocking the firefly's attack.
2) Bring Lightning Mech forward one square and zap the firefly, chaining the zap to the firefly boss as well.
3) Use Boulder Mech to launch a boulder at the same firefly, simultaneously killing it so it won't damage your Lightning Mech and pushing the boss firefly to the left so its attack will miss the buildings.
4) End turn with zero damage to mechs and zero damage to buildings.
@@raphaelellis5969 Actually better to use Boulder Mech to throw the firefly leader into the water.
@@Jiorgon It’s a boss, it isn’t killed by falling in water, it just cancels their attack for that turn. The strategy they explained would do more damage.
Thanks for having the names of the games in the discription, going to look up Opus Magnum: 2017 right now. Looks interesting!
With FTL, you can win every single boarding encounter by leveling up your medbay, and then just venting every single room except the medbay. If your medbay is too small to support all your crew, you can vent the medbay as well to prevent the enemies running to it.
Another variation on this theme is to upgrade the doors to max level. Then just vent whatever room the enemies are in. Every time they break through a door to the next room, you vent that room too. They will die before they get to your guys if you do it right.
I know, what you were trying to say about the removal of the mana system in Hearthstone(and i agree), but it's still ironic to hear about Blizzard removing some variance from their card operated slot machine.
Remember that it isnt the part that has to do with you putting money into it, and you can see why it wouldnt help them to have such variance. Hey, if their game is perceived ass less random, that means people would be less likely to consider that part.
His point on the nuance of color commitment didn't really have to come in one package with the land system's variance though. There's a good reason many Magic-style modern card games(Duel Masters, Wixoss etc) replace it with "sacrificing" hand cards into mana to fuel other cards, or something similar
I just discovered your channel, and your vids are high quality dude, keep it up and I have no doubt that you'll be big one day
Thanks for the support!
Oh, he'll be big alright
I think this counts as one day!
10:15 why dind't you vent the ship and hid in the medbay? You hadn't picked up a door system, so an upgraded medbay might be a requirement if you don't have any good crew. More than arbitrary, it has a long feeback loop, mening you don't see the direct consequences of your actions, to go deeper into that, check out Slay the spire, random enemies, random cards, but everyhting goes back to how you deal with it.
Yup, FTL is more about prediction, sound logic and game knowledge, than dealing with the immediate problem at hand.
yeah, but in slay the spire i got fucked by that crow bitch because i was using a power intensive deck (power per turn, metalicize, necronomicon+that one card that does like 30+ damage in a single hit all paired with that card that doubles/triples power cards, and Reaper for sweet sweet heals) all my effort was wasted cause i didn't know the fucker had a second stage that basically gives him a 1up... FUCK THAT CROW!
Thank you. This was exactly the comment I was going to make. I disagree very strongly with the video maker's comment of FTL being a game with 'too much variance'. With bad ships like the stealth cruiser, you certainly are going to have a lot of failed runs, but his example of a failure state was completely avoidable with good planning. Part of 'getting good' at FTL is experiencing the random, negative outcomes that can happen during random events, and planning ahead so that those events don't end your run. A 35 scrap door upgrade probably would've saved that run, which is worth getting if you have a squishy crew.
I'm still sad that you don't get nearly enough views on these, I've been sharing them where I feel it would be accepted when I see a new one. :)
Thanks very much, I'm sure it's probably brought a few new people to the channel!
Hopefuly, I heard some good compliments about this video specifically when i shared it today!
I've certainly followed the channel because of this video. I enjoyed it!
subprogram32
You are a true fan! 👍
I on the other hand, this is the first video I've watched. Just dropping by
I'm sorry but that's not a correct reading of how entropy works. Entropy is defined as the sum of p_x log p_x over all x, where each x is a possible configuration of the system, and p_x is the probability of x occurring. For a deterministic system, either p_x = 0 or p_x =1. In either case p_x log p_x = 0, so the entropy is zero. Thus, games like Opus Magnum, Chess, and Go all have an informational entropy of 0.
Also, you don't have to know what your opponent is going to do to have perfect information. You only have to know what state the game is in, and have no randomness. Go and Chess and both perfect information games because you can see the whole board. Poker has imperfect information because you can't see the opponent's hand. The distinction is important because if a game has perfect information, there exists a single optimal strategy that you can derive using the MiniMax Algorithm.
