I'll continue to update the spreadsheet and the BASIC program with the additional serial numbers that have been submitted since recording this video, so if you check out the spreadsheet now it won't exactly match the video anymore. If enough new data comes in that the estimate is substantially off, I'll make a new video about it. Download the .d64 (you can use it in emulators such as VICE): 8bitshowandtell.com/prg/tank1581.d64 and spreadsheet: docs.google.com/spreadsheets/d/1n6Y8h68BaSaLpueSVqrsrzfD5US1TJexpwFYLVVPp88/
This serial numbers looks odd. They are almost like braked in clusters (inside single prefix). Have you think about that serials coult go like this: 4001 4002 4003 ... 4299 and than, by any reason they can jump to 5001 5001 5002 ... like a year of production on chips, if you don't know that fact, you would calculate number of chips wrong. Same may be here. And all calculations may be wrong based on that.
Okurka: I'm working through all the new numbers I got, though the highest JE4 I've seen is 001296 which will "just" amount to a boost of 1500 or so. Of course "substantial" is subjective, but I was thinking if it was outside the estimate by more than +/- 10,000 unit then I'd probably make a video.
The last 1581 was a model for german market. Funkentstört is definitely german an DBP stands for Deutsche Bundespost, translates to german federal postal service.
Going back to my college statistics class.... For a professional grade poll, such as what you'd use to call an election, we need about 382 serial numbers. But, with the data we have, we can be confident the margin of error is 12 percent. Since we don't know where one of the series started, that may mean as few as 43,000 drives, or as many as 70,000 drives. Very cool project, and you've collected enough data to tell us how many more drives we need to have a really good idea of the actual number. Very cool project! (And yes, my 1581's serial number was part of your data set.)
Hey Robin. Enjoyed your video. As a hobby, I collect commodore disk drive serial numbers from eBay pictures and other sources to try to find out how many disk drives still survive in the wild. I have over 100 serial numbers for 1581 drives in my list. If you want a bigger sample size to get a more acurate estimate, let me know where ti send them.
Yes, please email me - address is on the "About" tab of my channel page: ruclips.net/user/8BitShowAndTellabout I'll add them to the spreadsheet, but ideally I hope someone will make a serial number registry for all Commodore products, such as extending the existing Commodore 64 one.
@@8_Bit Hey, Robin, I have an idea on finding out a lot more about how many were sold and how those serial numbers are coded. There's a guy on RUclips, Bil Herd, who was a Commodore engineer when the 1581 was released and he might be able to shed some light on it and maybe even give you some more exact answers. Here's his channel information: ruclips.net/user/BilHerd
My mother worked at a Pay 'n Save (discount and drug store) at the time (early 80's) ..the store also sold 2 computers TI99/4A and C64 (they also sold the Coleco Adam but did not sell many.. almost all returned as defective)..she estimated that the store sold about 5000 C64s and that about 20% were defective or had a problem that the customer returned it for..the store was told to remove the serial number from the computer and cut the serial number from the box (returning them to Commodore for refund to the store) and throw the computer in the dumpster...toward the end as they were discontinuing both computers the TI sold for $5 and the C64 sold for $25 ..the store had a few C64s left and probably 25 TI's that the store gave to employees .. Mom got a TI and brought it home i played with it for about 2 days and it ended up in the closet.. During this time i worked for Tandy Corp servicing the TRS line..
Always a treat to see another of your informative videos. I'm somewhat experienced with statistics and the math when working with probabilities and random samples, and your explanation of the german tank problem is spot on! I hope videos like this will convince younger viewers that the math they study in school is actually very useful later on in life, something many young people don't realize when they are in school ;)
Robin, By the time I had assembled and tested all the 1581 case's I got from Computer house in Chatham, I was up to nearly 900 of them over that 10yr period it took to sell them. I still have about 60 or so empty shell cases (for project boxes) with scratched face-plates. You can take photos of all those at the next WOC you come up for....Joey JP PBM
@@8_Bit That's what I'm saying...They ALL do! And they *should* be the later runs that were left unsold at Commodore-Pharmacy Rd. before contracting ComputerHouse -Chatham to remove the mechs for use in the new A500's that were being made at the time.
It is only guess. There were factories in Germany, the US, Hong Kong and China and Taiwan. The number made in the US must be known because the tax office want their money, in Germany not only the production numbers are painstakingly noted down, also on which day, month and year what specific serial number was produced. In Hong Kong the number don't say much. One roll of stickers with some numbers on it could be copied for the other shift. From the factory complete truckloads were sold without Commodore's knowledge. I assume there were not many who had a diskdrive with their C64, something like four to seven percent?
I am so glad I cleaned my fingernails that day! ;) I can believe the number was around 60,000. It was at the end of the life of the C64 line - and sales had been pretty much dying in America since 87 or 88. This would have been a luxury item for the die hard users but probably not a common item. I had retired my C64 in 1988 after I got my Amiga. I had heard about the 1581, but honestly wondered who would have bought it. Of course, now I own and love mine!
Well, having own a 1581 drive back in 87, I didn't realize how rare these drives are, I until started looking on eBay. Though, I regret that my c128 collection was stolen out of storage from decades ago. Yes, I found it really neat using 3 1/2" floppies over the 5 1/4" format. Liked the increase storage of 800k. I think I was a bit impressed of it's performance over the 1571 drive, since I believe it utilize buffering to write a track at time.
Former GEOS power user here, and boy, that 1581 was one heck of an addition to that graphical OS, even more with TopDesk, the German fork of the GEOS Desktop with fully evolved windowing technology including resizing them and scrolling within them. 😮
Joe Palumbo (JPPBM) had hundreds of 1581 without the actual 3.5" drive inside - the same drive that equipped the Amiga 500 - that he sold for a very low price over the years. I bought 5 in all from him and gave one to Gideon Zweijtzer, the maker of the 1541 Ultimate carts and the Ultimate 64. My last 1581 is serno HB 014755.
