Editorial on Ellison, McNealy, and national ID card

Clay Shirky clay@shirky.com
Thu, 25 Oct 2001 13:44:45 -0400 (EDT)


> Any ideas on howbig this DNA GUID might be (in bits)

Very good question, don't know the answer.

Genotypes are differentiated by analysing genetic sites called Short
Tandem Repeats, or STRs or microsatellites.  These are places where
all members of a species have some short DNA sequence repeated in
tandem. (STR is sometimes pronounced 'stutter', for obvious reasons.)

The sites where these STRs occur are the same, as are the sequences
themselves. What differs is the number of repeats.

I seem to remember (though can't easily find now) a reference saying
that commonality in two samples of something like a dozen STR sites is
enough to rule out false positives. 

Given that any numerical space that could positively rule out
namespace collisions in DNA would need to be only order 10**10, I
suppose this means that the number of repeats thet could exist at any
given site is only in the hundreds.

-clay