Curious about Computer Chess?

Howdy! It’s the Chess Friends!

#HelloWorld

Recently, we came across an online arena where top chess engines clash against each other, 24/7, nonstop! That prompted our favorite question: WHY? In this blog post, we unravel the mysteries of computer chess and share how we use engines in our own chess studies.

3… 2… 1… GO!

Good Game, Robot!

Check this out, friends: the top two chess engines in the world are Stockfish and Leela Chess Zero (Lc0), both over 3800 ELO (Fun fact: “ELO” is another way of saying “rating”, after chess rating inventor Professor Arpad Elo!) For comparison, the highest human ELO is around 2900. That means engines can calculate deeper than any human, by far.

These engines work using neural networks, which are modeled like our brains, using machine learning AI (artificial intelligence) and layered “neurons” that get inputs, apply some operation to those inputs, then generate outputs. Besides daily arena, there is a Top Chess Engine Championship (TCEC) among the best of the best engines! In the Season 27 Superfinal, Stockfish beat Leela 58-42. 

Stockfish uses a “brute-force” approach. It can evaluate millions of positions per second, with its efficiently updatable neural network (NNUE). It even works on your phone or laptop, which may not have that much space or speed. Stockfish is considered a bit stronger and is definitely the most widely used engine, but Leela has beaten it, particularly using supercomputers, which have much more computational power. Leela’s neural network is based on self-play, and it has played over 2.5 billion games! For comparison, GM Hikaru Nakamura has played over 65,000 chess games online, while The Man Benji has played over 2,000, The Myth Sarang 5,000, and The Legend Vivi 500.

Leela learns and improves every game utilizing Monte Carlo Tree Search (MCTS) and probability to evaluate and analyze positions. Curiously, the way Stockfish and Leela evaluate chess positions are not the same. Stockfish tends to play moves that appear very strange to the human eye, due to how deep and how quickly it calculates. Leela’s evaluations tend to be closer to equality, and plays more human-like moves, so many feel Leela is better in advancing both the science and art of chess learning.

And here is what our “why” comes down to, friends: Chess is a science and an art. Watching engines play one another inspires new creative ideas and plans that we, humans, have never tried before! Advances in science inspire creativity in art and creativity in art motivates advances in science. It’s a beautiful cycle of learning.

How do we, The Chess Friends, use chess engines to grow in chess?

The Man Benji

Here are a few ways I use chess engines:

1) I review my games with an engine to look for ideas that I missed. I also hunt for brilliancies! A brilliant move is usually sacrificing material for a big attack or a super clever tactic! Very few moves earn this prestigious mark (!!), and it’s always fun to see your moves being awarded such distinction. Only an engine and your best chess friends can tell you how many brilliant moves you had! As I mentioned in the “Would you rather” blog (read it here), it’s more fun winning using brilliancy than higher accuracy!

2) Before a tournament, I sometimes use an engine to quickly review opening lines. It’s better to be safe than sorry!

3) I use an engine for endgame training. What I do is I simply challenge the engine, Stockfish on the highest setting on Lichess.org, to play a pre-set winning (or drawing) position against me, so I can gain confidence in converting wins (or holding draws) against someone who never gives up!

4) When I am tired and want to play chess without any pressure, I turn to a chess bot to wind down with an easy game. You know those games, like promoting to six knights against Martin on Chess.com or listening to Oscar yapping on Duolingo. Don’t judge me, they are funny!

The Myth Sarang

I think it's good to analyze popular positions with engines to come up with interesting, unconventional plans. Engines are a good way to learn common lines in openings and various deviations and how to play them. That said, it's very important to play lots of games without using engines to practice these plans and make sure you really understand how to apply them. There's a big difference between theoretical knowledge and practical knowledge. I also like looking at positions and trying to figure out whether it is winning, drawn, or lost. Engines can be helpful to check thinking in these situations and tablebases can verify your thinking in endgames with fewer pieces. 

The Legend Vivi

I honestly don't use engines much. I think a lot of people use them too much and because of the evaluations that are not human, they think a move is "terrible" or that a line "can't be played" because of some weird combo of moves that no human could find. I like to play with humans, learn from my coaches, and be creative in my chess. Sometimes I come up with interesting plans and put them in an engine to see what defenses it comes up, and it can help me be even more creative in those scenarios. I honestly much prefer working with my coaches, great humans, rather than engines. 

There you have it: three chess friends, three unique opinions. Do what works for you!

Before we sign off, here are a few knee-slappers for the road:

What does a baby chess engine call its computer father?

Da-ta!

***

What does the baby chess engine have for breakfast?

Goo-goo-bytes!

***

What is a baby chess engine’s favorite opening?

Bongcloud!

***

What does the baby chess engine say when it’s sleepy?

It’s time for an App!

Three Cheers, Fellow Future Master Chess Friends!

The Man Benji, The Myth Sarang, The Legend Vivi