Post

So they tried stylometry and that didn't work. Then a modified version and that didn't work. They then found 3 “markers of sociolinguistic variation” that matched Back and Satoshi. And then tried an ad-hoc filtering method on a pool of candidates which gave Back as the best match. Then they stopped.

Old Unreliable

@oldunreliable.bsky.social

8 Apr

This is kind of reminiscent of polygraphs, hair analysis and blood spatter pattern analysis in forensics. These methods may initially have seemed plausible, including to the layperson, but on closer examination turned out to be less reliable than claimed. I view this identification as speculative.