3 ms·
This evidence is inadmissible in general public. You can't go on Twitter, say that the writing style of that and this person is 98% similar per the state of the
by akomtu 5y ago
This evidence is inadmissible in general public. You can't go on Twitter, say that the writing style of that and this person is 98% similar per the state of the art SGERT model and convince the public with that. It's also trivial to for a doxing target to dismiss this evidence as another flaky piece of software that's confusing writing styles. This kind of software is useful, but for more specialized purposes: forensics, intelligence.
- snerbles 5y agoI'd really hate to ruin your naïveté...but sufficiently riled-up mobs have gone after individuals for much, much flimsier evidence than "because this AI says so".
- akomtu 5y agoIn this case a made up evidence would suffice.
- corndoge 5y agoYou don't need any evidence at all to convince the online public of anything it wants to believe, especially if it gives it a target for vitriol. I have to agree with my sibling commenter that believing otherwise is naive given the wealth of evidence to the contrary.
- exporectomy 5y agoAfter they're identified, keen humans will go looking for stronger clues. Maybe the different accounts told the same anecdotes or show they have the same set of opinions or knowledge or the posting times are similar but never coincident or whatever other human-readable evidence.
- uniqueid 5y agoYes. The time of day the user posts can reveal to their time zone. People often leave comments that reveal their age or gender. If the user mentions a business or product name, it might only available in certain part of the world. Many people reuse account names on other services and their content their may have more clues... etc.