The longest CCRL game ever....

[reflectionofpower]

A good illustration of why none of the top programs test with repeating time controls, only with increment. There are better ways to spend one's time (or your computer's time) than playing out drawn endings forever at a repeating time control. Repeating time controls are pretty much obsolete even for humans, more so for engines. You can get a noticeably higher quality game in the same average time with increments. But I'm not complaining, Komodo seems to top the rating lists either way. I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

BUT Bobby has the patent on it : http://www.google.com/patents/US4884255

He had the patent on that particular clock, but the Micromate clock was already sold to the public in 1980, nine years earlier, with instructions on how to set it for "Accumulation" (same as incrrement) and crediting me as the inventor. I suppose the Micromate manufacturer could challenge the patent, but the clock was no longer in production by 1989 so it would not pay for him to do so.

I was just showing that Bobby has the patent. I realized that you had thought of this concept earlier.

[lkaufman]

A good illustration of why none of the top programs test with repeating time controls, only with increment. There are better ways to spend one's time (or your computer's time) than playing out drawn endings forever at a repeating time control. Repeating time controls are pretty much obsolete even for humans, more so for engines. You can get a noticeably higher quality game in the same average time with increments. But I'm not complaining, Komodo seems to top the rating lists either way. I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

BUT Bobby has the patent on it : http://www.google.com/patents/US4884255

He had the patent on that particular clock, but the Micromate clock was already sold to the public in 1980, nine years earlier, with instructions on how to set it for "Accumulation" (same as incrrement) and crediting me as the inventor. I suppose the Micromate manufacturer could challenge the patent, but the clock was no longer in production by 1989 so it would not pay for him to do so.

I was just showing that Bobby has the patent. I realized that you had thought of this concept earlier.

I'm not a lawyer, but I believe the Fischer patent would be ruled invalid if challenged on the grounds that a clock was already on the market years earlier with the idea.

[reflectionofpower]

A good illustration of why none of the top programs test with repeating time controls, only with increment. There are better ways to spend one's time (or your computer's time) than playing out drawn endings forever at a repeating time control. Repeating time controls are pretty much obsolete even for humans, more so for engines. You can get a noticeably higher quality game in the same average time with increments. But I'm not complaining, Komodo seems to top the rating lists either way. I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

BUT Bobby has the patent on it : http://www.google.com/patents/US4884255

He had the patent on that particular clock, but the Micromate clock was already sold to the public in 1980, nine years earlier, with instructions on how to set it for "Accumulation" (same as incrrement) and crediting me as the inventor. I suppose the Micromate manufacturer could challenge the patent, but the clock was no longer in production by 1989 so it would not pay for him to do so.

I was just showing that Bobby has the patent. I realized that you had thought of this concept earlier.

I'm not a lawyer, but I believe the Fischer patent would be ruled invalid if challenged on the grounds that a clock was already on the market years earlier with the idea.

I don't disagree with you there.

[Norm Pollock]

One issue about CCRL that I pointed out several years ago was the issue of many games forced into a draw adjudication after 250 moves (500 plies).

The results of ccrl 40/40 games ending in 500 plies (250 moves) are 136 games, all 136 were draws.

Looking at the 100 previous plies combined with the next 100 plies (400-499 and 501-600), the results are 727 games, 621 draws (85.42%), 56 White wins (7.7%) and 50 Black wins (6.88%). A total of roughly 14.5% non-draws.

An average of 3.63 games per ply, far less than the 136 games for just 500 plies.

And by inference, 14.5% of the games ending in 500 plies would have gone on to have a non-draw result. Also perhaps one of those games could have exceeded the 351 ply longest game in length.

[IanO]

I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

That's a neat bit of history, Larry! I remember how slick those first digital clocks like the Caissa were. I even started designing my own clock to learn how to program an 8048.

But I don't recall the Micromate. Do you happen to have a picture of one?

[Xann]

Hi Larry,

I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

Increment is good. Even better when added before every move, although the practical impact is negligible. I would like to know whether the decision to add it only after testing the remaining time was due to (non-maths) constraints in the clock design. It feels so wrong to me that I would at least prefer that there was no choice ...

Note that I haven't checked which version Fischer advocated.

Thanks,

Fabien.

[Graham Banks]

A good illustration of why none of the top programs test with repeating time controls, only with increment. There are better ways to spend one's time (or your computer's time) than playing out drawn endings forever at a repeating time control. Repeating time controls are pretty much obsolete even for humans, more so for engines. You can get a noticeably higher quality game in the same average time with increments. But I'm not complaining, Komodo seems to top the rating lists either way. I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

A number of engines did not cater for incremental time controls, and this was an issue when the major rating lists were set up because it would have restricted the range of engines that could be tested.
Not so much of an issue today of course.

The thing that I personally like about repeating time controls is the consistency of quality play throughout the games.
With incremental time controls, this can suffer.

[lkaufman]

I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

That's a neat bit of history, Larry! I remember how slick those first digital clocks like the Caissa were. I even started designing my own clock to learn how to program an 8048.

But I don't recall the Micromate. Do you happen to have a picture of one?

No, I don't have a picture. They were big, heavy clocks with one side covered with buttons. More like a computer than a clock. The manufacturer didn't know it could support what we now call increment until I told him about it and gave the instructions on how to program it. He then enthusiastically promoted the idea, giving me credit.

[lkaufman]

Hi Larry,

I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

Increment is good. Even better when added before every move, although the practical impact is negligible. I would like to know whether the decision to add it only after testing the remaining time was due to (non-maths) constraints in the clock design. It feels so wrong to me that I would at least prefer that there was no choice ...

Note that I haven't checked which version Fischer advocated.

Thanks,

Fabien.

I think clocks have been inconsistent on that point, whether to add the time before or after the move. I never considered that important enough to even express an opinion. In the U.S. it is more common to use "delay" (also called "Bronstein") where the clock justs waits X seconds before ticking down. The advantage is that there is no reward for super-fast play or for playing waiting moves to build up time. The drawback is that once your main time is gone you have to make every move in X seconds, no chance to make easy moves faster to save time for hard moves. For computer testing I think increment makes more sense than delay; engines won't abuse it.

[lkaufman]

A good illustration of why none of the top programs test with repeating time controls, only with increment. There are better ways to spend one's time (or your computer's time) than playing out drawn endings forever at a repeating time control. Repeating time controls are pretty much obsolete even for humans, more so for engines. You can get a noticeably higher quality game in the same average time with increments. But I'm not complaining, Komodo seems to top the rating lists either way. I admit to a certain bias, as I invented increment play (for the Micromate digital chess clock around 1980), not (as widely believed) Bobby Fischer.

A number of engines did not cater for incremental time controls, and this was an issue when the major rating lists were set up because it would have restricted the range of engines that could be tested.
Not so much of an issue today of course.

The thing that I personally like about repeating time controls is the consistency of quality play throughout the games.
With incremental time controls, this can suffer.

This objection is minimized by setting the increment to 1% or more of the base time, as is becoming standard now in engine testing. On average there is just less need for taking time in the endgame. Of course some endings really require it, but many are just hopelessly drawn and the time spent on them is wasted. A one percent increment is a good compromise.