This is a great idea. I love it! I don't see any problems, but I do have a question. If you draw while using 5.5 or more points, do you get zero points or negative points?
I also have a suggestion: If a player wins, they get half of the points their opponent spent on that game.
@sicilian3 said in #2:
> Good Idea! I hope it is implemented in tournaments one day.
Not in regular FIDE tournaments but in special tournaments that aren't rated.
@ErzyaKS said in #3:
> You can go another way. You play using an engine, but you not only make your move, but also predict your opponent's move.
> For each correct prediction you get 1 point. As a result, the price of your victory can be very significant. Perhaps the phrase 'your victory' is not entirely correct. Let's say that 'your victory together with the engine' over an opponent will have a certain price.
Both ideas can be their own variant.
@dimkadimon said in #4:
> I think this is a brilliant idea and something that should work quite well in practice! Knowing when to use this super-power would be part of the challenge. If used correctly you would find the best move that you would have otherwise missed. If used in a dull position, it won't give you any advantage.
>
> It can also throw you off your game. AI may find a winning sequence of moves, but it is very long. Knowing just one move from that sequence may not be enough and lead you to a more complex position than you would like.
My idea to fix this: After consulting the engine, they can choose not to use the engine's move and only spend one point. They still lose one point because they can just play a waiting move and play the engine move in their next turn.
@NDpatzer Also, especially in losing positions, the engine move is not always the best move.
@Vihaan7387 said in #11:
> This is a great idea. I love it! I don't see any problems, but I do have a question. If you draw while using 5.5 or more points, do you get zero points or negative points?
> I also have a suggestion: If a player wins, they get half of the points their opponent spent on that game.
The question of whether points should be allowed to go negative is a good one and I think this idea needs to be play-tested to find out what makes the most sense. My first thought is to let points go negative in general (including asking for help and losing) so that players won't just keep asking for engine help turn after turn in lost positions.
@NDpatzer said in #16:
> The question of whether points should be allowed to go negative is a good one and I think this idea needs to be play-tested to find out what makes the most sense. My first thought is to let points go negative in general (including asking for help and losing) so that players won't just keep asking for engine help turn after turn in lost positions.
Do you agree with my suggestion about getting half the points your opponent spent if you win?
@Vihaan7387 said in #17:
> Do you agree with my suggestion about getting half the points your opponent spent if you win?
@notenoughqueens made a similar proposal, and it's another thing I think would just need to be tried out. Can't say I agree or don't at present, but the idea of making victories against a consulting opponent more valuable could be good.
About science pursuit orbits and rating orbits.
Performing comes at the price of reduced exploration; according to certain analytical points of view (RL or DRL, in machine learning or well-defined-domain-AI).
It might take more time or more life-spans when erring on the exploration side of that dilemma.
The technical details about how the points are spent and received still need to be figured out imho, probably easier to do after a bit of real-life experimentation too. But I love the idea as a whole, please invite me if you ever want to try that in an online tournament.
Reminds a little bit of the "chess 2.0" mechanism where there is an extra currency (betting points) and at the time of capturing pieces players get to "bet" that currency in an auction to nuke the capturing piece - in that there is a new trade-off between the OTB game and external help to get right as a player.
Some questions
- as other raised, when you "spend" points, does your opponent receive them ? Do they go back to the game pool / opponent only wins them if he wins the game ? Are you capping the number of times anyone asks for the AI best move to "5 times", or is there a balance mechanism ? i.e. if White asks twice, and Black asks twice, are we back to an equal balance, or have both players lost something.
- This also reminds me of a handicap mechanism - assuming players of differing strengths (does this game still have ELO at all ??), if I am 2600 facing a 2800 player, maybe they are hoping to win 10 points, but I am only looking to earn 6-8 points out of this game, so a way to rephrase the setup is "can I beat or draw this stronger player given that I am allowed 1 AI clue". If I were only ranked 2200, maybe I would be happy to draw by using 4 AI hints, that's always better than losing, isn't it ?
- If I spend 1 points on "what is the eval", can I spend 1 more point to get the best move before I make any other decision (getting a rebate as I just spent points), or would it cost me still the full 2 points (seems silly). Would be nice to say e.g. "is white ahead ?" and if no, stop spending, but if yes, then spend the extra to know by how much, and then maybe I can find the winning path by myself.
- Do players get to pick their favorite engine (potentially building a playstyle with it, or putting skill in figuring out the best engine out there suited to the format), or does the tournament supply the same AI to everyone ?