How many c64 sold?

[Gryzor]

A nice video.

Spoiler: ShowHide

Around 12 million

[Prodatron]

There is always this myth, the C64 has been sold so much more than any other 8bit homecomputer.

But when you compare it with e.g.
- Amstrad PCW: 8 mio
- MSX: 9 mio

It's not such a big advance. Maybe only compared to the CPC (3 mio) and the ZX Spectrum (5 mio)
Source: english Wikipedia article of each system

[Gryzor]

True. I think that the 30 million number has been proven as bullshit for a while now.

[ComSoft6128]

But what about the American market size?
Then the European.....

[MaV]

Michael Steil got quite a bit of flak from some for his estimation by applying the "German Tank Problem" to the serial numbers of the C64s.
Turns out he is proven right. And yes, quite a few guys who worked in sales at Commodore said about the same.


How many Commodore 64 computers were really sold? – pagetable.com
German tank problem - Wikipedia

[Gryzor]

Yes I remember that. It kept me busy reading about the theory for a bit

[tjohnson]

I only ever remember coming across 1 commodore 64 in my childhood, pretty much everyone I knew owned either an amstrad or spectrum.  I wonder how many uk sales there were.

[Bryce]

Michael Steil got quite a bit of flak from some for his estimation by applying the "German Tank Problem" to the serial numbers of the C64s.
Turns out he is proven right. And yes, quite a few guys who worked in sales at Commodore said about the same.


How many Commodore 64 computers were really sold? – pagetable.com
German tank problem - Wikipedia

The problem with using the German Tank method for non-war related production is that production volumes vary a lot in commercial products, military production during a war is "produce as many as possible all the time".

Bryce.

[Gryzor]

The problem with using the German Tank method for non-war related production is that production volumes vary a lot in commercial products, military production during a war is "produce as many as possible all the time".

Bryce.

Can you expand on that? I mean, in war or peace alike, it's always a matter of resources and competing needs. Even in peace you can reverse the question somewhat and the production schedule becomes "produce as many as possible all the time" as long as such price points are met to follow the demand elasticity. Inversely, on the other hand, "as many as possible" isn't really logical even in a war of attrition without setting the parameters of how many you actually need (=sell) or what the cost in terms of other goods is. Would it really matter if you could produce 1.000.000 harnesses a month for an army that's reliant on jets, say?

I'm really not sure about how much sense I'm making - apologies, have been up since 4am today But the bottom line is, why wouldn't the GTM(TM) apply in a consumer good?

[pelrun]

The German Tank Problem doesn't depend on production speed at all, though, only that the serial numbers used were consecutive - or could be converted into a consecutive sequence with a bit of arithmetic.

[Gryzor]

That was my understanding too. How does speed or quantity affect distribution?

[MaV]

Michael Steil explains it well in his blog post. He breaks the serial numbers down to board revisions which all start at 0, and almost all of the numbers increase linearly. One graph shows the serial numbers sorted by board revision then by number.

Sure, there are some assumptions being made, but having a big database of serial numbers, and differentiating between various board revisions over the full life-time of the C64 surely helped a lot in shrinking the error margin of the estimation quite a bit.
Michael Steil's blog post does explain in detail how he came to his conclusion.

[eto]

The German Tank Problem doesn't depend on production speed at all,

As fas as I understood, they used the math to calculate the production rate of tanks and other material. However, the German tank problem itself does not consider production rate at all. It's just "how many items have been produced until now". If you apply this every couple of weeks, you get the production rate indirectly.

[arkive]

Turns out he is proven right.

There are no "proofs" in vast historical subject like this one - only estimates. C64 is not my favourite micro but I have no problem with the high end of the 12-17 mil estimate. Overall it was the best machine, popular all over the world, so why not?

[MaV]

Michael Steil is a Commodore fan. If he and Commodore History's video both point to the total sales being lower, one an exercise in statistical theory and the other concrete sales numbers from the annual sales reports of the 80s, and if the estimated numbers confirm themselves, that is pretty much the best approximation to the true sales number which we probably will ever know.
Is "to be proven right" in this context not a valid phrase to express confirmation? Such as "This Youtube video has demonstrated that he is right"?

I own a few Commodore C64's and Amigas, there's nothing wrong with them (for the record).
Some assume that the sales numbers were much higher to which I will point out that there is good "proof" that it was 12 million give or take. I know a lot of Commodore users, and this will be a nice topic while having a few beers (probably/hopefully even soon), right before we return to listening to the latest SID tunes and trying to make out the flaws in the current SID replacements.

[arkive]

It's irrelevant if these sources are Commodore fans or not, since they are dealing with a subject where it's nigh on impossible to provide definitive answers. And "Proven right" is a pretty definitive statement, meaning that the 12 million figure is absolutely right. Coming from two (2) sources it's not a very convincing proof. The 17 million figure is also based on annual sales reports and is no better or worse than the other claims.

The point is, that these guys "know" how many C64s were sold. Me, I don't know. Maybe it was 12mil, maybe 17, or maybe 13,455. For that reason I prefer using estimates when talking about this subject, instead of absolute figures.

[MaV]

Look: "The estimation has been proven right". I'm not revoking what I said, but I'm not going to endlessly argue about this. 17 million is way over the mark, about 12 million is right. This is as close as it gets with the evidence that we've got.

The 17 million figure adds 5 million C128 sales to supposed 12 million C64 sales, whereas the video has evidence that the sales numbers (12 mill.) *include* C128 sales (IIRC, in the video he shows that the C64 family contains all types of C64s and the C128, re-watch it). It's all there.
I can accept your 17 million figure as unlikely based on the evidence from the video, but unless we find the definitive answer, we're at an impasse.

My statement " Michael Steil is a Commodore fan. " is done more in self-defense, as some people react quite aggressively when someone from "outside" makes claims on their beloved subject. It's in the manner of your " C64 is not my favourite micro ...".

I'm happy to return to this thread to tell you I'm wrong, should I ever hear of the 5 mill. C128 as a separate sales figure to the C64's.

But there's really nothing to be added to that argument now, right?

[Gryzor]

It's irrelevant if these sources are Commodore fans or not, since they are dealing with a subject where it's nigh on impossible to provide definitive answers. And "Proven right" is a pretty definitive statement, meaning that the 12 million figure is absolutely right. Coming from two (2) sources it's not a very convincing proof. The 17 million figure is also based on annual sales reports and is no better or worse than the other claims.

The point is, that these guys "know" how many C64s were sold. Me, I don't know. Maybe it was 12mil, maybe 17, or maybe 13,455. For that reason I prefer using estimates when talking about this subject, instead of absolute figures.

I really don't get your point... You accept something vague, you don't accept something more concrete because "you don't know"?