🔥 CHESSIVERSE turned me into an AI-bot!... which crushed me! | Chessnut Go | Travel Chess!

🔥 CHESSIVERSE turned me into an AI-bot!... which crushed me! | Chessnut Go | Travel Chess!

Avatar of vitualis
| 8

#chessiverse #bots #chessnutgo #travelchess #vienna 

Click to subscribe to my YouTube channel!


🔥 Coupon code: CHESSNOOB64

It’s been a slightly busy week for me in terms of travel. Last weekend I was at the RACGP annual conference in Brisbane. This week, I attended the Equally Well Conference held in Adelaide, which was excellent.

I didn’t manage to get a game of chess in during the conference but found myself with about an hour free at Adelaide Airport on Friday evening, waiting for my flight back to Sydney. The Qantas Business Lounge was spectacular, and IMHO, much more comfortable than in Sydney and Melbourne!

If you find yourself with the opportunity to use the Qantas Business Lounge at Adelaide Airport, take it as I don’t think you’ll be disappointed! 😙🤌

So, I pulled out my ultraportable Chessnut Go electronic chess board and played my newly released AI-bot on Chessiverse! Man vs machine…

It’s the AI-powered… me!

Game 1: AI-me lures me into the “anti-Vienna”!

I foolishly started with 5|0 blitz and the AI-me crushed me on time, though I feel like I had the “moral victory”… 😅

I had the white pieces in the first game and led with the Vienna Game (1. e4 e5 2. Nc3). My bot responded with the Falkbeer Variation, and of course, I played the Vienna Gambit (2… Nf6 3. f4). The bot paused for a few seconds, and then played (3… d5), the Main Line. Nice.

So, captures-captures (4. fxe5 Nxe4) and I went with the Paulsen Attack (5. Qf3) and AI-me immediately responded with the Bardeleben Variation (5… f5). Oohh, very nice! You see, this is in fact my usual and recommended way of playing AGAINST the Vienna Gambit; to bait the Vienna player into entering the Main Line, with the Bardeleben Variation practically being an “anti-Vienna” at beginner-intermediate levels! I felt very proud of my bot! 🥹

I’ve previously written about the Bardeleben Variation in the Vienna Game Main Line, as an excellent anti-Vienna approach for Black!

However, as an enthusiastic Vienna Game player, I do have a way to play against the Bardeleben Variation, which is the Heyde Variation with the immediate (6. d4). The logic is that we cement in our e5-pawn, which happens to be a passed pawn! Importantly, this anchors some control of the centre for White, even though this isn’t technically the most accurate move. A challenge for White is that Black often has substantial attacking chances in the Vienna Game Main Line, and if we lose control of the centre with a mistake or two, the game can rapidly go pear-shaped.

Against the Bardeleben Variation, I like (6. d4) the Heyde Variation!

And it worked! AI-me immediately traded knights (6… Nxc3? 7. bxc3), and this is a common beginner-intermediate mistake! In fact, from this point onwards, Stockfish rates me as substantially ahead of my bot for the remainder of the game, even though I make a few mistakes, and didn’t see my bot’s blundered bishop on turn 11. For the remainder of the game, I set up a reasonable kingside attack, but not a killer blow. Simply, my bot outplayed me on time, and I flagged on turn 21. Drats! 😂

Game 2: Argh!! My bot beats me on time again!

Rematch! I played Black and this time, AI-me led with the Vienna Game and this time, I attempted to lure them into the Main Line with (2… Nf6). Although my bot’s description states that they play the Vienna Gambit, perhaps they were perturbed facing their fleshy progenitor (!), and instead, played the unusual (3. Bc4), the Stanley Variation.

Against this, I don’t have anything better than dragging it back towards something solid, so (3… Nc6), a “two knights” approach, with the expectation that it’ll pull the game into the Berlin-Vienna Hybrid, or if White abandons the Vienna, a Four Knights Italian / Giuoco Piano structure. The logic is that while Black doesn’t gain an advantage, neither does White.

Indeed, by the end of turn 5, we’d transposed into the usually tame Giuoco Pianissimo, but then, AI-me, as if bored, as if frustrated by the development, struck with a very aggressive double “knight attack”: (6. Ng5) and then (7. Nd5?!). My bot was testing me!

I thought about the rigorous counterattack with (7… Na5) but decided to play my own Romantic knight attack with (7… Ng4!?). A notion (against humans, and maybe humanised-bots): when the opponent makes an unbalanced attack, they can be vulnerable to an attack on their king, both sound and unsound as their defences will typically be relatively thin. With my knight attack, AI-me with the White pieces had an immediate problem of their f2-pawn that had to be resolved, and I was essentially inviting a mistake.