These are all questions about the class of game these systems fall into from a game theory perspective and what algorithms can be used to find optimal strategies, not how complicated they are
What is the intuitive meaning behind "p_x log p_x over all x" though?
I mean, how could it be described in "layman's terms" similar to how f'(x) = df(x)/dx can be described as "the instantenous rate of change of f(x) equals to its gradient"?
@@revimfadli4666 you can also write it as "Expected log probability". So if you take a random sample, how likely were you to get a similar sample on average? With a really tight distribution, any sample has high likelihood. If a distribution is very spread out, than all your samples will have really low likelihood. Basically it's a way of measuring how a random a distribution is
In your FTL boarding situation, you should have stuck two guys in the medbay and vented the rest of your ship. At that point, the challenge would have been to not lose life control and doors to weapons fire and to try and keep the rest of the crew alive (by not venting the entire ship and using tricks like that.
FTL is a lot more skill-based than people realize. I've put quite a lot of time into the game and I'm now winning more often on hard than I used to on normal early on.
Perhaps that's FTL's problem too. Sometimes, the random number generator rolls an enemy that's just the perfect counter to your ship out something that's just really strong (eg double bust 2 midway through the game), which you may not like, but it's the situations you feel you had no control over that come on top of these that make the game feel like it's out of your hands. Its steep learning curve basically compounds the RNG issue.
11:48 Use the launcher mech to push the boss to the left, use the grapple bot to pull the exploder into the line of fire of the lightning bug so that it dies before going off and blocks the lightning bug’s shot (usually the exploders go off at the end of the vek turn), and use the lightning mech to do some chain damage to the boss or the other two vek.
Alternatively, if you have building chain for the lightning mech and your grapple mech has enough hp to survive it, move it out of the way of the lightning bug’s shot, then shock the central buildings to kill the exploder and damage the two strong enemies. Then use your grapple mech to pull the exploder producer into the line of fire of the other lightning bug. Finally use the launcher mech to move the boss to the left (it has 2 hp left, so it can definitely tank the lightning bug’s shot if you don’t kill it with the rock.
I know I’m a few years late, but that is far from the most impossible situation I’ve seen.
FTL does have a lot of RNG, but the best of the best have a ~90% win rate. On hard. Even I'm >50% on hard if I were to start over today. To illustrate, all advanced players could handle that stealth B (which is the worst ship in the game on hard) boarding event you showed without losing crew or taking ship damage regardless of where the enemies spawn, via venting oxygen to manipulate AI boarders. The FTL devs intended a low winrate on normal, what they got was players screwing around with triple ion builds on hard for easy wins and a lot less variance than most people think :).
i've won about twice, downloaded the infinite space mod (or whatever it's called, it allows you to just fuck around for as long as you want) cause that bossfight can go fuck itself (it isn't necessarily hard, but it's annoying for the game to end when the random pleb battles are what i enjoy the most) and the captains edition mod, and never looked back.
Well, abusing exploits is essentially like using a god mode cheat to win. Wonder how they still manage to lose 10% of the time...
Venting your ship was put in the game for a reason, and that reason wasn't to pretend it was an exploit in obvious use scenarios. Not everyone who plays the game better is using "exploits" :D.
@Cudgeon
"Abusing exploits"? Are you serious? Strategically venting your ship's oxygen is how you're SUPPOSED to fight boarders!
Cudgeon Have you ever actually played/seen FTL before? There aren't a ton of 'exploits' that exist, and most are actually just intended but slightly unintuitive mechanics. It's also not an easy game.
Just found this channel and great vid. One trivial gripe I have though is that you keep your text on the screen just a little shorter than necessary. I found that if I didn't rush through reading what you said on screen then I wouldn't have time to completely read what was put on.
Again great video, very informative thank you.
Great video, well explained, lots of examples. Awesome.
Also, tip for Into the Breach - 3 Boulder Mechs yo! Just rock it up!
I just watched all your videos in two evenings. They are very interesting and insightful, thank you for taking time making these for other people to enjoy!
10:23
Giant alien spiders are infact no joke.
