Discussion about R16,P17,M16

At position R16,P17,M16, @mark5000 started the conversation:

Should we demote any of these? O16, P15, and R13 may be suspect, though AI doesn’t find a big difference.

( FYI @shinuito )

@aesalon said:

Maybe. Almost all of them are already demoted because they aren’t current. I would love to see some user insights about this kind of question.

( FYI @shinuito )

I do think some should be demoted further but we ought to come up with some criteria instead of winging it.

In some sense, too many Ideal choices does seem like an issue. (it might not be clear enough that there is a distinction between Ideal moves. We have Current vs non-current and variation ordering)

However, we should avoid being too quick to place judgement. I think we want to be helpful but not too judgy. AI are very confident that Q15 is the best move, where do we draw the lines?

My feeling on this position is that Q15, P14, K17, R13, and R12 are played more than 5% of the time by professionals(according to ps.waltheri and a modern filter) so I would at least preserve those as Ideal. I also like O16 as a suggested ‘Basic’ move and it was the most popular move historically but this isn’t all that objective.

I agree that Q15, P14, K17, and R13 should stay Ideal.

I think R12 should stay Good. The modern filter on ps.waltheri captures games played since 2000. The first 16 of these years were pre-AI. After AI, professionls value the sides a lot less. You’ll see usage drop off dramatically if you apply the “new” filter instead of modern.

While that is true, it isn’t very consistent with a lot of previous work we have both done. It’s also what the ‘Current’(it should be more visible somehow) tag indicates. A current node almost always passes the AI win % sniff-test whereas Ideal non-current may not. It was satisfactory to professional players for years, after all.

I could change my opinion if you can observe or have a source saying when R12 is a situational move as I think that’s a large indicator of how we have been using the Good label.

edit: Small sample too. It’s like a 20ish game difference between IDEAL and GOOD move? It’s a fine line to draw.