Bots Are Funny With Opening Theory

Probably around a C, but enough to be taken somewhat seriously - by no means is this an opening worth worrying about though; other choices seem much more dangerous in my opinion.

The Slav Defense! Cool! That was like 10 moves of slav theory 

about c to e i guess

Glad you thought it was cool, but actually ALL 15 moves I listed was opening theory I had memorized. There are some key variations and sidelines to know along the way though. How does an 1100 bot follow this deeply into book?

At the last move, ...O-O was the book move chosen in the game, but ...O-O-O is also theory and even ...b5 is "book" (although less common). This is why learning openings is more about learning ideas and patterns than just memorizing lines. Many sidelines and variations are too much to remember so the chess player must understand the position instead

Yea I agree but from what i was told High Level bot like Masters or Such Do the best move predicting the players option so they lay out traps or such if the play takes a long time looking at every move (Time depending on Knowledge of the game) the player has to lay out moved preventing these "Traps" Or moves or i think they will loose, In my opinion. Bots are very weird in what they do lol but depending on the level of the bot (High=Every move has a meaning sometimes) and (Low=Every move has some what of a meaning behind it but the player can easily sometimes see it) What i think depending where your pieces are or openings the bot can see it and play their best according to their level. 

Bots do play differently than people (why I recommend playing people as default instead of bots), but what you were told isn't exactly 100% true. It is sort of like half the story. 

Bots mainly play weird to us because they can't "think" in human ways we do - at least not yet anyway. Traditionally, top human players like Grandmasters, tend to think in terms of patterns, structures, pieces activity, initiative, counterplay and elements like these difficult to assess concrete value towards. 

Conversely, bots/engines like Stockfish assign values to EVERYTHING internally, even things like how much a certain file is worth or value of a passed pawn based on surrounding factors like how far advanced it is. 

Another big difference in mindset is that humans just calculate a bit and then draw on intuition, but computers don't have intuition, so they calculate WAY more. Exciting are newer engines like AlphaZero which play a little more "human like" with exchange sacrifices and positional play.

Long post already but now to how bots are "weird" at different levels. It is because bots calculate a ton and then assign values to each line/move. They can't calculate tired, drunk, or hungry like human players do lol so a 3000 bot will consistently play 3000 level etc. whereas human players might have off days and other times they are in the flow. 

Since bots can't "play tired" etc. (or play easier), they can't play a lower level like 1000 by calculating etc. A 3000 engine will always play 3000 level and that is it. 

How chess.com (and other computers) can set "levels" to bots is by having a stronger bot occasionally play an inferior move; the more frequently inferior and more severe are for "lower levels." This is the reason why an 1100 Laura-bot can play GM-level opening theory (or at least expert level depth). The 1100 bot is operating off of opening programming/calculation of a 3000-ish engine xD 

Yeah, computers play funny in human eyes. They play super accurately then throw on random moves, unless highest level just that we can't always exploit it

Makes a lot more sense thanks xD

This has been a complaint of mine for a LONG time: bots know WAY too much theory, even against rare openings. Even a 1000 bot will know 4 or 5 moves deep of opening theory against the Dutch, whereas the VAST majority of humans at 1000 know zero.

Instead of giving low-rated bots an opening book, this is what needs to happen: chess.com needs to look at the games of actual players (they have a huge database of player games so they can do it). They need to then in the opening have the bots play like human based on a probability distribution. If 1000 rated players play e5 against e4 50% of the time, then bots should play it 50% of the time.

When black plays the Dutch, the most common response by masters is g3, and this is 2.5x more often than the second move Nf3. Lower rated players, because they don't know theory, play c4 about 3x more than any other move. g3 is the **8th** most common reply. Why? They don't know theory and treat it as a weird version of the Queen's Gambit, which is the only thing they know.

But the 1000 bots? They almost always play g3. This is just wrong and needs to be fixed.

Does that mean all the lower rated bots are smurfs? 

lol not exactly smurfs; more like a hustler letting you win by playing randomly bad moves at certain points in the game xD

Play the jimmy bot you will get a laugh out of it lol

The sheer ridiculousness of this move... I guess this Jimmy fellow thought hanging his queen and rook wasn't enough so he decided to hang his bishop too for good measure.

The immortal gluttonous knight