Houdini 3 running for the IPON

[IWB]

[...snip...]
EDIT: Thinking about it a time control like 1 move in X minutes is the logical conclusion out of the argumentation for a repeating time. Boring in drawish endgames but somehow logical.

The best timecontrol is "x moves in y time repeated", because the
program knows in every stage of the game exactly how much time is left.

Best wishes,
G.S.

That is true for sudden death as well!

But why 40/x? Why not 20/x or 10/x or 1/x? 40/x is because it was always that way. There is no other reason and that is why I called it "crusted" (I think that picture wors only in german).
As all programm use that x in 40/x individually the time control is nothing else then helping the engines/authors to play best chess within the given time. As time is part of the game if should be hanlded by the engine and not by the testers!
Thinking that helping to an end the logical time control would be 1/x as here the engine always now how much time it has for a single move.

Bye
Ingo

[Modern Times]

Yes some engines get into time trouble towards the end of their repeating time control allocation, but it is less of a problem than with fischer or game in x controls. And at least it re-sets again with the next allocation of time.

[IWB]

Yes some engines get into time trouble towards the end of their repeating time control allocation, but it is less of a problem than with fischer or game in x controls. And at least it re-sets again with the next allocation of time.

That is exactly my point. With this "re-set" YOU as a tester help the engine! YOU decide what is best game play, not the engine! Why do YOU want to help the engine, and interfere with its time alocation algorithm when doing a rating list?
If an engine gets into time trouble so be it. Time is a part of the game and if an engine can't hanlde that it is its problem not mine as a tester! An enigne with a good timing algo should be rewarded for it while a repeating time control penalises this good engine and helps engines with a less good timing algo!

Bye
Ingo

PS: I know it is a theoretical issue as no one knows how big this penalty or reward is, but why doing it at all?

[ThatsIt]

[...snip...]
With this "re-set" YOU as a tester help the engine! YOU decide what is best game play, not the engine! Why do YOU want to help the engine, and interfere with its time alocation algorithm when doing a rating list?
[...snip...]

And why helping an engine with an increment as you do (IPON) ?
Best wishes,
G.S.

[Graham Banks]

Yes some engines get into time trouble towards the end of their repeating time control allocation, but it is less of a problem than with fischer or game in x controls. And at least it re-sets again with the next allocation of time.

That is exactly my point. With this "re-set" YOU as a tester help the engine! YOU decide what is best game play, not the engine! Why do YOU want to help the engine, and interfere with its time alocation algorithm when doing a rating list?
If an engine gets into time trouble so be it. Time is a part of the game and if an engine can't hanlde that it is its problem not mine as a tester! An enigne with a good timing algo should be rewarded for it while a repeating time control penalises this good engine and helps engines with a less good timing algo!

Bye
Ingo

PS: I know it is a theoretical issue as no one knows how big this penalty or reward is, but why doing it at all?

We don't interfere in any way during the game. It's not like we sit at our computers 24/7 adjusting time controls in every game. That's just brainless.

You choose an incremental time control, we choose a repeating time control. As I said before, it is a matter of preference. There is no right or wrong way. Both have validity.

Ingo - please don't take offense, but you shouldn't be trying to preach that you know everything about what is right and what is wrong about running engine v engine testing.
You do a great job as I said, but you're not the world authority on it.

The results of all testing groups plus others who run engine v engine testing all contribute to the big picture.

[Modern Times]

With a repeating time control, I think the list has a better chance of showing how strong an engine is for analysis.

With an Fischer time control, perhaps the list potentially gives a better indication of how an engine will perform in tournaments because many do have that sort of time control, or a game in x. HOWEVER that all becomes meaningless, because many engines play SMP in tournaments and SMP efficiency is a factor, most use strong custom books, most will use at least 5 men tablebases if they support them etc etc.

So - no right or wrong answer. You take your pick and run with it. One is not better than the other.

[IWB]

[...snip...]
With this "re-set" YOU as a tester help the engine! YOU decide what is best game play, not the engine! Why do YOU want to help the engine, and interfere with its time alocation algorithm when doing a rating list?
[...snip...]

And why helping an engine with an increment as you do (IPON) ?
Best wishes,
G.S.

Yes, you are right, the +3 is doing the same "helping" (avoiding time battles) as repeating time controls.
You might not know it, but I mentioned several times on the CSS forum (and I think here as well) already that, if I would have to start the IPON again, I would play with an incerement of +1. And the + 1 is just a compromise to avoid some possible software/network/hardware lag.
In short, the +3s is the biggest downside of the IPON in my eyes - as 'I' see it today!

I would have to possibilities. Game should be finished in 15min. That is a reasonable compromise between fast and thorough. This translates into something like 6+1 or 7+1 (needs some testing) or I play shorter to get more games - 3+1 ... 5m+1s or 300s+1s looks promising! (That is 740-800s in average)

So, what about your repeating time controls?

Bye
Ingo

[IWB]

We don't interfere in any way during the game. It's not like we sit at our computers 24/7 adjusting time controls in every game. That's just brainless.

Right, who said you interfere in every game? That would be brainless. "You" interfere with the genereal decision, yes. But to say that is not brainless but logical because you avoid bad decisions of an engine. Any ARGUMENT against that?

You choose an incremental time control, we choose a repeating time control. As I said before, it is a matter of preference. There is no right or wrong way. Both have validity.

See my answer to Gerhard and my previous postings. Do they both have validity? No. The best would be a sudden death. Thats what I said right from the beginning. And I did it with argument but you felt attacked!

Ingo - please don't take offense, but you shouldn't be trying to preach that you know everything about what is right and what is wrong about running engine v engine testing.
You do a great job as I said, but you're not the world authority on it.

I do not preach and I never said I am THE authority, I argue - you jump in from time to time because you feel attacked and defend with this accusations about preaching or some ipontifi.... Sorry Graham, arguing is fine with me, accusations aren't, do you have arguments as Gerhard did it?

Bye
Ingo

[Modern Times]

5m+1s is a very nice time control. It is what is used in HG's monthly online tournaments.

I strongly dislike sudden death time controls like 5+0

[IWB]

With a repeating time control, I think the list has a better chance of showing how strong an engine is for analysis.

I don't know, I have to think about that, but we are making rating lists which ar ement to show plaing strength, not nalysis strength. To show that (analysis strength) some important parts are missing in our "rating lists" (e. g. backward analysis)

With an Fischer time control, perhaps the list potentially gives a better indication of how an engine will perform in tournaments because many do have that sort of time control, or a game in x. HOWEVER that all becomes meaningless, because many engines play SMP in tournaments and SMP efficiency is a factor, most use strong custom books, most will use at least 5 men tablebases if they support them etc etc.

That all correct. A live tourney is somethign different than a ratinglist!

So - no right or wrong answer. You take your pick and run with it. One is not better than the other.

Hmm I dissagree, but I proposed something better for both! Sudden death leaves the timing decision to the engine as it sould be for a rating list!

Bye
Ingo