Did Chess.com Upgrade the Bots???

Fwiw Joel definitely seems stronger.  I have drawn with him 5 times since July 30th.  Unless I'm at my very best he can't be beaten. I'd say his real rating is around 1400-1500.  Still not a true 1600 imo, but clearly stronger than the old Joel who seemed to be at the 1300-1400 level.

I've been playing Nelson a lot, and have had the "usual" difficulties beating it over a year or so, and hundreds of games, i.e. losing most of the time. Now, after a few days hiatus, defeating it is suddenly ridiculously easy: it blunders queens like no tomorrow. i can effortessly win almost all of the games. I wish, but doubt that I just suddenly woke up twice as smart yesterday.

I played all the easy and most intermediate bots a few weeks ago.  Just yesterday I started over. All the easy bots were far easier than I remembered from just a couple of weeks ago. As you said, they all make hung blunders with their queens instead of a more random mix of mistakes and blunders with all pieces. The intermediates also blunder their queens more than a similarly rated person would. Tl;dr I concur that they are dumber than before and in a bad way.

Well, I have been playing old, dedicated chess computers and their gameplay is much more 'normal'. If you're a beginner and giving the low purchase price of such machines they represent far more value that a chess.com subscription!

If you're a beginner and want something more modern and available - for Xmas! - then I would recommend the Millenium Karpov Chess School computer - which can still be purchased new from Amazon UK, beware that the instructive booklet comes in two versions, German and English.  It's a neat little (child sized) computer with a maximum playing strength of 1000 to 1200 imo (quoted as 1500+), which comes with a booklet of 100+ learning positions all build in.

Bah just checked looks like this model has been replaced with the similar looking Europe Chess Champion - no booklet etc but 7 additional games. Not tried.

If you can afford it the Millenium Chess Genius is one mean machine/toy - 1700+ , but double the price.

However, note that on ebay you can get used older 1990s machines for the price of a modern pizza, most of which are 1000 - for two pizzas you can nab a1700+ elo one!

PS: I have bought half a dozen of these oldie chess computers - all working - from ebay in the last few months. - quite unbelievable there appears to be no market for them - they all play 'normal' chess of various strengths. I am even thinking of give-up on chess.com as my 'babies' have more personality than the on-line bots. The very least chess.com could do is give people talking Bots - I mean how difficult would this be?

i kinda thought the opposite as some. Against bots, i can beat a much higher rating than I can online players, so i thought the bots were a little weak compared to their ratings.

I find myself sometimes beating chess.com bots rated 400 points higher than me, where they hang pieces and play passively. That just doesn't happen against humans.

For real, they got alot better. I always beat Sven like he is nothing but since yesterday he won with me once,, later I beat him twice again but it was no easy ride for me. Yeah I'm a chess noob, but bots got much better than they were for example 2 weeks ago.  I've also defated Nelson already but now he just destroys me. Even Emir got better, yes he blunders alot but still he is playing much more solid game then he used to.  So unless my skill just dropped significanlty in last few days, they must have upgrated them. 

I took a bit of a chess vacation (work and all). Now I am back and the 1500 CPU bot that was good fair competition for me is crushing me. That brought me here, so not my imagination. Like people say, it spots mistakes and takes advantage much more aggressively in ways I can’t anticipate. I knew the 1500 bot was not really 1500 human quality, but it would have been nice to know about the change before losing 18 out of 20 or so recently. On the other hand, I know know some of my moves were not so great after all. I guess I’ll try 1300 next and eventually find a new partner. I figure it is just chess inflation…

If you paid for a subscription they make the bots a tad easy so you dont end membership do to losses

Oh, come on!

@Boliviandefense, could you please show me some saved games? Feel free to send me a private message, or post the game links here.

@Babyagun, just wondering if there have been any updates to the 1400 bots recently (in the last few weeks)?   I noticed when playing Ali, Mateo and Boxbox over the past few weeks I'm winning pretty much every game whereas beforehand I'd maybe lose 40%. (or more like 50/50 with Mateo).  I'd love to believe I've simply improved my game but the stats seem too good to be true!

@Bradleygotch, you improved your skills. The bots haven't changed. However, there have been some slight changes recently on Android that shouldn't affect 1400 bots, but if you're playing on an Android smartphone, please let me know.

@BabYagun that has made my week, thank you! 15 consecutive wins does seem too good to be true but far be it from me to look a gift horse in the mouth! I am indeed using an android smartphone - interested to know what those changes were ?

Bradleygotch, I cannot reveal more details. Thank you for confirming that you play on Android. This is interesting. I will take a look at the most recent games of those bots.

@BabYagun, no problem, but do let me know if you figure out it really was too good to be true !

Glad this thread is here! Confirmed that I'm just getting worse practicing every day.

@BabYagun Thank you for engaging with us on this matter, your comments have been insightful and given me much to ponder.  I absolutely appreciate that you are tweaking the bots to make them more accurate to their playing levels, thank you for your dedication.

Let me ask you a question:  Do you think it possible to post when the bots are tweaked?  I don't mean posting updates like "Enhanced Emir by increasing X" or "Changed Emir by lessening Y".  I would be happy with just "Enhanced Emir on X date".

The reason I ask is that ever since I've been reading this thread, I find that I've been second-guessing myself when I play the bots.  Last week, I played Emir and handily beat him five times in a row, no sweat.  But yesterday, I lost to him twice and barely squeaked out a victory in a third game.  Of course, I can't help but wonder, "Was Emir enhanced?  Or am I playing horribly?"  If I knew for sure that Emir wasn't tweaked, then that would answer that question.  Just a suggestion...

@redapplesonly, thank you for your kind words. This is not me tweaking the bots, this is a team work and even a result of work of a few teams.

The bots are based on the Dragon (Komodo) chess engine. The goal of the Dragon team is to make the engine stronger, stronger and even stronger with every update to compete with Top Chess Engines and make Dragon the strongest chess engine in the world. The strongest engine should play like... well... an engine, not a person.

This is contrary to the goals of the guys involved in setting up bots. These annoying guys undermine and weaken the engine, make it do silly mistakes and say weird phrases like: "Meow! I am become Mittens, the destroyer of kings. Hehehe. I am inevitable. I am forever. Meow. Hehehehe." When the Dragon developers read these phrases, they probably think: "What are the guys from the bot team smoking?!", when they see Levy Rozman's video where the streamer wins against Martin-bot despite Martin having 30 queens ... they highly likely scream and cry. They definitely cry. Maybe they are crying right now.

However, they make the Dragon (Komodo) flexible. It is tricky to make bots playing with similar strength on web, Android and iOS. It is hard to make them playing more human-ish. It's hard to get them to play at the same level in the opening, middlegame and endgame.

I don't promise you to post news like "We changed the ELO 800-1200 bots based on feedback. Now they make a little more mistakes in the middlegame." or "Adaptive bots adapted too quickly due to a bug. Now the bug is fixed." or "Android devices now use the latest version of Komodo." here. But I can promise you to take your feedback and pass it on to the team members.

All I know for sure is Mittens (ELO: 1) plays WAY above its rating.