Idea For Better Chess With Today's Technology

[towforce]

Inspired by my own post in another thread (link), here's a plan for a better chess program with what we have today. First, the background (copied from my post in the other thread):

1. Having too little knowledge (implying a smaller NN) is fatal. We know that hand-coded evaluations are uncompetitive, because they lack important chess knowledge - even though they can get ridiculously high NPS (nodes per second) counts

2. Having too few NPS per second is fatal. Even the best NNs clearly have some missing knowledge, because to get all that knowledge, they have to be bigger, which leads to lower NPS. These big NNs lose to smaller, nimbler NNs that can generate a higher NPS count. Therefore we can say that at the current time, even the biggest NNs lack essential knowledge that only a deep search (hence high NPS) can provide

So: both high NPS (deep searching) and large NNs miss essential aspects of chess. Here's my remedy: do both!

1. On one set of threads, have a simple HCE (hand-coded evaluation) searching as deeply as it can

2. On another set of threads, have a large NN doing a shallow, but highly intelligent search

3. From both sets of threads, have a consolidated list of moves to avoid

4. Choose the favourite move from step 2 (the NN has better positional knowledge than the HCE, so it is ultimately the better entity to make the final choice) which is not on the list of moves to avoid in step 3

In selecting a chess move, the most important thing is to avoid a mistake, and using both a large NN and a deep search is probably the best way to avoid a mistake with today's technology - while still using the wisdom of the large NN to choose the best move that probably isn't a mistake.