It's about having perfect information at the current turn. However, yes, building can resist an attack on chance, that is the only element that breaks the perfect information label, since it's a dice roll. Not knowing what your enemy will do doesn't mean you don't have perfect info, otherwise, only puzzle games would have them.
I think the fun part of strategizing is not only making plans but also making back up plans to that plan. It makes you feel smart and it’s fun!
When you introduce RNG as a mechanic, there is still mathematically a "best" strategy to beat it. It just changes from building a strategy which always wins, to building a strategy whose outcome has the highest probability of giving you a win.
Many of the questions in this video are purely semantic (e.g., RNG versus what counts as perfect information). I would say that a (let's say one player) game with RNG can be counted as perfect information so long as it actually tells you the various RNG probabilities. That is: it should be the case that given any strategy, you can theoretically calculate the probability that your strategy will work or not. Then things are not all that different to a classical perfect information game like chess: it's still the case that there's a mathematically "best" strategy.
That last part is the most accurate representation of how the ItB forums were like back in the day.
The building defense chance is such a great gameplay mechanic. You never want to take building damage so it would never come into play for professionals, but on the off chance you made a mistake you can pray for this chance to give you one more chance. It softens the punihment for playing poorly but doesn’t incentivize playing poorly because you’re only allowed one repeat.
Great video, and finisher. Thanks Adam. I'm glad your channel is getting the attention it deserves. Your content has great form. Your videos support your in depth analytical commentary with engaging gameplay footage, side notes, and humour. I've only recently discovered your channel, and subscribed. Your content reminds me of Mark Brown's Game Maker's Toolkit, who I'm also a fan of. It's inspiring to see more mature and intellectual content like this. We need more of it. Games are pioneering works of art, so it's great to see them acknowledged with appreciation and analysis as they grow into more immersive experiences.
You mention World of Warcraft in your videos. Do you play MMOs? It would be great to see your take on the MMO genre: Why it has quietened recently, and what developers could focus on to help their new MMOs and communities succeed. Bless Online was the most recent hope we had for the genre, but unfortunately that appears to have fallen before it can walk. It's a genre that I really want to become great again. I believe it is the future for games, as I think most genres will eventually have a mass interactive community from within the game world. I understand they are the most expensive and complex genres to get right, and therefore risky from a business perspective, but there must be some lessons we can learn on how to build and deliver them. It feels like the core experiences need to be defined and refined so they can communicate these unique qualities to the community, so the game can sell itself as something different, as well as helping players justify why it is worthy of their time investment, and build the hype on solid honest ground. It's not an easy fix, but I think Bless Online, if anything, showed that it is a fix (pun intended) that a lot of gamers yearn for.
Thanks again for the videos. Looking forward to your future content.
Peace
Playing XCOM2 and Into the Breach are the two times where I was playing legit 4D chess, with all the info in front of you, plenty of time to think it through, and a feeling of 'Aha!' When i successfully pulled off a 'thinking with portals' grade mind game and crushed em
For me thoes "portals" tend to be perfectly lobbed hand grenades, it's odd to me how out of everything in the game where exactly the grenade lands isn't randomized but the damage deals is
Wow, this is almost no views for a video that got a shout out from Mark Brown.
Edit: Your channel seems to have blown up massively since I left that comment, and rightfully so!
I mean, I would contest your point on informational entropy, because there IS a genre that has small to none - puzzle games. Puzzle games are based entirely around having small or nonexistent informational entropy, expecting the player to be able to juggle *ALL* of the rules of the game in their head at once, and think several - entirely predictable - moves ahead. Many puzzles games, in fact, have no randomness at all.
I think my eyes bugged out when I saw you had less than 1k subs, well here's one more to that goal.
Hacknslaashing 60k now
Another good video. Great job verbalizing what people subconsciously feel.
What was the FPS you placed at 5:32? I know it's another checkbox to tick before rendering, but please add game names in the corner. I see you do it sometimes.
Tom Sko It's the new Dishonored game. Death of the Outsider, IIRC.
