Champions Showdown - Chess 9LX

[NickFaulks]

You know nothing - you do have to actually play the variant to understand it and I do so can produce more then one

But you haven't, have you?

[Joey Stewart]

Yes, I played it quite a lot, I wouldnt make such a sweeping statement without knowing something to back it up. I appreciate the intention of fischer chess but too often it does not create an enjoyable novelty position and it sounds worse still that GMs are now trying to analyse it with engines, thus negating both the avoidance of computer theory and the hope of a good game hence rending it worthless.

[Ian Thompson]

OK, I'll bite; you have seen this Nick? http://www.computerchess.org.uk/ccrl/40 ... able_start

The last row in the table seems to be at odds with the average White score obtained in millions of other games:

Code: Select all

RNBQKBNR	634	32.6%	46.6%	0.42 MB

Wouldn't you expect a similar percentage score from this mirrored position?

Code: Select all

RNBKQBNR	492	32.7%	58.4%	0.35 MB

[Matt Mackenzie]

People have been saying that classical chess is "played out" for at least a century - it very much isn't.
(and is highly unlikely to be in any of our lifetimes, either)

That is people referring to chess at the very top level.

I know, but even then it isn't really true - not now or indeed at any point in the previous 100-odd years.

[NickFaulks]

Yes, I played it quite a lot, I wouldnt make such a sweeping statement without knowing something to back it up.

But you're keeping that information to yourself, because it is of proprietary value. Fair enough.

[Geoff Chandler]

I know, but even then it isn't really true - not now or indeed at any point in the previous 100-odd years.

I know, I was agreeing with you, chess won't be played out.
But it is often mentioned due to top level tournaments and the high level of draws
Capablanca feared for chess in the 1920's he was referring to top level chess.

'Capablanca believed that chess would be played out in a few decades
(meaning games between grandmasters would always end in draws)'

https://en.wikipedia.org/wiki/Capablanca_chess

Thankfully he was wrong.

[Geoff Chandler]

Hello Again Matt,

Following on...if anything we, the punters are being chess'd out.

The candidates, the Olympiad and the British all in quick succession.
We are getting swamped with games we will never catch up on as one
'don't miss this' tournament follows another with barely a week in between.

The circus continues with this 960 variant and the Julius Baer Generation Cup.
The only interest of which is will Carlsen play and make a statement.
The chances of perhaps seeing the Carlsen v Niemann game are dodgy
because so many people will log on to see it the site will most likely crash.

[NickFaulks]

OK, I'll bite; you have seen this Nick? http://www.computerchess.org.uk/ccrl/40 ... able_start

Yes, a while ago I looked at this in some detail in connection with FIDE work.

It actually offers only weak evidence that the starting position has any effect at all. The distribution of White scores is very much as you would expect randomly, except at the top end where you find a bunch of 3 sigma outcomes that might be expected to be 2-2.5 sigma. Not much, but enough to suggest that something is going on.

The last row in the table seems to be at odds with the average White score obtained in millions of other games:

Code: Select all

RNBQKBNR	634	32.6%	46.6%	0.42 MB

Yes, that is weird and looks like a bug. Are they somehow giving Black the first move when they don't change the starting position?

Wouldn't you expect a similar percentage score from this mirrored position?

Mirrored positions are interesting. They are distinct because of the castling rule, but how much difference can that make?

Take a look at the best position for White ( 4 sigma ). QBBRKRNN. To my untutored eye, it is possible to believe that White's diagonal based pieces are perfectly placed to launch a blitzkrieg attack. But what about the mirror, NNRKRBBQ? Looks just as good to me, but comes in at just average.

We must be grateful to the computerchess guys for doing this really valuable work, but sample sizes ten or a hundred times as large would be good.

edit : DeepMind could have put this to rest in minutes, but sadly that circus has left town.

[Kevin Thurlow]

One thing I have noticed from playing 960 events is that the better players still tend to score better. At the Fischer Memorial at Reykjavik in 2018, I got 4/4 against lower-rated players and 0/5 against higher-rated players, which admittedly might have been affected by "Swiss Bounce". In correspondence chess, you do play two games against each opponent with the same position, to even out any positions where white might have an advantage. But https://www.iccf.com/event?id=95664 suggests that the higher-rated players more or less do better. (Tie-break is number of wins.)