I'm very familiar with the traditional methods of sequencing, including the commonly used forensic techniques[1]. I used to work at a well-known genetics lab way back when everybody used Sanger sequencing. I left right around when the first pyrosequencing machine was available, so I'm not familiar with the newer techniques that were developed after that point.

My question is what these newer methods are storing. Elsewhere in this threat someone mentioned that 23andme was able to inspect >100k SNPs, which would be outstanding data if they can produce it with an acceptable error rate.

> the chance of multiple matches can be less than the worlds population.

See, this is exactly what I was talking about: it depends on what question you are trying to answer. If have a sample from the scene of a crime, we can get very high confidence when matching that sample to even a large number of suspects. The "paradox" happens when you try to ask a different (but related) question that changes the search from one-to-many into many-to-many. Even with a low probability of a match, the larger number of possible matches makes false positives very common.

(if you prefer, when the prior probability is extremely low, testing entire databases will tend to find a lot of false positives)

DNA analysis is a powerful technique, but you still have to be careful to make sure you are asking the right question.

[1] there is viability in how many which unfortunately are somewhat variable, but I'm ignoring that