Cmon. There's no such thing as a 'false positive' in code crashes. Its a binary thing - correct code doesn't do it. So something is wrong with an OS that crashes for somebody. Its got nothing to do with selection bias.
And if it crashes for certain people chronically, then there's a syndrome that's reproducible. Meaning it is likely to affect a lot of people in similar circumstances.
Crashes aren't new. OS X has had them for a long time, they're just less common than some other operating systems.
The question is whether you can go from "2 or 3 friends of mine said they have more crashes than before" to "10.9.3 is more prone to crashing than 10.9.2." For all we know, the entire sample is using some 3rd party utility that crashes under the new update, which is bad, but is also not something you can completely blame Apple for.