A colleague of mine told me that his father invented a scheme to create seemingly random serial numbers to not show how many units were built to their competition. In essence a PRNG with a seed, guaranteed to repeat only after billions of something.
Always great to see others in the C= community make reference to people you know. It is humbling to see how connected we are, regardless of where we live. I knew Joey of JPPBM for a couple of years before I moved from Toronto -- he's been selling gear since the 90s! Doug of 10MARC is a regular participant in our virtual Seattle C= group meetings.
Yes, I enjoy the friendship of the C= community just as much as the computers. This will be the first year since 1998 that I won't be travelling to even one retro computer show; I'm missing seeing friends (and interesting computer stuff)!
@@8_Bit The current virtual meetings via video chat aren't that bad, but lacking in hands-on detail of course. A few months ago we met Nick Vivid -- he joined us and showed his BBS ("Afterlife", using C64 and WiFi modem) as well as talk about his work on CCGMS. And every now and then someone will get their hands on some modern Amiga or C64 gizmo and demo that too. Cool stuff! If you ever wanted to join us one evening and see what we're up to, I'm sure we could schedule a suitable day and time. :)
@@nichderjeniche Thanx for reminding me. Just found it, but it unfortunately doesn't have a sticker at the bottom anymore... So guess i can't help anyway :)
Old Video, but one of the biggest problems with the 1581, was that it basiclaly was only good for data storage. Almost every game ever made was made for the 5-1/4 disks and the 1541. But since I ran a BBS, I had 4 1581s at one point. Til I got the Lt Kernal up and running. Im sure this is why the never sold nearly as many as 1541s.
I'm sure you know this already, but the "annoying space" that leads numbers generated by STR$ is actually in the sign position. When the number is negative, the negative sign goes there and there's no space. This means that your RIGHT$-justify trick would still align the number correctly if the result was negative without explicitly checking for it.
Interesting question, Robin. I have two 1581 drives without serial numbers. I bought on 11/21/1993 my first 1581 drive from a dealer called "2fach Computer" in Germany. The second drive I bought later on ebay.
That sounds a lot like the capture and recapture method on Biology, you are supposed to capture a sample and mark all individuals, then you make a capture again and count the number of individuals that you re captured (the marked ones) and divide by the total individuals (marked and unmarked), you extrapolate that and estimate the total population, with certain reserves depending on the specie.
I just did a quick eBay search for active and already finished 1581 bids. There are some interesting serial numbers as well e.g. JE4 000309 , JE4 000098 , HB 016405 , 00493 , 000201 , 001083 etc ...
Well, there are many 1581 without serial, as they were not made by Commodore. Commodore sold the parts and another company assembled them in Germany or so, at least that is what they wrote in the 64er German magazine that "1581 is back".
@@8_Bit archive.org/details/64er_1994_06 Article is on page 26. However, I could not find any reference that it was assembled in Germany. It just says that the new 1581 are coming from an american distributor and that externally they are not different from the old version.
@@stefanegger Yes, I do not doubt that. With "not being externally different" they only meant the physical dimensions and weight. They didn't mention the serial stickers, maybe they just didn't find that newsworthy enough. Anyway, I've found another little tidbit mentioned here (in the small blue box): archive.org/details/64er_1995_04/page/n13/mode/1up ... It says that during the termination of several field warehouses in the USA large quantities (whatever that means) of 1581 were discovered.
That's very fascinating. I wonder if there's any way to estimate the number of 1581 drives that still exist using some sort of statistical analysis. Also, has anyone done similar analysis for the 1541 and 1571?
motherboard serial numbers don't count toward total sales of a complete 1581 drive. it's entirely possible there were more 1581 boards made than drives, and they were used for other things. edit: although, I doubt they made ˜16,000 more than the drives
I checked your work in Excel. The spreadsheet doesn't completely match with your inputs; for example your HB sheet contains 13 entries while your program data shows 12. The Bayesian estimation numbers are fairly close. In general statisticians would also calculate a confidence interval around the point estimate. As others have noted, it would be unlikely that serial numbers started at 1. As a statistician, I sometimes try to estimate the number of taxicabs within a new city by collecting a small sample of medallion numbers at the airport. Yes I am weird.
Since I recorded the video I've been given more data and have updated both the spreadsheet and the program in the .d64 file to match, that's why the spreadsheet doesn't exactly match, but the .d64 does. You can use it in an emulator like VICE.
Wow, I wish I knew you were doing this. Mine is one of those weird shiny label ones with serial HB015791 and the floppy controller is WDC'85 WD1772-PH (8627).
5:36 It might possibly say "Moder" instead of "Model" due to a mistyping or the fact that many Eastern Asian languages (in Hong Kong's case Cantonese) have difficulties discerning between "L" and "R"
I have a suspicion that the "JA" and "JE" series are separated for the North American and European markets. That odd one out looks to be a German model since it has "Funkentstört" (radio frequency interference shielded) on the sticker. Could even be a GDR (East German) counterfeit. How big are the chances of Commodore having a sense of humour and making the 1581 in a 65,536 unit limited run? ;-)
Toward the end of my C64 (as primary) days, I wanted a 3.5" floppy, but I ended up getting an A600 with the A64 kit. That was a rough emulator thing, most of those games ran far too slow.
I think the numbers are too low since you only sample from the areas where you reached with your call for serial numbers, and there are certainly countries with different model types that did not go into this very much, like germany.
You notice that the motherboards both had numbers below 1,000, so the company in general may not have the practice of starting serial numbers from 10,000. So he first estimate 64,000 also is much closer to the motherboard estimation.
MODER 1581 looks to be a german tag. This may have been a repaired 1581 in Germany. That would mean this replacement tag for original tag. According to google translate "funkentstört" = interference suppressed Or maybe this was a counterfeit 1581.
