Chess 960. Good or bad for chess as a game.

960 is dumb (some positions are just bad or uninteresting)

What they need to do is take the top 10 starting positions... have GMs vote on them or something. They'll be both equal and complex enough to be interesting. Then use those positions for 1-3 years and repeat the process.

Something like that would be interesting, but 960 is too sloppy IMO. I wouldn't want 960 to become popular.

Could you give an example of a dumb position?

Something like this is a dumb position?

The thing is that all bishops start in the corners. There is no question about how they will be developed, all bishops will be in a fianchetto

I can't find the interview, but Aronian was saying something like in some 960 positions the best moves are obvious and you have no choice. Like if a bishop is on a1 or h1. Something like that. Or the pieces are uncoordinated and you only have one sensible way to develop them.

Also some positions favor white, so they're not fair.

But some positions are like the standard position where there are many good options and it's an equal position. I'm saying using those makes sense, but not the bad ones.

 Thank you for your prompt well thought out  reply.  Which if the 960 starting positions do you think make sense.

one of the 960 positions is the regular position. That means that 960 can be easy to set

I'm not good enough to know the difference haha

But a few times I did put some top GM 960 games into stockfish... and I was surprised that (in at least these few games I looked at) they were playing the #1 choice of the engine even on moves 1, 2, 3, etc. I'm guessing at least one of those positions were one of the not good ones.

The case you make is valid. I had not thought about it until I read your post. Given that part of my income come from doing probability I would imagine that at least one setup was "dumb". In fact using GM to judge position and have them vote I would estimate using the empirical distribution that 2.5% or 0.025 *960 = 24 to 48 position were dumb. Anyway, it is nice to have a discussion with someone who does not become apoplectic.  

My point before I had read your post was:

What I like about 960 is that one has to determine what strategy to employ to develop your piece to ideal squares and of course tactics. For example, if I am playing open positions I know development and tactics are very important (i.e. the Evans Gambit). Where as in closed positions I understand the pawn structure and where to put the pieces is important (i.e. KID ) But, Rios, book Grand master Pawn Structures is very helpful.

However, give the plethora of possible openings in 960 it forces the player to be more creative.

Yeah, it's a good point. Top players struggle with getting winning chances without taking big risks because everyone has done so much opening research.

A counter point to 960 I've made in the past is that you can just as easily diminish the skill element as enhance it though restrictions. At the extremes you have a game where anything is possible, which is arguably the most complex, but also with no rules it has no skill. Rules create that space where strategy is possible. And of course the other extreme is nothing is permissible, so there is also no strategy there.

So I argued that 960 positions is too much. Preparation and research is part of the skill. With too many possibilities you might as well remove the opening phase altogether and just make the players start from random positions where 5 or 10 moves have been played... another way to say it is that the opening position is too complex to figure out OTB. You'd need multiple days, weeks, etc.

So what I'd like to see is something like 10 positions chosen for some amount of time (like a year). Then when two players meet in a tournament, they can each reject 1 or 2 starting positions. Then with the remaining 6-8 positions 1 is selected randomly.

I think this keeps things fresh enough to not suffer from over analysis, but keeps it limited enough that players can prepare and do research on their opponents.

Another way to determine the positions might be to do a stimulation.  For example a colleague of mine pointed out that Fisher d6 bust to the King's Gambit was shown to be correct after 30 million games with engine vs engine. This was a phone conversation so I did not fact check. But thanks for the imput