When you get boarded in FTL, you man someone in security (if there is one), then open all the hatches to the boarded room to deplete oxygen giving you an hp bonus when they battle the boarding party. meanwhile target their teleporter, healing bay, and oxygen tank to suffocate the opponent and deplete your opponent's personelle health. you get bonus scrap for not damaging the ship as much. FTL is as random as xcom, but there's always mitigation strategies
The oxygen strat works so well you can pretty much have non-essential rooms perpetually oxygen depleted to ward off boarding parties... unless they target your security room or they have those aliens that can survive without oxygen
Great games that are actually 100% deterministic get away with no randomness whatsoever:Bullet Hell shooters like Ikaruga and Music games like Guitar Hero.
They raise the skill level needed to beat them (on the highest difficulty) to such ridiculous levels and add a high score on top, that perfect information doesn't even matter, because you won't be able to play remotely perfect anyway unless you're a god with hundreds hours of practise.
Also old school adventures but they're different because they may very well "feel" random because of weird puzzles and aren't even games anymore on their second playthrough and more like clickable movies.
also boardgames like chess, checkers, 4 in a row, go
ikaruga has quite a bit of randomness in the suicide patterns and the game actually spawns new enemy formations if you speedkill stuff. also lot of bullet hells such as psyvariar have loads of rng in the bullet patterns and player must rely on twitch reflexes instead of pure memorization
Entropy in information theory and physics are quite different when you start to really look at them. In practice they are inverted compared to each other in meaning.
In physics entropy is the number of different states that looks the same to the observer, and it therefor represents hidden information. In information theory, entropy represents how much information is in a state that is fully observable.
Math wise these things are close together, as a fully random system would have maximal entropy in both and a homogene system would have minimal. The problem is in the interpretation of information, where entropy in physics describes loss of information, it describes gain in information theory (or technically capacity or value).
I find the explainations of informational entropy and variance given here to be quite confusing and lacking, if not downright missleading.
There are special problems with the use of destintion between negative and positive variance, because variance as mathematically defined cannot be negative (unless we are dealing with some esoteric probability theory, with say events with negative probabilty or imaginary outcomes). Variance is a measure of how different the possible outcomes are expected to be. This can come from random positive events, negative events, or mixed events.
There are several problems with the informational entropy, but a lot of it might be related to the discrepency between using information as in "full information", and the information lost to entropy, which is the part of how a system can be practically random (psudo-random).
Full information is related to physics, where full information basically means that you have all the information necessary for the future development of the system to be deterministic and unique for the given initial state (which is what you have full information on). This doesnt mean you necessarily know how to calculate it, or that it is even possible to calculate such a future state (there are a host of physical problems, such as the 3-body problem, that even given full information is not precisely calculatable - though we can approximate them). In game terms this means that while you that in a full information game you are presented with all the information necessary to determine what will happen, but the problem might be so complex that it will be very hard to determine what actually will happen. This is how we build puzles and mazes, you have all the information of how to solve them, but you still have to figure out how to actually do it in practice.
Did you really have to include the FTL theme music? :( now I have to go and play it again...
i kinda like how, in both FTL and Into The Breach, your whole run can be completely ruined by one, tiny, screw up. Even if, before then, you were doing perfectly, it keeps you on edge and teaches new strategies to try and prevent the same outcome.
I'm kind of in the opposite realm. That's what I dislike most about Into the Breach (not as much FTL because I still feel like there are more viable ways to play and beat the game and more ways to take calculated risks -- there's a bit more "freedom of player choice" in FTL even if it's still rather restricted).
My favorite games are like Fallout 1&2 and X-Com where you have tons of freedom in how you want to play the game (thanks in part to the RNG), and it's not so easy to say what is a "right" move or a "mistake"... there might be more risky moves like firing a shot that aims for the enemy's eyes that has a high chance to miss but could change the tide of battle if it succeeds, or safe moves like firing a low-damage shot that is almost guaranteed to hit and keeping your squad in defensive formation which might only help a little bit to sway the battle in your favor but has a high chance to hit and a low chance of getting your squad in a vulnerable position.
I love when games make it difficult to say what is a right and wrong move and instead makes it about calculating risks and weighing pros and cons. Into the Breach offers many cases where there is only one or two right moves and many wrong ones, as in making a blunder in Chess which costs you your queen -- it's kind of a black and white game that way that doesn't give many viable choices to win. I never really enjoyed that type of Chess-style puzzle gameplay -- I'm much more into games that give me lots of viable options and I weigh the pros and cons of each one.
