Suggestion: Leftover kings, etc.

Currently if a player resigns or flags his king stays on the board as a +20 prize. Other players compete to capture it.

My suggestion is: The king (only the king, not the gray army) should be controlled by simple AI. And this AI should move the king in random directions. The moves should be legal (the king should not step under a check) and that king should not capture any alive pieces.

This way that king becomes more difficult target. Not just a pile of gold laying on some square.

I understand that the developers need to spend some time to make the AI. But this is quite a simple task:

1. Add 8 squares around the king to the list.

2. Remove the squares with any alive pieces from that list.

3. If some squares are attacked by some alive piece of any player - remove them too.

4. If 1 or more squares left in the list then select one of them randomly and move the king to that square.

This is the algorithm.

come on... it is chess, not casino where lucky players win

@iplychss, having a resigned player being controlled by an AI in some way might actually decrease the amount of luck required to win. Right now when a player resign/timeout the player that is lucky enough to be in a good position to capture the king gets an advantage.

If something like this would be implemented I would probably allow the king to capture pieces and always capture the most valuable piece as possible when it can (when that does not put the king in check). Otherwise getting your queen unprotected close to the king might make things still too easy.

While I do see some advantages with this idea, I am not sure if it is good in practice. It might be more frustrating to play against an AI than the current implementation. The movement of the AI can change the outcome of the game. The randomness when the king moves will happen more times than the "seemingly randomness" timeout/resignation. This said I do not actually mind the current implementation.

player that is "lucky enough" to be in a good position is usually the one who spent resources to force an opponent to resign or timeout

Martin0, you do understand me. I also prefer more simple solution, but could not find one by reading this forum. And this is a real problem currently. I did not count it precisely, but have a feeling that near 40% of games currently has at least 1 player resigning too early at a random moment. They can fight, but click "Resign". And we can not remove the "Resign" button at all. Because the players will just close their browser and force other 3 players to wait for his "flag" to fall down.

Capturing a leftover king now is a matter of luck. So we can just throw a dice and look who gets those 20 points. 

Points related solutions are not ideal too. For example, we can split 20 to 3 equal parts (7 rounded) and give +7 to each remaining player (or +10 if there are 2 players left), but it is also not fair, because one (or 2) of them attacked the resigned player when other(s) did nothing. And sometimes no one attacked the resigned player, he just abandoned the game for no reason. The remaining players do not deserve those points also.

So, I thought about the AI. Of course it is also not a perfect solution. But a player catching a moving target deserves those points. And if there is a chance that chess.com developers will do the AI some day... then we can suggest some improvements. Like replacing the random moves with some logic. Like Martin0's "eat most valuable piece" or so.

While this is sometimes true it is not always the case. And when someone disconnects or accidentally looses on time there were no player which forced that to happen. 

iplychss, unfortunately "usually" in your phrase should be replaced with "sometimes". Especially if the players' rating is less than 1300.

I think such early resignation do not happen at higher level though

To be more clear, I like the idea of having an AI control a resigned player. However I am not sure if it really solves the problem if there is an AI that only controls the king. Making a "bad" fix might not solve anything. Ideally I would want an AI that controls all the pieces, but that would require a lot of development time. Maybe the current solution can stay until the developers have more time for a more ambitious solution, such as creating an AI that can control all the pieces.

an AI that controls all the pieces already exists, it was possible to play 4-players vs AIs on hellochess.com

and I agree if an AI has to control a player, it can't control only the king, but has to control all the pieces

still, not sure this would help fix the problem

If there already is an AI, then I think it is worth experimenting with and see if it has potential to be a good solution. I assume the AI should be unable to get points, but other players can gain points by capturing pieces owned by the AI. The AI could also auto resign (like the resign in the current implementation) if there is only 1 player left that is not an AI remaining. Of course it is important that 2 AI's resigning in that way gives the final player 40 points.

I still see problems with having an AI in that way. Assume there are two players remaining and one AI. The AI is completely dominating in the position and the players only tries to not be mated by the AI for as long as possible. Then there might be some luck to which player survives to the end and get 20 points after the checkmate since the AI would resign.