5:40 is produced for the german market, "Funkentstört nach DBP-Vfg" It should be "Modell" not MODER" maybe lost in translation, also the missing Umlaut "Ö" in funkentstort My 1581 (bought on ebay) had this typo, too. I'm interested if every german model got this MODER typo.
A quick look on ebay: a different label for the german market exists www.ebay.de/itm/BOXED-Commodore-Floppy-Disc-1581-fur-C64-C128-S-N-309-LOW-SERIAL/164339062808?hash=item26435f1818:g:53MAAOSw6q5e7e7s Problem for you: it is a JE4
I have two of those too. Put an Amiga drive in one and a modified PC floppy drive in the other. I heard at one time, he (Joe) had a wall of his garage covered with a stack of those empty 1581 shells. No idea how reliable that info was though...
Or, is it just a bit with tri-state logic and an enable signal? Besides, Old school electronics would set an address to read, then poll it after it was stable, otherwise the float could give you a non-zero or non-one result. And I'm not sure the game grid was a stable place to be anyway.
This is a good application of theory to make estimates. After seeing this video, my estimate would be around 50K of drives made to be sold (this would be a good target number for manufacturing). This would be based off the popularity of the 1541 drive, and the arrival of the 1581 in the C64 life cycle. I don't remember the 1581 being as popular as the 1541 in software offered. So, the demand wasn't there as Commodore hoped. Add in the other computers being offered, and there wasn't too much a need for this drive, unfortunately. As far as the serial numbers, I don't believe that there a lot of jumping around (besides starting at a certain number). It's a lot easier to keep the numbers sequential for record keeping and recall purposes.
It would be interesting to see a plot of the motherboard serial numbers vs the case numbers. I would like to know how many of them are in cases ahead of or behind the numbers from one another.
I don't know if this is relevant, but I've got two of the "chrome" label HB series that were sourced from the pile of drives that Commodore stripped the drive mechanisms from. As far as I know, those drives came from Canada.
There are lots of differing ways manufactures allocate serials. Tektronix in the 60/70's allocated serials starting at 101 anything lower was a prototype. Another scheme I heard of was constructed like this - Country(1d), OEM(2d), plant(1d), line(1d), year(2d), week(2d), sn(4d) starting at 1000 so a serial like CFCAB89381047 was only the 47th unit made on line B week 38 of 1989 by Foxcon in china (this was an easy way to tell if the unit was still under warranty of if they had production problems it allowed quick identification of quarantine units etc..) so if that line only made say ~900 units per week there would be huge holes in the sn allocations. NOTE: this schemes ordering was scrambled so as not to be as obvious, modern HPE computer serials are like this.
I also have some MODER. Its just the very early ones (all have low serial) and its a typing mistake. Normally it shoulds say MODEL.
4 года назад
I would have taken the sum of all serial numbers, divided by the number of serial numbers observed, multiplied by 2 and subtracted 1. If that number is higher than the highest number observed I would have stuck with it, otherwise I would have used the highest number observed plus half the difference for the estimate. (Strange that Wikipedia does not mention this approach, I should update that article...)
No way that commodore could have sold more than 65.536 of those 1581 disk drives: One more and my world would have crashed due to an "illegal quantity error".
Wait, what? ... I started this video and thought to myself: hmm, I own a 1581 drive. Wonder what my serial number is? Well, it turns out... The label is missing... So, whatever number you eventually land on... Add one more 👍
C64 BASIC does have log, but the Bayesian estimations require at least 3 samples because part of the equation is (k-2), and unfortunately I only had 2 data points for a couple series. If we get more data, we can add those to the program too. Spreadsheet and BASIC code is downloadable from the video description if anyone wants to give it a crack.
The last one is made for Germany. But Moder is not a word in german, we write it Modell and the it should say Funkentstört the o has no dots. Since all Commodores i know where made in north America or West Europe i think it's a drive i would like to see the inside of.
The "Moder 1581" really looks like a clone. They also lost the " on the ö when copying the label. Which is kinda weird, as China wasn't into wholesale hardware cloning until two decades later..
That's a neat trick to right-justify the numbers. What I would have done (and probably actually _have_ done before) would have been much more complicated.
As usual, a great informative video. Could you fudge the results so it shows there’s around 2 million 1581’s? That might drive the eBay value down so I can buy one. After that, change it back ;). Just kidding. Look forward to your next video.
Or maybe it’s an inside joke by someone at CBM; they made exactly 65,536 of them (=0x10000), but deliberately kept a few, so that your estimate would be low. Who knows, maybe they planned on making 65,536 C64s, too. 😏
We won't truly know unless we collect all the serial numbers! :) So yes, this estimate is built upon the assumption of sequential numbers starting at 1.
Yeah, a "trit" is a trinary digit, like a bit but instead of just having two states, it has three. You can look up "ternary numeral system" for more info.
Yes, it ended up that I could have done that. At first I was planning on also implementing the Bayesian approaches, and was thinking I'd need to collect everything in an array to possibly do some extra comparisons or calculations. But that fell through when I realized I needed at least 3 data points for those calculations. The program already worked, and I didn't refactor.
I have no reason to doubt your estimate, but...it just feels like there should have been a whole lot more than that. (No proof to back that up either.) If this is correct, then I own ~0.005% of all 1581's ever made. Best drive ever made for the C= 8 bits, IMHO.
Nice explanation and it is sad we don't know for sure how their schemes worked or their exact sales. It is fun to postulate though . In regards to the C64 numbers, I don't buy that 12.5 mil number because it is much lower then what has been stated by Ex-Commodore employees that were there. Plus if it is listed in the Guinness Book of Records as 17-20 mil wouldn't there need to be proof for them to list it at all? You would think for a computer company Commodore would have data coming out their ears but sadly not everything survived of those days.