Great video! Love your cadence and playful approach to providing some awesome insights into game theory. I don't play games often these days, but Into The Breach has got me hooked! It's really polished and loads of fun.
Have a great day!
-J
Your content is really wonderful, you've easily earned yourself a new subscriber! Very interesting analysis of things I take for granted a lot.
I love that I can totally screw up a move in ITB and immediately think, "yeah, that was my fault".
In FTL, its all about how well you could react to certain events, and how good certain ships are at reacting to certain events as well (which is why i make the assertion that stealth type b is the worst ship in the game). Into the breach is about how proactive you can be given every position the vek spawn in and where to place your mechs etc. Naturally in reactive games the bullshit-o-meter is very high where in proactive games/situations it feels like it's YOU'RE fault for doing something wrong.
what is the name of that 'molecular assembly' game at the beginning?
I have no idea why your channel isn't bigger, keep up the good work!
I've actually seen a sort of middle ground between the MTG mana system with colors and the Hearthstone one without colors, in a flash game called "Mytheria", which is fairly old at this point. In it, each turn you generate one "Power" (essentially mana) of one color of your choice OR draw a card. Generating power doesn't just add 1 to your current power, it adds 1 to the amount of power of that color you have each turn.
So, if you're running a blue deck and you generate 1 blue power each turn (alongside the 1 it gives you for free, since your starting power matches the color your deck has the most of) you'll have 1 power on turn 1, 2 on turn 2, 3 on turn 3, so on and so forth, no matter how many you spend.
Headstone has color in how cards are devided into classes and you can't middle-class
Forgive my ignorance but what was that dice rolling program?
Dude that level you showed at the end is actually quite ez to solve. I know it’s a joke but still. Use the rock thrower to block the boss’s shot. Grapple the other beetle that attacks the building onto the new enemy spawn spot. Ant electrify the explosive goo with your mech
Nice video, thanks! Subscribed and going to watch more of your stuff. :) I agree with some of the random features of Into The Breach, they really add to the game. The Grid Defense mechanic however.. I don't know. I wish they would've implemented something more meaningful for going beyond powergrid level 7. But I'm having a hard time myself at coming up with a better idea.
You're right about one thing here: the presence of what are essentially die rolls for grid defence And spawns make the combat imperfect information even if the rng in that regard is kept to a minimum.
However, when it comes to games of perfect information, knowing what opponent's will do on future turns (as opposed to simultaneous or prior turns) is not required. You just need to be able to work out their set of options. Otherwise games like chess, checkers, and go could not fall into the category of "perfect information" despite the fact that they are the bastions of this category.
Here off a rec from Game Maker's Toolkit. I'm sure you know this already, but... You deserve more views. Subscribing now! Great video!
I know not many players would agree, but I think that Europa Universalis has too much RNG, as much as I love the hell out of the game, the weighted random starts for each nation, while great for replayability, made learning the game an absolute hell for me, as even the easiest nations are prone to random game ending scenarios (e.g. As a new player playing Castille, France rivaling you or allying Aragon is essentially a game over, and you can do very little to affect this, other than restarting the game until you get a favourable start)
Even as a relatively experienced player, sometimes the only option is to restart your game until the RNG rolls in your favour for a decent start (e.g. Playing Byzantium and hoping that the Ottomans don't immediately invade you before Serbia or the Anatolian minors)
The game does, in that regard, have endless replayability, but I personally hate that the game essentially forces me to restart the game just to hope for a beginning situation that I can deal with at my skill level.
Does anybody have an opinion on this? Am I just whining about the game and my lack of skill, or is it a genuinely frustrating mechanic in your opinion?
