May 11, 1997, was a turning point in the history of mind sports. On that day, reigning world chess champion Garry Kasparov lost a match to the computer program Deep Blue. The creator of civilization proved weaker than his own creation. And although many, including Kasparov himself, immediately challenged the validity of this result, it soon became clear: a new era had begun.
Several years passed, and the rapid advances in hardware and software finally dotted the i's, forever changing good old chess and reducing grandmasters to the role of trained monkeys, diligently memorizing computer-generated lines in order to outshine another similar monkey at the board. A peculiar epilogue to this process can be seen in Elon Musk's mocking remark, who, after a dispute with the former champion, declared that Kasparov plays chess almost as well as his iPhone.
Go held out longer. The complex strategy of the Eastern game proved too much for programs built on classical principles. However, the advent of neural networks breached this bastion as well. On March 15, 2016, AlphaGo, a program created by a division of Google, convincingly defeated ninth-dan Lee Sedol. Over a glass of sake. Without clinking glasses.
Dots remained practically the last corner free from computer dictatorship. Not because it was unique, of course, but simply because no one was particularly interested. Experiments with artificial intelligence began almost as soon as the first game clients appeared (the benchmark example being Pavel Torgashov's 2002 article ), but this was all done at a purely amateur level and had no serious prospects.
The first more or less intelligent Dots bot appeared in the Sport Dots app (I think it was in 2018). The developer decided to experiment with neural networks, which were gaining popularity at the time, and launched it on Anton Gorodetsky's server. The program went through several training cycles and achieved a fairly decent level of performance (with some easily detectable vulnerabilities, of course). The project's demise put an end to the experiment.
Amalia is also worth mentioning—the brainchild of Bartek Dyda, who occasionally released his little bot for fun on Zagram. But it was designed for Polish rules and only played well (really well) on a very small board, instantly becoming lost on a larger one (possibly due to a simple lack of hardware). However, I haven't seen it on the server lately and don't know if Bartek is still developing it.
However, none of this posed a threat to the champions. The Dots players dug in their little clearing and indulged in pure creativity, looking down on their colleagues groaning under the iron heel. Personally, I admit, this state of affairs suited me just fine. Yes, it was obvious that Dots was fundamentally no different from other intellectual games, and that creating a "terminator" was just a matter of time and developer skill. But the lack of a strong bot meant they didn't have to worry about the fairness of online tournament results and lent Dots a certain romantic aura (the last bastion of humanity!), so the desire to delay the inevitable prevailed over curiosity. Alas, progress is inexorable.
Ivan Kochurkin, aka KvanTTT, had been working on Dots AI for a long time (for example, back in 2011, he and Evgeny Kurnevsky, aka keij, created a bot for the iconic Points XT platform). But his real breakthrough came when he tried to adapt KataGo, one of the best neural network Go programs, to Dots (a perfectly logical decision, given their similar game mechanics). In late 2025, teaming up with Nikita Shokarev (who, as it turns out, had long dreamed of creating a "champion killer"), he perfected his idea, introducing KataGoDots , an engine for creating Dots bots, to the public.
The progress, it must be said, is impressive. Work continues, but already this bot is practically invulnerable in close combat and only gets lost in large spaces (plus, there are still some artifacts, such as a lack of understanding of ladders). If you understand these weaknesses, it's not difficult to beat it (though the games tend to be artificial and uninteresting), so full-fledged matches against a human are out of the question for now. However, it can already be used quite productively for developing tempo sense and practicing forced combinations. More is to come.
As much as it's a shame for our fellow human players, we can't remain on the sidelines of progress. Artificial intelligence has long been firmly established on chess and Go platforms, where it's not only a reliable competitor but also a powerful analytical tool (a teacher, essentially). Now, it seems, it's Dots's turn to take hold. Fortunately, Ivan and Nikita aren't hiding their know-how and are willing to collaborate with any platform.
С недавних пор свой маленький терминатор завёлся и на BBS. The bot, based on the KataGoDots engine, goes by the pseudonym Tomas de Torquemada and is ready to test your Dots prowess at any time of day or night. You can summon him from the online players list on the app's main page. He'll play with red and according to your settings (for live players, as a reminder, it's the opposite). Tomas knows how to resign in desperate situations, doesn't accept requests for backtracking, pauses, or draws, but will agree to abort games. No rating is awarded for playing with the bot.
Now that it's in the cards, let's think about how to use it for good (and how to deal with the inevitable accompanying problems, of course). As a last resort, we always have the option of paper and pens. Silicon monsters certainly won't be getting there anytime soon.
Maybe.