The one aspect of that C64 number estimate that I don't fully understand is how Steil says that North American machines are over-represented in the registry, and NTSC machines represent only 1/6th of total Commodore 64s produced. I'd like a more thorough explanation and examination of that; I might look into it myself. I want to clarify that I think Steil is brilliant and I have huge respect for him.
That could certainly be it; unfortunately I don't have much information on where the drives were sold. The JA numbers are much higher than the JE, maybe even 6x as many JA than JE. But given that these drives were sold from 1987 to 1990 or so, which is when North American C64/128 sales were supposedly tanking, and European sales were way up, that seems contradictory. Michael Steil, in his article calculating total C64 sales which inspired this video, claims that North American C64s only account for 1/6th of all C64s produced. I find that difficult to believe, especially in light of these 1581 numbers. Does that mean that per C64 system, a 1581 was 36x more likely to be purchased for an NTSC machine than PAL? Despite the stories, it was mostly just the UK that stuck with cassette on the C64. Germany and many other European countries switched to disk. Well, seems I have a few thoughts about this! Thanks for the comment.
@@8_Bit The only C64's I ever ran into in the wild here in Norway, used cassette decks as opposed to floppy drives. I should probably do some more research on this in the future, but I always had the impression that computers like the C64 was used almost exclusively by "kids" (teenagers I suppose) over here, and those using computers for more serious purposes (i.e. "adults") had an IBM-compatible. And since music cassettes were readily available, it would be very convenient for a young person to use that as their storage medium for their computers. Also it allowed for very easy software piracy, as a lot of people had access to a stereo with dual cassette deck with high-speed dubbing.
@Tom M33 Don't misunderstand me. I never claimed the C64 couldn't do office type work. It certainly could. But a large quantity of systems were marketed and sold due to its gaming abilities.
We are talking about the drive shown at 5:30. I agree it's likely a bootlef, as they tried to print the German equivalent to an FCC compliance statement on it - and failed in the usual way: They messed up the umlaut. It should read "FUNKENTSTÖRT n. DBP-Vfg 1046/84", with abbreviations expanded, it reads "Funkentstört nach Deutsche-Bundespost-Verfügung 1046/84", which roughly translates to "Interference suppression according to German Federal Postal Service bill 1046/84". In Germany, the Federal Postal Service also was responsible for telecommunication. It went so far that they had the monopoly on all kinds of communication between different premises, and thus also were responsible for keeping the radio spectrum clean.
I'll continue to update the spreadsheet and the BASIC program with the additional serial numbers that have been submitted since recording this video, so if you check out the spreadsheet now it won't exactly match the video anymore. If enough new data comes in that the estimate is substantially off, I'll make a new video about it. Download the .d64 (you can use it in emulators such as VICE): 8bitshowandtell.com/prg/tank1581.d64 and spreadsheet: docs.google.com/spreadsheets/d/1n6Y8h68BaSaLpueSVqrsrzfD5US1TJexpwFYLVVPp88/
I just sent you an email with attached pictures showing the serial numbers from my two 1581 drives.
This serial numbers looks odd. They are almost like braked in clusters (inside single prefix). Have you think about that serials coult go like this:
4001
4002
4003
...
4299
and than, by any reason they can jump to
5001
5001
5002
...
like a year of production on chips, if you don't know that fact, you would calculate number of chips wrong.
Same may be here.
And all calculations may be wrong based on that.
The estimate is substantially off when you add series JE4 to the spreadsheet.
here my contribution: imgur.com/gallery/0K76mMr
Okurka: I'm working through all the new numbers I got, though the highest JE4 I've seen is 001296 which will "just" amount to a boost of 1500 or so. Of course "substantial" is subjective, but I was thinking if it was outside the estimate by more than +/- 10,000 unit then I'd probably make a video.
Well that was a more detailed response than I expected! 🤣 Nice work Robin, cheers! 👍🕹
Your friend in retro, Perifractic
Oh-oh so you're only "some guy" among your YT colleages 😭 This guy is brilliant, folks! 👍
The last 1581 was a model for german market. Funkentstört is definitely german an DBP stands for Deutsche Bundespost, translates to german federal postal service.
Going back to my college statistics class.... For a professional grade poll, such as what you'd use to call an election, we need about 382 serial numbers. But, with the data we have, we can be confident the margin of error is 12 percent. Since we don't know where one of the series started, that may mean as few as 43,000 drives, or as many as 70,000 drives. Very cool project, and you've collected enough data to tell us how many more drives we need to have a really good idea of the actual number. Very cool project! (And yes, my 1581's serial number was part of your data set.)
Hey Robin. Enjoyed your video. As a hobby, I collect commodore disk drive serial numbers from eBay pictures and other sources to try to find out how many disk drives still survive in the wild.
I have over 100 serial numbers for 1581 drives in my list. If you want a bigger sample size to get a more acurate estimate, let me know where ti send them.
Yes, please email me - address is on the "About" tab of my channel page: ruclips.net/user/8BitShowAndTellabout
I'll add them to the spreadsheet, but ideally I hope someone will make a serial number registry for all Commodore products, such as extending the existing Commodore 64 one.
Man, that is the most niche hobby I've ever heard of! Could be useful to Robin, though.
@@8_Bit That's where I got the idea of collecting serial numbers for disk drives, from Peter Rittwage excellent C64 serial number registry.
@@8_Bit Hey, Robin, I have an idea on finding out a lot more about how many were sold and how those serial numbers are coded. There's a guy on RUclips, Bil Herd, who was a Commodore engineer when the 1581 was released and he might be able to shed some light on it and maybe even give you some more exact answers. Here's his channel information: ruclips.net/user/BilHerd
This is the magic of the internet, wow!
My mother worked at a Pay 'n Save (discount and drug store) at the time (early 80's) ..the store also sold 2 computers TI99/4A and C64 (they also sold the Coleco Adam but did not sell many.. almost all returned as defective)..she estimated that the store sold about 5000 C64s and that about 20% were defective or had a problem that the customer returned it for..the store was told to remove the serial number from the computer and cut the serial number from the box (returning them to Commodore for refund to the store) and throw the computer in the dumpster...toward the end as they were discontinuing both computers the TI sold for $5 and the C64 sold for $25 ..the store had a few C64s left and probably 25 TI's that the store gave to employees ..