Do you seriously think that france rivaling you and making an alliance with aragon is a game over, god you are bad at the game
Huge youtube as background noise type. Victim of auto play & recommended videos, and, having enjoyed mark brown, discovered you're channel, and subsequently, through the algorithm, this video. I loved ftl, I mean loved ftl, having gotten deep, and I mean, deep, into star trek at the time of my discovery of the game, I was, the, perfect, victim, for that game's design. So, upon, the first, abstract whispers of this games existence, I was obviously hype for this game, got recommended this video, and, subsequently, found out that this game, had come out, went to buy on steam, found out mac wasn't supported, proceeded to go out and buy the cheapest, all in one, windows desktop, I could find, and am currently out 325$ to play a $15 steam game, all because of your video, truly I am a whale for this industry, and to this mk ultra'd, Manchurian candidatesque type of advertising.
11:51
1) Use Hook Mech to pull the blob into the path of the firefly, both destroying the blob before it can damage anything and blocking the firefly's attack.
2) Bring Lightning Mech forward one square and zap the firefly, chaining the zap to the firefly boss as well.
3) Use Boulder Mech to launch a boulder at the same firefly, simultaneously killing it so it won't damage your Lightning Mech and pushing the boss firefly to the left so its attack will miss the buildings.
4) End turn with zero damage to mechs and zero damage to buildings.
3:52 what game is that? it's not in the description
nvm i found out; It's Opus Magnum
What I like in "Into the Breach" is that the combat rules do not depend on RNG. You always know what your mechs can or cannot hit and how much damage will be dealt and what other effects will the hit apply(pushing enemy, applaying acid etc.) I don't mind the rest of the RNG since it creates variety and encourages replaying the otherwise short campaign. For comparison in other games (as Darkest Dunegon or X COM) you have a chance to hit a damage range and chance for effects to apply like critical hits. It makes for a very intense experience but sometimes you feel cheated. At some point the randomness probably evens out(if it is a "fair"RNG) but this doesn't change the fact that sometimes you just want to scream at the monitor. There is a tactical game about football "Football: Tactis and Glory" where quite a lot of people felt cheated by the RNG and despite the devs explaining that it is a totaly fair RNG they decided to implement an optional "less random" system so they can keep people happy. For instance a player wants to pass the ball and has a "Pass" ability of 30 and needs to roll at least 20 to make a successfull pass. In the original system a number between 1 and his skill (30) will be generated. In the "less random" way a number between 15 and 30 will be generated thus decreasing the "range of possibilities" in favor of the player. I chose some arbitrary numbers but I think that the idea should be clear.
I'm studying electronics and communication engineering Shannon is my hero and I love games at the same time, you title intrigued me. I was always interested in the technical mechanics that game designers used. I love your video please make more on the same format.
Lots of love
I agree. There is still random generation in "Into the Breach". But the results of what the player does are not random, which is what I love about it.
I'm about 59 hours into it so far and I've only come up against two 'game over' moments where the game screwed me. And even then there was probably a couple of things I could have done earlier in the campaign to keep my power high enough to not be so close to defeat in that moment.
Great video :)
What's the game at 5:32?
Also, you misspelled 'Hearthstone' as 'Heathstone' in the description.
Alex Baldwin Looks like the new Dishonored.
In the last into the breach level you can block the alpha vek with your lightning mech and attack with it. You move your boulder mech and shoot the explosive goo and lastly you can either block the firefly vek with your grapple mech or you can move in the lane above it and grapple the goo launcher vek so it takes the hit from the firefly.
Another great video Adam. As you seemed to out of no where start this channel with some pretty impressive content do you work in the industry at all by any chance?
Nope, I'm just currently between fulltime jobs so I fill the time by making these!
1:15 You say that Stompy mech is a bit rubbish? Combined with the nanite bonus, it's incredibly powerful. Now, Zenith guard's laser mech... there's a weapon that's difficult to control. On your assertion that there is no focus on strategy, there are plenty of strategies to utilise, e.g. using Abe Isamu to pilot mechs that damage themselves, or Prospero on the grapple tank, or the Flame squad laying down fire, and even the placement of mechs to corral Vek.
Love the ending ^_^ I think FTL has the best music of any video game ever. I had it as my ringtone for a while. But I totally agree to what you said about it. It can be so frustrating at times
Thanks that you address this topic! It was an interesting for me to hear of a negative variance, which I was taught is always strictly positive in mathematical terms. Thanks to you and the so called Internet I now know that variance in accounting is used to describe the deviation from your expectations. Before I internally translated what you referred to as 'positive' and 'negative variance' to positive or negative reinforcement/reward.