I don't like the idea of an AI playing at all. The only option worth considering would be an AI that always moves the king closer to the middle, unless obstructed, but even that feels wrong.

In practice, I don't think this is a big issue at all actually, I disagree with the premise that mostly luck is involved. From what I've seen, if a player is in better position to capture a king, it's because that player plays more offensively. One part where luck DOES play a role, is that it very heavily affects the opposite player. Not only is he way less likely to capture the king, but with the opposite player gone, his chances of winning decrease drastically.

I don't see how an AI would fix that problem unless it's a strong AI (in which case, I'm totally against it anyway). In many games, I either ignore a player's threats or play more aggressively than usual, because of perceived strength. If that player resigns and the AI takes over, it will be at a different level (stronger or weaker). That might greatly affect my ability to continue the game.

Gilberreke, I am glad you admit that the luck does play a role. And if, starting from now, you will pay more attention to this matter (reviewing early resignation cases) you'll see that in most cases luck plays the main role.

I think the argument that an AI would change the strength of a player is very valid. I still think it is very unclear if having an AI take over a player would be an improvement or not. There are merits to both and I would need to test before having a clear opinion of what I think is better. Like I said earlier though, I don't think the current solution is necessarily bad, just that it has downsides with it.

That is why I propose that AI should control only the king, not the full army. This way the AI can not play stronger than the player and won't change the balance.

I had thought of a similar solution before. Simply disallow resignation and if a player leaves add an AI that would control the players remaining army. The AI would be very weak but still good enough to not simply give away pieces.

I agree that the AI should be weak. I'm slightly worried that just keeping its pieces safe and capturing unprotected pieces might give 1100-rated players (and weaker) some trouble. You have to take some consideration of the weakest players.

@BabYagun If you look through the forums, you'll see that the previous post about this topic is mine, so I have been paying attention to this for a while now, I still disagree it's mainly luck, except for the opposite player.

Game design is my main hobby, and one of the biggest biases is based around luck. Humans are very bad at natural statistics, so they will assume everything is either nothing but luck (poker is all about luck, you need good cards) or luck is something that changes over time (I'm having a bad stroke of luck, as I threw a lot of ones today). Neither of these is true. Professional poker players consistently win against amateurs, so it can't be luck. Statistically speaking, throwing the same number multiple times in a row is not that rare, but we think it is.

The truth of the matter is that from any board state, you can usually grab the queen of another player in two or three moves. If you look closely, you can predict the shortest path the other player will take and contest those squares. That is, unless someone was already putting pressure on that king, in which case, it's only fair he gets to grab it first.

So, I agree that within a single game, it's unfair to the opposite player when someone drops out, but I disagree it's based on luck. Luck means high variability, I don't see many variations favoring a single player, it mostly plays out the same. The two side players usually get there within 3 moves and probably like 75% of the time, they get there exactly at the same time, contesting the capture.

Now, a big issue, to me, is the rating system. ELO is measured over a rating period of one game. Better weighted systems like Glicko recommend 5 game rating periods and Glicko 2 even recommends 10 to 15 game rating periods. In the current version, the rating period is measured over 1/3 of a game (this is to say, for a single game, it rates you three times separately against each player). The result is that ratings will go up and down too quickly, imo, making a single event, like a player dropping out, way too important for your rating, which feels unfair. Unfair does not mean luck however.

Sorry for the long post, I just find these topics endlessly fascinating

We will be moving to Glicko in the near future.

I know this subject is about so much more than my single comment. But this game has a level in that you need to decide if it's worth your time/development to put out knights or bishops in a 1 move to the other kings set up on the chance they will drop the game. I see it as a part of the game and a choice to make. I cant think of maybe 1 time where I could not have developed a piece to get to all dead kings and either take the queen that beat me there or take the king. That said most times I don't develop to that extent early since it makes  a mess of the rest of my game. But its still a choice which means luck has very little to do with grabbing a dead king (I will give you the one time you can't grab it or a high value piece is when someone has done all the work and is sitting right on top of it before they resign. 

So if people look at this as an added layer to the game I'm not so sure its a problem? *personally I could go either way just food for thought.