Mom got a TI and brought it home i played with it for about 2 days and it ended up in the closet..
During this time i worked for Tandy Corp servicing the TRS line..
Always a treat to see another of your informative videos.
I'm somewhat experienced with statistics and the math when working with probabilities and random samples, and your explanation of the german tank problem is spot on!
I hope videos like this will convince younger viewers that the math they study in school is actually very useful later on in life, something many young people don't realize when they are in school ;)
Robin, By the time I had assembled and tested all the 1581 case's I got from Computer house in Chatham, I was up to nearly 900 of them over that 10yr period it took to sell them. I still have about 60 or so empty shell cases (for project boxes) with scratched face-plates. You can take photos of all those at the next WOC you come up for....Joey JP PBM
Cool, thanks Joey! None of those remaining shell cases have serial numbers I assume?
@@8_Bit That's what I'm saying...They ALL do! And they *should* be the later runs that were left unsold at Commodore-Pharmacy Rd. before contracting ComputerHouse -Chatham to remove the mechs for use in the new A500's that were being made at the time.
It is only guess. There were factories in Germany, the US, Hong Kong and China and Taiwan. The number made in the US must be known because the tax office want their money, in Germany not only the production numbers are painstakingly noted down, also on which day, month and year what specific serial number was produced. In Hong Kong the number don't say much. One roll of stickers with some numbers on it could be copied for the other shift. From the factory complete truckloads were sold without Commodore's knowledge. I assume there were not many who had a diskdrive with their C64, something like four to seven percent?
I am so glad I cleaned my fingernails that day! ;)
I can believe the number was around 60,000. It was at the end of the life of the C64 line - and sales had been pretty much dying in America since 87 or 88. This would have been a luxury item for the die hard users but probably not a common item.
I had retired my C64 in 1988 after I got my Amiga. I had heard about the 1581, but honestly wondered who would have bought it.
Of course, now I own and love mine!
Well, having own a 1581 drive back in 87, I didn't realize how rare these drives are, I until started looking on eBay. Though, I regret that my c128 collection was stolen out of storage from decades ago. Yes, I found it really neat using 3 1/2" floppies over the 5 1/4" format. Liked the increase storage of 800k. I think I was a bit impressed of it's performance over the 1571 drive, since I believe it utilize buffering to write a track at time.
@@marcuswilliams3455 oh how terrible that your C128 was stolen! Yikes!
Former GEOS power user here, and boy, that 1581 was one heck of an addition to that graphical OS, even more with TopDesk, the German fork of the GEOS Desktop with fully evolved windowing technology including resizing them and scrolling within them. 😮
Fascinating! Thanks to everyone who volunteered their serial numbers.
Joe Palumbo (JPPBM) had hundreds of 1581 without the actual 3.5" drive inside - the same drive that equipped the Amiga 500 - that he sold for a very low price over the years. I bought 5 in all from him and gave one to Gideon Zweijtzer, the maker of the 1541 Ultimate carts and the Ultimate 64. My last 1581 is serno HB 014755.
A colleague of mine told me that his father invented a scheme to create seemingly random serial numbers to not show how many units were built to their competition.
In essence a PRNG with a seed, guaranteed to repeat only after billions of something.
Nice to see JPPBMs photo in the video :) I have two 1581's in storage makes me want to pull them out and take a look at the serial numbers.
Michael Steil is an amazing presenter. The has done the best 6502 explanation ever. Also, his AGC presentation was really great too.
Always great to see others in the C= community make reference to people you know. It is humbling to see how connected we are, regardless of where we live. I knew Joey of JPPBM for a couple of years before I moved from Toronto -- he's been selling gear since the 90s! Doug of 10MARC is a regular participant in our virtual Seattle C= group meetings.
Yes, I enjoy the friendship of the C= community just as much as the computers. This will be the first year since 1998 that I won't be travelling to even one retro computer show; I'm missing seeing friends (and interesting computer stuff)!
@@8_Bit The current virtual meetings via video chat aren't that bad, but lacking in hands-on detail of course. A few months ago we met Nick Vivid -- he joined us and showed his BBS ("Afterlife", using C64 and WiFi modem) as well as talk about his work on CCGMS. And every now and then someone will get their hands on some modern Amiga or C64 gizmo and demo that too. Cool stuff! If you ever wanted to join us one evening and see what we're up to, I'm sure we could schedule a suitable day and time. :)
This needs a second part with more entries from the yt audience!
Yep, i have one too
@@zaitarh give him your serial :)
On eBay is one now with this serial number:
JE4 000309
@@nichderjeniche Thanx for reminding me. Just found it, but it unfortunately doesn't have a sticker at the bottom anymore... So guess i can't help anyway :)
Old Video, but one of the biggest problems with the 1581, was that it basiclaly was only good for data storage. Almost every game ever made was made for the 5-1/4 disks and the 1541. But since I ran a BBS, I had 4 1581s at one point. Til I got the Lt Kernal up and running. Im sure this is why the never sold nearly as many as 1541s.
I'm sure you know this already, but the "annoying space" that leads numbers generated by STR$ is actually in the sign position. When the number is negative, the negative sign goes there and there's no space. This means that your RIGHT$-justify trick would still align the number correctly if the result was negative without explicitly checking for it.
Interesting question, Robin.
I have two 1581 drives without serial numbers. I bought on 11/21/1993 my first 1581 drive from a dealer called "2fach Computer" in Germany. The second drive I bought later on ebay.
Me too, my 1581 never had a serial number! Or a type label. It only had a warranty label on the back (not on one screw like others have).
Are you counting variants of the drive, like the 1581-II ? My drive has the serial number BLS 088478.
