1.g4 opening is losing?

[Leto]

At regular ICCF time controls (10 moves in 50 days) I'm certain I can beat 1.g4. I'm playing a game on ICCF against an 1.f4 master and we're now in the middlegame and I'm feeling very good about my chances at winning. 1.g4 is worse than 1.f4 I'm sure.

[Ovyron]

At regular ICCF time controls (10 moves in 50 days) I'm certain I can beat 1.g4.

If you posted this back in September 2019, I'd have told you "and I'm certain I can draw you!", but after all I've seen, I believe you.

[whereagles]

what's all this talk about "demonstration is not a proof"?

they're the same, as far as I know

[jp]

It sounded like "demonstration" was referring to someone winning a CC game with Black.
Do you think "demonstration" is too strong a word for that?

[Zenmastur]

what's all this talk about "demonstration is not a proof"?

they're the same, as far as I know

They are not the same. If I produce all ducks in such a manner that any interested observer can see them and they are all white then I have demonstrated that all ducks are white. It's not a proof. It doesn't rely on proven mathematical theorems used in some particular order to “prove” a new theorem, namely, “ All ducks are white”.

It sounded like "demonstration" was referring to someone winning a CC game with Black.
Do you think "demonstration" is too strong a word for that?

A proof tree “IS” a demonstrations and not a proof. That is why it's the last word on if the mate is correct or not. There are no mathematical theorems involved. It's simply a demonstration by enumeration of all possible variations. This is also why they are almost useless in mathematics in general. Any time you are dealing with infinite sets, it become impossible to demonstrate anything. Even with large finite sets demonstrations become useless. Chess proof trees are a perfect example!

E.g. I was going to post a mate in 50 as an example and then I started doing some rough calculation on the size of the proof tree. It turns out that it would take in about 10 octillion yottabytes plus or minus a few orders of magnitude depending on the average number of mover per position for the losing side, the amount of transpositions and assuming you could encode it at one byte per position.

It couldn't be stored, as there isn't enough storage on this planet for that. Even if you could processes it at 10^18 nodes per second (seems unlikely) it would take 10^32 seconds to verify it. Just for reference this is approximately 10^14 times as long as the universe has existed.

So, pray tell, what use would you have for such a thing. You can't store it and you can't process it so what good would it be to you or anyone else?

Even if we take an EXTREMELY small proof tree, something that could actually be stored, and someone has some chance of processing it within a reasonable period of time, say 10 exabytes, what good would it be? A much more pertinent question would be, would it be worth the time and effort to create it?

While I might entertain arguments about the length of a mate, for me, arguing that I can't produce a proof tree and therefore there is no mate seems utterly ridiculous. Further, going on and on about 32-man table bases is even worse. They can't be made to exist by human endeavor since the number of subatomic particles which it's possible for us to manipulate is much smaller than the number of entries in a 32-man table base.

Enough said!

Regards,

Zenmastur

[Ovyron]

Well, I think that my game with Harvey did "demonstrate" the strongest possible defense and attack of 1.g4 d5 2.g5, so at least that'd be proof that 2.g5 loses. You're welcome to disagree, and suggest a better way for white to defend, but as you go deeper your "improved" defense will look worse than what was played (or both end in mate scores, so your improved defense holds for more moves, but still loses).

1.g4 d5 2.Bg2 is having a similar treatment right now, with David Vs. Goliath added (I'm winning as black despite big hardware disadvantage and 23 nps Leela.) It could demonstrate how 2.Bg2 is lost as well.

It's conceivable that the same treatment could be had for the other second moves, so 1.g4 would have the same lost status of black starting without a queen in the opening position (that also can't have a proof tree, but if you think it's conceivable that black can save that, something might be wrong with you...)

[Alayan]

The mathematical meaning of "demonstration" is "proof". In french, where this word comes from, a mathematical proof is called a demonstration.

They are one and the same, a demonstration doesn't need to involve brute-forcing all possible cases ; indeed most do not.

A brute-force proof is called a "proof by exhaustion", not a "demonstration".

[jp]

A proof tree “IS” a demonstrations and not a proof.

No, a proof tree would be a proof too.

(Also look back at Uri's post. But in either form it is a proof.)

The mathematical meaning of "demonstration" is "proof". In french, where this word comes from, a mathematical proof is called a demonstration.

I have to say, I don't think modern mathematical terminology (in English; I don't know about French) gives a technical meaning to "demonstration". I mean, 1697 was a long, long time ago.

[Ovyron]

Well, whatever the meaning of the word "demonstration" is, I'd like to see a demonstration of a draw line for 1.g4 where no obvious blunder by black happened.

That would be your black swan and would put the onus of proving 1.g4 is lost on our side.

[jp]

You'd be happy with a non-French, non-17th-century demonstration.