High entropy games/systems aren't completely free from optimisation and strategies to maximise outcomes. The optimisation methods are just not as simple as perfect systems and due to the highly probabilistic and non-linear mechanics, require machine learning for optimisation.
Great example, AlphaGo. TD-reinforcement seems the way to go.
It's balancing player information with strategic uncertainty in a specific way. It offers positive variance and places the rewards on the player's skill over RNG by doing so.
It's an interesting take on the genre that makes strides by way of the ingenuity required to circumvent the restrictions placed on the developers. I think I may have to check the game out now.
Check out Prismata! It has an amazing take on perfect info: There's only rng BEFORE a match, which results in random units each match (identical for both players), TONS of branching complexity and limited time. You can check out Lunarch Studio's videos
on the spirit of FTL, never EVER send crew memebers against giant spiders, only take the blue options in that event.
11:48 hook mech to H4, grapple the blob over to G4, (the firefly will destroy it on its turn), electric mech to B5, fry the firefly on B6, if the leader firefly is low-health enough, kill it, otherwise push it. You don't get many kills but you also don't take any grid damage.
I'm amused you mentioned Civilisation's goody huts as never being negative, because that wasn't the case in the early games, and indeed, if you look into the files of Civ 5 some of the negative possibilities(Like barbarians spawning), still exist, but commented out. Whether that's still true with Civ 6 I don't know.
Hey man your channel is really good
Great talk about a game I just started playing, thank you!
lol, that outro.
The answer to solving that is not allowing more then four bugs to exist onscreen simultaneously. Less than three is ideal, even with four you run a gambit.
Physically speaking entropy has nothing to do with caos or disorganization. It is simply the amount of possible states a system has.
Which system has bigger entropy: (1) a bucket with still water, or (2) a bucket with water that you swirled with a piece of wood and now has a vortex?
The answer is (1), because (2) has less available states for every particle (they are moving less randomly hence are more organized), even though (2) is apparently more chaotic. Also, (2) will eventually become like (1) and as is known, the entropy of a closed system will always increase. Therefore (1) must have bigger entropy than (2).
0:17 how is that game called? Edit: 3:39 Opus magnum? Ok
While you ranted at the end, I was solving the map. Played too many times, I have...yet regret nothing, do I.
you can't just say that and then not explain...
11:49
Main plan with divergence of how to uss the grapple tank:
1: Use the launcher mech to move behind the lightning mech and hit the firefly attacking the lightning mech, pushing the queen firefly off it's attack. If you have an upgraded launcher, that's a vek down. if not, you still protect your buildings for another turn.
2: lightning mech moves to the space adjacent to the firefly and the scorpion. Prepare to shock, depending on what the grapple does.
3: as said before, divergence occurs here.
A) Tank moves to edge of map opposite of the starting positions of the launcher mech and grapple the blob into the path of the firefly attacking the upermost town. Lightning the bugs for damage.
B) Lightning for damage, then move grapple tank into the edge of the map opposite the queen firefly and most of the vek and grapple the blobber vek into the path of thr firefly.
City damage: 0.
Lvl of aggression: less than favorable, but considering power grid, acceptable.
I do have alternate plans, but they take power grid dmg, so I don't like them as much.
That's the other great thing about this game, is the "chess puzzle" aspect. Screenshot and share! It's fucking great.
Actually, I find every Into the Breach level has a scale of solutions. If I spend 10 minutes finding the best one for round 1, then on round 2 I can get the best solution. As such, i can easily knock up 4 perfect islands. (Well, Ol' Horace made me lose a star on island 4, so not quite)
Also, that round at 11:50, you want to make sure when you use Blitzkreig to get an extra power off early, (grab a time traveler who has one from previous timelines is the best way) and power the Building Chain on the Lightning Mech. The armor of the Hook Mech then allows you to chain lightning through it. I find Silica is stupid OP in the Lightning Mech, as you can just sit it next to a cluster of buildings and never move.
In this situation, should you have Building Chain powered, you could:
Move the Boulder Mech right behind the Lightning Mech, fire a boulder on the Vek attacking the LM to push the vek monstrosity in the bottom row. Drive the hook mech into that spot next to the goop bomb, (the tile where it would get hit by both the bomb and the firefly) and pull the firefly up close.