Is there any info how good those wartime estimates turned out to be?
I'll bet I've got a half dozen 1581's. I wouldn't guarantee any of them to be in working order, though. Does that matter?
Digging the Bedford Level Experiment. Thanks for the introduction!
Mine is like new, has no sticker on the bottom of the drive, but the serial number on the the board inside is JA0024448.
using a commodore as a tool is great, most people would bring out their phone or their windows computers. Love the video
"You might recognize that thumb". LMAO, greatness!
😄
2:57 I just love intelligent humor! 😂🤣🙃👍👍👍
That sounds a lot like the capture and recapture method on Biology, you are supposed to capture a sample and mark all individuals, then you make a capture again and count the number of individuals that you re captured (the marked ones) and divide by the total individuals (marked and unmarked), you extrapolate that and estimate the total population, with certain reserves depending on the specie.
I just did a quick eBay search for active and already finished 1581 bids. There are some interesting serial numbers as well e.g. JE4 000309 , JE4 000098 , HB 016405 , 00493 , 000201 , 001083 etc ...
Well, there are many 1581 without serial, as they were not made by Commodore. Commodore sold the parts and another company assembled them in Germany or so, at least that is what they wrote in the 64er German magazine that "1581 is back".
Very interesting, do you know what issue # or date that 64er was published?
@@8_Bit archive.org/details/64er_1994_06
Article is on page 26. However, I could not find any reference that it was assembled in Germany. It just says that the new 1581 are coming from an american distributor and that externally they are not different from the old version.
@@guntergreil5082 OK maybe not Germany but they are different, they do not have serial number sticker on bottom.
@@stefanegger Yes, I do not doubt that. With "not being externally different" they only meant the physical dimensions and weight. They didn't mention the serial stickers, maybe they just didn't find that newsworthy enough. Anyway, I've found another little tidbit mentioned here (in the small blue box): archive.org/details/64er_1995_04/page/n13/mode/1up ... It says that during the termination of several field warehouses in the USA large quantities (whatever that means) of 1581 were discovered.
That's very fascinating.
I wonder if there's any way to estimate the number of 1581 drives that still exist using some sort of statistical analysis.
Also, has anyone done similar analysis for the 1541 and 1571?
motherboard serial numbers don't count toward total sales of a complete 1581 drive. it's entirely possible there were more 1581 boards made than drives, and they were used for other things.
edit: although, I doubt they made ˜16,000 more than the drives
That was seriously cool ! Greetings from a fellow software developer.
I checked your work in Excel. The spreadsheet doesn't completely match with your inputs; for example your HB sheet contains 13 entries while your program data shows 12. The Bayesian estimation numbers are fairly close. In general statisticians would also calculate a confidence interval around the point estimate. As others have noted, it would be unlikely that serial numbers started at 1. As a statistician, I sometimes try to estimate the number of taxicabs within a new city by collecting a small sample of medallion numbers at the airport. Yes I am weird.
Since I recorded the video I've been given more data and have updated both the spreadsheet and the program in the .d64 file to match, that's why the spreadsheet doesn't exactly match, but the .d64 does. You can use it in an emulator like VICE.
Wow, I wish I knew you were doing this. Mine is one of those weird shiny label ones with serial HB015791 and the floppy controller is WDC'85 WD1772-PH (8627).
It's almost like an episode of Vsauce or Numberophile. Very interesting as usual, thanks Robin!
Sypialnia Studio If they didn’t make 65535 I’m disappointed!
18 dislikes from Commodore's "secret manufacturing plant" their secret is out.... lol
5:36 It might possibly say "Moder" instead of "Model" due to a mistyping or the fact that many Eastern Asian languages (in Hong Kong's case Cantonese) have difficulties discerning between "L" and "R"
I've heard about this "problem" since the 1980's. It's still unreal to me that a company would not track every unit produced! Just unreal man! lol
My 1581 has no label. I bought it new at the computer store. I think, it never had a label.
I have a suspicion that the "JA" and "JE" series are separated for the North American and European markets.
That odd one out looks to be a German model since it has "Funkentstört" (radio frequency interference shielded) on the sticker. Could even be a GDR (East German) counterfeit.
How big are the chances of Commodore having a sense of humour and making the 1581 in a 65,536 unit limited run? ;-)
Toward the end of my C64 (as primary) days, I wanted a 3.5" floppy, but I ended up getting an A600 with the A64 kit. That was a rough emulator thing, most of those games ran far too slow.
1581 Was 3 years available. Calculating with 250 working days = 750 working days. 64k/750 Is 85 drives produced per working day on avarage.
I think the numbers are too low since you only sample from the areas where you reached with your call for serial numbers, and there are certainly countries with different model types that did not go into this very much, like germany.
You notice that the motherboards both had numbers below 1,000, so the company in general may not have the practice of starting serial numbers from 10,000. So he first estimate 64,000 also is much closer to the motherboard estimation.
MODER 1581 looks to be a german tag.
This may have been a repaired 1581 in Germany. That would mean this replacement tag for original tag.
According to google translate "funkentstört" =
interference suppressed
Or maybe this was a counterfeit 1581.
does this 1581 have a pc-compatible drive mechanic inside?
should we create a 1581 rebuild with a compatibility mode for 1541?
Nice, Just got a 1581 drive myself. What a great drive! Just need to work out how to create disks on a modern PC using D81 images!
5:40 is produced for the german market, "Funkentstört nach DBP-Vfg"
It should be "Modell" not MODER" maybe lost in translation, also the missing Umlaut "Ö" in funkentstort
My 1581 (bought on ebay) had this typo, too. I'm interested if every german model got this MODER typo.
A quick look on ebay: a different label for the german market exists www.ebay.de/itm/BOXED-Commodore-Floppy-Disc-1581-fur-C64-C128-S-N-309-LOW-SERIAL/164339062808?hash=item26435f1818:g:53MAAOSw6q5e7e7s
Problem for you: it is a JE4
Thanks, I've now been sent 4 drives from JE4!
