Can cheating be helpful, and the solution to cheat detection.

Closer to the end I give a solution to cheating, that has never to my knowledge been presented before, and it might solve the problem of effective cheat detection nearly entirely. But first the context:

It is highly possible that a statistically significant percentage of cheaters are using engines during games, because they believe it’s an optimal method for improving understanding of the game.

Whether or not this is true, if people believe it is true, that makes it significant, because it means honest-intentioned people may feel an obligation to cheat. Chess professionals who ordinarily would get no joy from winning a game using an engine, may feel a strong motivation to at least try it out in case it is actually beneficial. It’s something that would require them to actually test it for themselves, because there is no collective understanding, or academic studies to answer the question of whether cheating can be helpful for learning.

Reasons why cheating may be helpful for learning:

• Analysis after games may not be as effective as analysis during games because the motivation isn’t tied to results: We know analyzing games with engines can help people improve, so why wouldn’t analyzing the game while you’re playing it also have that effect? It seems like it may be more beneficial in one sense, because the person may be more engaged with the position/care more about it, and the knowledge from the engine might hit deeper.
• Exposes blind spots: Being able to reference the engine might make them aware of moves they didn’t even consider, and therefore expand their understanding of those types of positions.
• Learning by immersion of engaging with good moves. With learning a new language we know immersion is the best approach: In my experience with computer programming, using an LLM to write code for me actually helps me learn quicker, because it’s able to immerse me in world I would otherwise not have direct access to. I’m already using the concepts before I’ve memorized the definitions, and then I intuit them quicker and deeper. It connects the book learning to the practical learning instantly.
• Learning openings by repetitiously playing them out: Memorizing opening theory is quite tedious, but if you get the opportunity to play some opening enough times by following an opening database, one might learn it quicker through the action of playing it, then by trying to hard memorize it.
• Allows lower level players to connect to higher echelons of chess: They say you are who you hang out with, and being able to hang out with people above your level might be helpful.
• Allows lower level players to connect to higher order abstract concepts: When I watch master games I see interesting abstract principals at play; I am able to understand these higher level principals when I see them, yet I am not able to get my games to those positions where these principals happen, because the “play style” at lower levels doesn’t allow for these positions.

This is compatible by the way with Anders Ericsson’s research, who showed superior performance stems from deliberate practice, not passive exposure. Getting to use an engine a couple times in a game isn’t passive exposure, because you still have to understand and calculate when to use the engine, and using an opening data base as well isn’t passive exposure, because the person will still be navigating the opening book to reach a position they desire.

Why I don't personally cheat

(You can skip this part, I’m just putting it in here to protect myself from the perception that I cheat, because I give reasons why it may be useful: To be clear I have never cheated even once, and I don’t plan on it, here are the reasons why)

1. I have a policy of never lying, so if I cheated, and someone asked me if I cheated, I would be stuck.
2. I play chess only because I enjoy it, so I don’t care about my rating.
3. I don’t really care about getting good at chess, because I don’t see how that would help me in life. Again, I just play chess for fun.

I think some of these traits may be anomalous: If someone thinks it’s okay to be dishonest sometimes, and they want to improve at chess, they may conclude "trying out" cheating is reasonable.

The solution:

First, if we can know somehow that cheating in chess isn’t helpful at all, then we should spread awareness of that.
If we don’t have that knowledge, then the following seems like a really good option:

Cyborg chess:

Create a space on some chess website, where people can use computers to various degrees, for example:

• They can reference the engine a given number of times during the game. (Engine provided by the chess website, just like in a post-game analysis.)
• The eval bar is visible
• They have an opening data base to reference.

Instead of making it a direct mirror so both players are cheating equally, it could be set up so there is a credit system that motivates higher level players to allow lower-levels players to cheat against them. Basically in order for you to be allowed to use computer assistance against higher level players, you must allow lower level players to use use computer assistance against you. Like a currency system.
The benefits are everything I listed above in "Reasons why cheating may be helpful for learning", plus the following:

• People may never put in the time and effort to properly cheat, because this cyborg chess option fulfilled all their desires.
• Amazing training data for anti cheat neural networks: \* Currently when it comes to training neural networks to detect cheating in chess the issue is that there is no sample data to train on. We both don’t have the data for clean games, or dirty games. The clean game data is polluted with an unknown amount of cheating, and the only dirty games we have are when cheating was done poorly enough to be detected by the current systems. In other words if we try to train neural networks using the current data, we will only get results as good as we currently can do. If we train our neural networks from a complete cheating data set, we might get nearly perfect cheat detection results.

The results:

Implementing this “cyborg chess” feature may:

• Bring more people into the game, like having training wheels on a bike.
• Allow people to learn chess more efficiently.
• Decrease the number of people cheating in the regular chess pool, by giving them a legitimate option.
• Actually facilitate the training of neural network that can detect cheating at a level never seen before!!

I think the training data that this will provide will result in ASTONISHINGLY powerful neural net capability to detect cheating. Any anti cheat team would be insane not to consider that. Literally every method for proper cheating could be reflected in cyborg chess and unlimited beautiful training data for every method of cheating could be represented. This needs to be looked into further. It’s possible that this training data will be powerful enough to end online cheating in chess as we know it.

Why this might make chess more fun.

Finally I will just put emphasis on one more feature: Currently if you want to play a game of chess, you will be matched with a person that is at your level: The reason this is the standard paring system is because It’s not fun to play with someone when there is no challenge. When I play with friends who are weaker then me, they don’t like playing positional odds (me being down a piece) because it doesn’t feel like a proper game of chess to them. But if for example I could say “You get to have the opening book, we are still having a proper game of chess, but you can just get out of the opening in safety, while learning theory in the process” I think they would be far more inclined towards these kinds of odds.
Additionally being able to play with more advanced players would just be extremely exciting! Right now that basically isn’t an option, because the games are no fun. If every player had the option to play higher rated opponents, in an equally dramatic and exciting game, and all they had to do was allow lower rated opponents to do it to them, I think we would see a lot more of that; I would personally do it.
I wont go into all the counter arguments here, for the sake of brevity (risk of dependence, gateway drug argument, etc)