[Coco] [coco] The German Tank Problem and estimating the number of color computers made

[Jayeson Lee-Steere]

Coco serial numbers do not appear to be sequential across model variations
and/or markets. For example, I've seen an E-board Coco 1 with serial number
well down into the D-board serial number range.

However I thought I'd give it a try with Coco 3's since there are
relatively few variations. I dug around my stash and pics on the net to
come up with this:

https://docs.google.com/spreadsheets/d/12UCzkGX7SALJKt20W8_JO5b8PmF5hPsmzjHv4n-kUMM/edit?usp=sharing

There isn't enough data to tell whether there are gaps in the ranges,
except for the large gap between Korean and USA manufactured machines.
Canadian market machines have this rather different metalized tag with the
serial number stamped rather than printed. I suspect these numbers do fit
into the NTSC numbering sequence, assuming the unhelpful first "1" digit
was dropped. Perhaps it was too hard to stamp the extra digit.

It is not clear to me whether the PAL machines are part of the NTSC
sequencing, or their own independent sequence. The ranges do overlap.

On Fri, Oct 21, 2016 at 4:13 AM, John Guin <johnguin at hotmail.com> wrote:

> The recent thread about math problems got me thinking about the German Tank
> Problem :
> http://www.statisticalconsultants.co.nz/blog/the-german-tank-problem.html
>
>
>
> We can use this rule of thumb to estimate how many color computers were
> made
> (my personal estimate from years ago was 1.3 million).
>
>
>
> Anyhow, here's how to do it:
>
> 1.       First, everything I have seen is that the serial numbers were
> sequential.  There might have been gaps, but that can be compensated (step
> 4)
>
> 2.       We get a list of all the color computer serials numbers everyone
> here owns
>
> 3.       Compare to the serial numbers produced.
>
> 4.       IIRC, each factory was sequential, but I do not know if the Korean
> factory picked up exactly where the other factory left off, but can adjust
> with a constant
>
> 5.       Then just apply the formula
>
> 6.       Repeat step 5 if we get new data
>
>
>
> Seem reasonable?
>
>
>
> John
>
>
>
>
> --
> Coco mailing list
> Coco at maltedmedia.com
> https://pairlist5.pair.net/mailman/listinfo/coco
>