Send a lightning chain through the cluster of buildings so it hits the goop, hook mech, and fire fly. The goop is dead, the Vek Monstrosity stalled, and no buildings were hurt.
Amazingly well explained!
Negative variance is mathematically impossible, as variance is defined as an expected value of a squared deviation from the mean. Negative mean with some variance would be an appropriate term.
(Fades away into 1s and 0s)
"Rocket Destroyed"
Moves to abandon time line XD.
Entropy in physics is not about chaos, it's about how many microstates correspond to a given macrostate. The higher the entropy of a given macrostate relative to other macrostates of the same system, the more likely that macrostate is to occur. It's also analogous and closely related, but not precisely the same as informational entropy.
Even in the most dificult of cases, there is always a way to beat the level perfectly. There is a chain reaction caused by every move you make in you turn. If you take the right move on you first turn, you are guarantee to win the battle perfectly as long as you keep making the right move on every turn.
If yall liked the ending song and don't know what it is, its the main theme from FTL. You should play FTL
"the game is 100% deterministic"
[4 minutes 12 seconds later]
"lots in the game is pretty random"
underated channel!
reminds me a bit of Mark Brown
sadly I dont know someone i could share this wonderfull video with :(
Cultist Simulator plays in a very similar way to Into The Breach in terms of "perfect information". If you know what you're doing there are very few things that can surprise you as you always know what's coming in the future and how to counteract that even when rng plays a role. The biggest challenge is actually paying attention to the game and not making mistakes. At higher levels of play, most of the failed runs are because the player did an obvious mistake or decided to take a shortcut.
That ending could have not been more perfect for the average Into The Breach player
11:50 grapple mech moves right to block projectile and pulls bomb into self killing it then either of the other 2 mechs block the other projectile
giant alien spiders are no joke ;D
I really enjoyed your video on Into the Breach, even though RUclips seems to think you're mostly talking about FTL.
While I agree with what you said about FTL, I also think the game requires a certain degree of skill to make it feel less punishing.
The Stompy Mech is awesome, okay? Once you've upgraded your healing, that whole squad is an absolute menace. (And you can always bring Abe to tip the scale.)
Also nice video thanks
I think i discovered a great channel today.
One important thing which I'm surprised this video doesn't really make mention of is the difference between random events that you can react to in a reliable and predictable manner, and actions you take which have a random outcome.
Into the Breach deals exclusively with the former. You may be presented with a random event that has no ideal solution, but at least you know what the immediate outcome of your solution will be. Having a difficult situation where you will incur some losses is frustrating, but it's not nearly as infuriating as a game like XCom, where you might have a clear course of action which you can confidently say will give you the best odds of a positive outcome, and then the game rolls a 1 and you lose anyway.
only 4:40 in, but want to point out that although bug spawns and what they'll do is random, you still have perfect information from turn to turn, and THAT was the goal, not perfect information at all times. if a game were to have only perfect information, nothing hidden at any time, it would have to have a high skill ceiling for executing a strategy. a good example is assault android cactus, or to use a more well known example, doom eternal. every encounter is the same every time and demons are quite predictable, so you can approach any given encounter the exact same way every time if you really wanted to, and there is a most optimal appraoch to each, but executing these strategies perfectly is nearly impossible
11:00
I hate that misconception, that colors are tied to drawing cards. You know what would have no RNG but still include Colors? A Mana Generation phase where you pick one of the 5 available options (you could even add a Grey color worth 2 colorless if you want to be really crazy).
Boom, you get Color Mana but also flatten the mana curve.
Have you checked out Mabinogi Duel? While the game only lets you take cards from 3 of the 5 colors/factions, you have to generate colored mana. The mana created is random, but the rest of the game's mechanics are heavily diminished. There is no drawing cards from a deck, creatures will attack specific targets, etc. I stopped playing it about a month afterward, but I found its take on CCG quite interesting.
Star chamber used to work in that way.
But it super dead.
Random person: RNG isn't that bad
Me: only gets one store in an ftl run
Man that FTL music is so good. What a great game.