And it worked. 😏

My bot blundered with (8. f3??), perhaps a natural beginner move to “save” the pawn, but not recognising the problem was the f2-square itself! The follow up (8… Nf2!) forked White’s rook and queen, and the evaluation firmly swung to Black according to Stockfish.

However, this wasn’t enough. I’d used almost two-minutes out of my five, while AI-me had used only forty seconds. The middlegame dragged on, and it soon became clear to me that I wouldn’t be able to convert to a win with my remaining time. I aimed for a draw by threefold repetition but simply couldn’t move my pieces fast enough on a physical board! The bot won again despite an unfavourable evaluation of almost [-5]! ARGHH! 😤

Game 3: A win by retreating to 10|5 rapid! 😏

Being determined to record at least one win before I had to board my plane, I decided that 5|0 was perhaps just too ambitious a time control and played the final game as 10|5. I had the white pieces again, and fascinatingly for a bot trained on my games, AI-me played a French Defense (1. e4 e6)! I almost never play the French with Black, but okay, I can do this!

Those of you who follow my channel will know that against the French, I lead with the offbeat and tricky Steiner Variation (2. c4). The idea here is that French players often get to dictate the structure of the game, and as it is in their sphere of experience, they’ll get the advantage. With the Steiner, I find that it often results in games that don’t look like the French, and indeed, we get tactical open lines by the middlegame!

I am impressed by how Chessiverse have programmed the AI-me, though this just might be my cognitive projections of my biases. Against the Steiner, AI-me played the curious (2… Bc5?!), which I could almost imagine is what a chess noob who has an unhealthy attachment to opening attacks (😏) might play in exasperation, should the opponent annoyingly scuttle their initial intention of playing a “normal” French Defense for once! 😆

I responded with (3. d4), putting a third pawn into the centre, threatening the bishop. AI-me then gave a check, and traded away their bishop, resulting in a somewhat weird position at the beginning of turn 5. Black had no development whatsoever other than a pawn on e6, while I had the c-, d-, and e-pawns commanding the centre on the fourth rank, and a slightly developed queen. The interesting thing on analysis is that somehow, this position isn’t actually that bad for Black, with Stockfish giving the evaluation at around [+0.7]!

However, although the AI-me was indeed a bot, it was not based on Stockfish, but had a neural network trained to play like a human noob. From a human perspective, White had a major advantage in development, and this carried into the middlegame. Where I meaningfully castled on turn 8, my bot castled on turn 10 but still had all their pieces on the back rank, except for a knight awkwardly placed on the edge of the board!

 Apparently, my bot is tuned to potentially choke under pressure (I feel personally attacked here!), and that indeed happened on turn 19 with (19… Ne7??), a catastrophic blunder. The knight previously defended the d8 square, and in the attempt to attack my queen, had exposed Black’s back rank! Firstly, (20. Qd8+!) and (20… Rf8) was the only legal move to block check, giving up the defence of the knight, leaving to its clean loss (21. Qxe7). Flustered, AI-me played another one-move attack (21… Rf7??), and this time, blundered [+M2] through their back rank: (22. Rd8+! Rf8 23. Qxf8#), GG! It feels good! Humanity wins! 🤣

* * *

Learn how to play the best chess opening attacks in the Romantic style with my new book, “Become a Chess Assassin!” available now on your local Amazon store!

Buy on your regional Amazon store! US | UK | DE | FR | ES | IT | NL | PL | SE | JP | CA | AU

Hi!  I'm vitualis, the chess noob, and I run the "Adventures of a Chess Noob" YouTube channel and blog.  I'm learning and having fun with chess! 

I restarted playing chess recently after my interest was rekindled by the release of "The Queen's Gambit" on Netflix.  I mostly play 1 or 2 games a day, and am trying to improve (slowly!).  I document some of my games and learning experiences on my blog and YouTube channel from the perspective of a beginner-intermediate player!


Subscribe to my YouTube channel! https://www.youtube.com/@chessnoob64


NEW BOOK November 2024: Become a Chess Assassin! Learn to play the best chess opening attacks. Don't miss out on your copy! Buy on Amazon for only USD $15.99! US | UK | DE | FR | ES | IT | NL | PL | SE | JP | CA | AU


Also, 50+2 Chess Quick Wins: Tactical ideas for exciting chess for beginner players on Amazon for USD $13.99! US | CA | UK | DE | FR | IT | ES | NL | AU 

 

Also check out my Twitch and YouTube channel where I play and stream my backlog of Japanese role playing games (JRPGs)!