I think I have one of those JPPBM units. I put an A500 mech into it and got myself a 1581.
I have one, I got an Amiga 500 for free and the drive works perfect.
I have two of those too. Put an Amiga drive in one and a modified PC floppy drive in the other. I heard at one time, he (Joe) had a wall of his garage covered with a stack of those empty 1581 shells. No idea how reliable that info was though...
How do I get a hold of JPPBM? Thanks.
Listening to the end music, Bit in TRON isn't a bit because it has 3 states: YES, NO and silence... and the song mentions this! 😂
Yes, Bit is really Trit! :) Full song video here: ruclips.net/video/Appmg64tZB0/видео.html
Thank you :)
Or, is it just a bit with tri-state logic and an enable signal? Besides, Old school electronics would set an address to read, then poll it after it was stable, otherwise the float could give you a non-zero or non-one result. And I'm not sure the game grid was a stable place to be anyway.
nice breakdown
This is a good application of theory to make estimates. After seeing this video, my estimate would be around 50K of drives made to be sold (this would be a good target number for manufacturing). This would be based off the popularity of the 1541 drive, and the arrival of the 1581 in the C64 life cycle. I don't remember the 1581 being as popular as the 1541 in software offered. So, the demand wasn't there as Commodore hoped. Add in the other computers being offered, and there wasn't too much a need for this drive, unfortunately.
As far as the serial numbers, I don't believe that there a lot of jumping around (besides starting at a certain number). It's a lot easier to keep the numbers sequential for record keeping and recall purposes.
I also have one (the Netherlands) but no serial sticker on the bottem. Don't want to open it, is there another way to read the serial?
It would be interesting to see a plot of the motherboard serial numbers vs the case numbers. I would like to know how many of them are in cases ahead of or behind the numbers from one another.
What drive mechanism is used in a 1581? If, as you say, some were parted out for Amiga use, I would assume mostly the Chinon 354 ?!
I don't know if this is relevant, but I've got two of the "chrome" label HB series that were sourced from the pile of drives that Commodore stripped the drive mechanisms from. As far as I know, those drives came from Canada.
If you want to send them in, email address is on this page: ruclips.net/user/8BitShowAndTellabout
@@8_Bit Sent!
@@f15simDid he send the labels or the drives? 😮
JUST KIDDING! 😊
I just got a photo of the bottom of my 1581 you said on Facebook sent you them any link how to do that?
Neat little project, sorry I missed out. I have 5 of those myself, but none of my serials would have changed the outcome.
If you want to send them in, I'm updating the spreadsheet still. Email address is on this page: ruclips.net/user/8BitShowAndTellabout
There are lots of differing ways manufactures allocate serials. Tektronix in the 60/70's allocated serials starting at 101 anything lower was a prototype. Another scheme I heard of was constructed like this - Country(1d), OEM(2d), plant(1d), line(1d), year(2d), week(2d), sn(4d) starting at 1000 so a serial like CFCAB89381047 was only the 47th unit made on line B week 38 of 1989 by Foxcon in china (this was an easy way to tell if the unit was still under warranty of if they had production problems it allowed quick identification of quarantine units etc..) so if that line only made say ~900 units per week there would be huge holes in the sn allocations. NOTE: this schemes ordering was scrambled so as not to be as obvious, modern HPE computer serials are like this.
Awesome estimation.
My 1581 was bought 1991. It was totally new, but lacked a Serial Number Sticker.
Interesting. Using the motherboard serial number would probably be a better method, but it's also more difficult to collect that number.
Right, there are several New-in-Box 1581 on eBay without the serial label.
Love the end music...
Thanks, it's from an EP my band did a few years ago: bedfordlevelexperiment.bandcamp.com/album/living-here-in-the-future
the real problem is to have a truly random sample out of a non-uniform distribution; the real world is globby. but 64k is a nice number :-)
Heh, I saw the request for serial numbers in the forum64.de. And now Robin makes a video about it. Small world. :)
I’ll share mine when I get back home. Any other devices we want to collect data on? I have a 128D as well.
I also have some MODER. Its just the very early ones (all have low serial) and its a typing mistake. Normally it shoulds say MODEL.
I would have taken the sum of all serial numbers, divided by the number of serial numbers observed, multiplied by 2 and subtracted 1. If that number is higher than the highest number observed I would have stuck with it, otherwise I would have used the highest number observed plus half the difference for the estimate. (Strange that Wikipedia does not mention this approach, I should update that article...)
No way that commodore could have sold more than 65.536 of those 1581 disk drives: One more and my world would have crashed due to an "illegal quantity error".
Wait, what? ... I started this video and thought to myself: hmm, I own a 1581 drive. Wonder what my serial number is?
Well, it turns out... The label is missing... So, whatever number you eventually land on... Add one more 👍
It's strange how many lack serial numbers!
So C64 don't have a logarithm function for Bayesian estimation?
C64 BASIC does have log, but the Bayesian estimations require at least 3 samples because part of the equation is (k-2), and unfortunately I only had 2 data points for a couple series. If we get more data, we can add those to the program too. Spreadsheet and BASIC code is downloadable from the video description if anyone wants to give it a crack.
The last one is made for Germany. But Moder is not a word in german, we write it Modell and the it should say Funkentstört the o has no dots.
Since all Commodores i know where made in north America or West Europe i think it's a drive i would like to see the inside of.
The "Moder 1581" really looks like a clone. They also lost the " on the ö when copying the label. Which is kinda weird, as China wasn't into wholesale hardware cloning until two decades later..
And now I want a 1581!!!
That's a neat trick to right-justify the numbers. What I would have done (and probably actually _have_ done before) would have been much more complicated.
