Come, deep learning models play starcraft, a game with a much thicker fog of war. They win at poker.

I doubt this variant would present a fundamental challenge to DL techniques.

Hill climbing or simulated annealing would work fine, wouldn’t it? You are constantly looking for a goal that maximizes the number of times you can check.

Check != Checkmate. Also you need to find the king before you can do that. And all the other pieces that protect the king. The problem is far more complicated than "throw this algorithm at it".

I recommend playing several games with someone else, and then rethinking your assessment.

> current deep learning tech can’t be used to build a competitive engine.

I’m sorry but I disagree with this point since the process of asking about pawns, knowing about the state of the game (checked, unchecked or checkmate) and knowing if a move is legal can be used with reinforcement learning to accurately model the unknown environment and build a play strategy.

I'm familiar with reinforcement learning and having done some private research in the subject, I disagree with your overall approach. While you might be able to make a basic model you won't be able to reach a global optimum with the vast variation in play you'll have after a dozen moves or so.

Deep learning can't really guarantee to find a global optimum in most nonconvex problems. However, this chess variant is definitely not something that is likely to pose serious issues to modern systems. DeepMind's StarCraft 2 AI does just fine handling vast variation and imperfect information.

> "this chess variant is definitely not something that is likely to pose serious issues to modern systems"

How do you know that? It must be experimented and proven. It's not something that can be handwaved. With machine learning, I would say "not possible until you prove beyond a doubt".

In what way is this variant a more difficult problem than StarCraft 2?

It's a different problem. SC2 is a realtime strategy game, and Kriegspiel is turn-based. The information asymmetry is different in both games, and the goal is different.

In SC2, you can perform realtime monitoring of changes and track them with multiple maneuvers all at once. In kriegspiel you cannot.

Also in SC2, when you see something, you know what it is. You don't know what kind of pieces are where in kriegspiel. Mistaking a queen for a pawn in kriegspiel can be a game-loser.

From a machine learning model’s perspective they are both turn based. In SC2/Dota case one turn is one frame. There is nothing fundamentally different between these and blind chess apart from the possible moves space being vastly smaller

As noted above, the other differences are that you don't know what type of piece you've located (if you've even managed to locate one).

So taking your analogy, and saying an AlphaStar game lasts 5 minutes, at 60 fps, thats 18000 frames. Kriegspiel games average about 50 moves total (or 100 half moves).

So lets tally these up: SC2 has 18000 frames where on visible turf, full knowledge of occupant enemy pieces are available. Kriegspiel has 100 frames where there is no such concept as visible turf of enemy pieces.

Deduction of if and which piece is occupying a square is the only way to have knowledge of the enemy, and the probability for the deduction to be correct drops significantly after each move.

With such a small amount of information available, the problem becomes significantly different.

The key to successful machine learning is available data. There is so little data that the model you are training will collapse or overfit quite quickly and become useless.

I thought DeepMind's Starcraft 2 AI uses maphacks - aka has full view of the entire map at all times.

The first versions used what you might call 'camerahack', it just sees the entire map at once but as the player himself would see (i.e. fog of war included). The latest version has camera control, only seeing one screen at a time like humans.

It doesn't.

On first view, that might make it a really interesting Incomplete Information game to consider for AI research.

Indeed it is! I have a publicity stunt in mind along those lines, when I launch the new design.