1. Sapling will give a result no matter how small the input text is. So it will give a result even if there is only 10 words, which is not a big enough sample size for it to give an accurate result.
The posts of the user in question are not short by any means, with around 800-1100 characters on average. Posts are long and detailed
his is why its good to use another detector (or two) in concert with Sapling, to rule out the chance that these are (semi) false positives.
That's the catch, though, otherwise, I would have already posted the username in public. This is because Google Translate uses AI. Sapling detects this while the other detectors don't.
IME, this is also right, Sapling is good at narrowing down spammers, but again, one detector (who has so many proven false positives) is not proof to accuse someone.I have heard back from @Ultegra134, and he has the same opinion as I have
. Thanks for all his efforts to go through all the details.