ah, well ... I had hoped to not get involved in this topic, but I guess I played a hand in it, a bit. I will, therefore, at risk of censure, put my 2 cents in. Note that I have been in touch with technolion privately about this matter.
I must disagree with some of David's objections, and will offer my own analysis - and why I believe that yukiblue and others are using software.
a. technolion did not use the word 'cheater', so it is unfair to assume that he was specifically making this accusation. However, he was offering a fair warning to the wise, and I agree with the warning. I warned him of this myself, just as I have warned other players here when I felt something was 'off'. This goes back to the days of the advent of the 'new' Twixtbot, since people were losing rating points right and left to a bot that was rated 1500 but was playing at the level of a 2600.
b. While I know and acknowledge that the person using the name 'yukiblue' exists, from what I have observed, my SPECULATION is as follows: I believe that he, or friend of his, have been clever enough computer programmers to create their own Twixt playing software. This is NOT Twixtbot - though, it is interesting that Bony Jordan made the code for Twixtbot openly available - and now, we are seeing the advent of some extremely strong players coming from nowhere. Personally, I find it strange to observe a player rise from the ranks of a 1500 strength to that of a 2300 strength in a mater of weeks.
c. As an observer of playing styles (and I have observed games using Twixt Commentator for many years now), I can recognize playing styles, to a certain extent. The new guys don't play in the style of the current Twixtbot. Instead, it appears that the software (if true) is using old techniques of 'brute force' that the old, weak Twixtbot used. The difference is that now, computer software operates much, MUCH faster. I would guess that new Twixt software would look considerably farther ahead, and could even be 'pruned' a bit to eliminate bad lines of play and analyze good lines better. The same thing happened with Chess software, which I have followed avidly - at first, brute force methods were used, and improved upon to reach ratings over 3000 (like Stockfish) - but then Alpha Zero comes along with the new technique of self teaching, self play of millions of games, and neural nets - and surpasses all the old software considerably. In the case of Twixt, the task of programming would be EASY to an experienced and reasonably skilled programmer - and now, armed with Alpha Zero techniques, Monte Carlo pruning, and extremely fast processing, it is not at all unreasonable to guess that new programs will start popping up. We never really had that problem before, since Twixt is a relatively obscure game - not known like Chess, Checkers, Go, Reversi, etc.
d. WHY would people write software and 'cheat (sic)' their way to the top?? Well - there appears to be some sense of accomplishment and pride (false, as it may be) associated with this achievement. I'm sure Bony Jordan was delighted when his Twixtbot proved to be the Champion at Little Golem (admittedly, players like Maceij and Florian did not participate). Thus, with no money, no trophy, no fame, no accolades whatsoever - programmers can enjoy seeing their software succeed, since it is a personal 'feather in their cap'.
e. I disagree with the statement that playing many games with Twixtbot will improve a player's strength; at least, not that much. Twixtbot's moves are strictly done by statistical analysis of millions of games - and human beings do not and can not have that capability. Again - I'll note that humans have studied the games of Alpha Zero (and Leela Zero) considerably, but the playing strength of the top players in the world has remained about the same - nowhere near the level of the software. The same is true of players who have played many games against Twixtbot - such as spd_iv and Sam Dakota - excellent players in their own right, who have not managed to figure out how to emulate the moves of the bot.
f. My guess is that yukiblue's hundreds of games were tests of his software against Twixtbot, allowing him to analyze the losses and tweak it to get better and better. His software finally managed a win. With enough tries, any good player (Messieurs Bush and Hensel come to mind) with care and determination might, indeed, score a win.
g. We note the disappearance of several top players, some quite recently. It is a FACT that some players absolutely refuse to play against software - the feeling is 'what is the point?' - even I am not all that keen on it. This is disturbing then, since with the Twixt takeover by software (or software assisted play from humans), it will change considerably the feeling and makeup of our little world, here.
OK, so maybe you can blame me - regarding what 'research' technolion has done before his 'accusation'. But to conclude (ah, finally - at last!), the research HAS been done, and this is a free and open forum (until it starts getting censored, like, say, Twitter?). He has a right to express his opinion, as do I. Btw, I have my suspicions regarding several other players here, in that the strength they are playing at does not quite seem legitimate; however, I will not name names in this 'open' forum. I am too nice a feline! Only ONE specific player has been named here, and I think it would be up to him to respond. Sorry about the language barrier - I guess Google Translate would come in handy to change Japanese to English.
(meow)