I think I once saw an even more elegant way to right-justify numbers buried in an old magazine or something, but have no idea how to find it again :)
As usual, a great informative video. Could you fudge the results so it shows there’s around 2 million 1581’s? That might drive the eBay value down so I can buy one. After that, change it back ;). Just kidding. Look forward to your next video.
Or maybe it’s an inside joke by someone at CBM; they made exactly 65,536 of them (=0x10000), but deliberately kept a few, so that your estimate would be low. Who knows, maybe they planned on making 65,536 C64s, too. 😏
Are the serial numbers really sequential though?
We won't truly know unless we collect all the serial numbers! :) So yes, this estimate is built upon the assumption of sequential numbers starting at 1.
I have a JA2 000047 - can send a pic, but not sure where to send. seems like JA2 is a different series altogether
Thanks, I've got that in the spreadsheet now
I need this for my emu sp12
what is a trit? is that a real thing in that song?
Yeah, a "trit" is a trinary digit, like a bit but instead of just having two states, it has three. You can look up "ternary numeral system" for more info.
I have an 1581 without a serial number on the case (no sticker). The Mainboard has serial # JE0000859. Let me know if you need a picture of that.
Hey,I have one that's ja1 series serial 0000192 and motherboard number 52 If you'd like to add it to your video let me know !
Great story...I love when math is used to save lives.
here’s my 1581 serial no. JE2 000375 manufactured in March 1988. Made in Taiwan
Instead of collecting all of the results into an array and printing them out, why not just print the formatted results as you compute them?
Yes, it ended up that I could have done that. At first I was planning on also implementing the Bayesian approaches, and was thinking I'd need to collect everything in an array to possibly do some extra comparisons or calculations. But that fell through when I realized I needed at least 3 data points for those calculations. The program already worked, and I didn't refactor.
8-Bit Show And Tell ah, makes sense!
The price of a genuine 1581 will now skyrocket based on these statistics. 😂
I have 4 of 1581 drives!
So, how many sx64's and gs's were made? 😅
It'd be interesting to know, especially the SX-64 since I have one. And I'd love a GS but the prices are crazy now.
I'm envious of people who own one.
I have no reason to doubt your estimate, but...it just feels like there should have been a whole lot more than that. (No proof to back that up either.) If this is correct, then I own ~0.005% of all 1581's ever made. Best drive ever made for the C= 8 bits, IMHO.
It is a great drive, though I think CMD's FD-2000 improved on it even further (except in the looks department).
Don't know from personal experience, although the FD-4000 (one of two or three in existence) is really awesome. I think it looks just fine.
Nice explanation and it is sad we don't know for sure how their schemes worked or their exact sales. It is fun to postulate though . In regards to the C64 numbers, I don't buy that 12.5 mil number because it is much lower then what has been stated by Ex-Commodore employees that were there. Plus if it is listed in the Guinness Book of Records as 17-20 mil wouldn't there need to be proof for them to list it at all?
You would think for a computer company Commodore would have data coming out their ears but sadly not everything survived of those days.
corporations like to fudge shipped units, spare parts inventory, etc...
The one aspect of that C64 number estimate that I don't fully understand is how Steil says that North American machines are over-represented in the registry, and NTSC machines represent only 1/6th of total Commodore 64s produced. I'd like a more thorough explanation and examination of that; I might look into it myself. I want to clarify that I think Steil is brilliant and I have huge respect for him.
I've got 3 drives, u want the serials?
If you want to send them in, I'm updating the spreadsheet still. Email address is on this page: ruclips.net/user/8BitShowAndTellabout
@@8_Bit Mail has been send...
Could it be that "JA" was manufactured for the American market, and "JE" for the European? Or is that too obvious?
That could certainly be it; unfortunately I don't have much information on where the drives were sold. The JA numbers are much higher than the JE, maybe even 6x as many JA than JE. But given that these drives were sold from 1987 to 1990 or so, which is when North American C64/128 sales were supposedly tanking, and European sales were way up, that seems contradictory. Michael Steil, in his article calculating total C64 sales which inspired this video, claims that North American C64s only account for 1/6th of all C64s produced. I find that difficult to believe, especially in light of these 1581 numbers. Does that mean that per C64 system, a 1581 was 36x more likely to be purchased for an NTSC machine than PAL? Despite the stories, it was mostly just the UK that stuck with cassette on the C64. Germany and many other European countries switched to disk. Well, seems I have a few thoughts about this! Thanks for the comment.
@@8_Bit The only C64's I ever ran into in the wild here in Norway, used cassette decks as opposed to floppy drives. I should probably do some more research on this in the future, but I always had the impression that computers like the C64 was used almost exclusively by "kids" (teenagers I suppose) over here, and those using computers for more serious purposes (i.e. "adults") had an IBM-compatible. And since music cassettes were readily available, it would be very convenient for a young person to use that as their storage medium for their computers. Also it allowed for very easy software piracy, as a lot of people had access to a stereo with dual cassette deck with high-speed dubbing.
@Tom M33 Don't misunderstand me. I never claimed the C64 couldn't do office type work. It certainly could. But a large quantity of systems were marketed and sold due to its gaming abilities.
that one from hong kong looks like a boot leg
We are talking about the drive shown at 5:30. I agree it's likely a bootlef, as they tried to print the German equivalent to an FCC compliance statement on it - and failed in the usual way: They messed up the umlaut. It should read "FUNKENTSTÖRT n. DBP-Vfg 1046/84", with abbreviations expanded, it reads "Funkentstört nach Deutsche-Bundespost-Verfügung 1046/84", which roughly translates to "Interference suppression according to German Federal Postal Service bill 1046/84". In Germany, the Federal Postal Service also was responsible for telecommunication. It went so far that they had the monopoly on all kinds of communication between different premises, and thus also were responsible for keeping the radio spectrum clean.
Make that two. We had one also.
MODER = Model Europe ?
Store clips looked like a Software Etc..
I have HB 14746 here :)