Posts:
64
Registered:
Jan 20, 2019
From:
Salem Oregon
Age:
48 Home page
Re: K10: The Supreme Overseer
Posted:
Apr 17, 2019, 7:22 AM
I think you are continuing to conflate opening theory with opening practice, which is what the DB actually is...a record of practice, the habits of players.
The number of games and players is nowhere near what completion would look like...sorry, it just isn't. We'd have lines worked out to force-wins all over the place, especially in the popular lines. We don't. We have "should be a win for P1 with best play".
Furthermore, most of the solid opening theory, loosely defined, happened just as you said...prior to the 1984 selling of Pente. 35 years is a long time. I think if you look back through some of Rollie's comments here, you'd see a similar train of thought...this isn't as done or solid as people think it is. It's where they stopped and moved on...
As for proof of incompleteness, I've resurrected and explored a few lines...exactly how much revitalization is necessary to prove that it can be done? Or that Pente opening practice hasn't made it to a solution yet?
Do you think Pente is solvable for P1?
I'm not saying it isn't...I have my own pet variation I'm working on just in case it is...but the idea that it IS solved is just not realistic...even among the tippy-tops...
Of course, I'm just a few hundred games in...not thousands...karlw asked what would get me to change my mind and I replied essentially that actual proof would. And it will if it arrives...just like my 3rd move experiments with him...until it does though, I'll keep my opinion on what's fact and what's dogma about where it is...
Re: K10: The Supreme Overseer
Posted:
Apr 17, 2019, 9:35 AM
@haijinx:
If and when you can give a final answer as to what White's best 3rd (or 3rds if there are multiple best) after K11 is, and for good measure M9, then I will tell you how J7 objectively compares to N10. I haven't seen anything convincing. Until then, all we have is the continually evolving record of master games, which paints a pretty clear picture IMO.
It's no use going back to yesterday, because I was a different person then.
Do I think that Pente is solvable as a win for P1? In the sense of always being able to locate a single move ( usually out of several) which leads to a sure win position for P1 from a given position or if the position is analysed as a sure win for P2 to rewind the position until a better P1 line is found? Yes, absolutely. Has it already been done? No. Non swap renju was shown to be a sure win in about 17 moves around 17? years ago. Captures don't complicate matters that much over renju and can lead to wins by captures which aren't available in renju. Progambler indicated he thought it was likely to be a sure win in 20 moves back around 2003 when he was still around - sounds accurate to me from what I've seen. With Deep Learning advances it would take relatively little time and effort to run self play AI for a few million games to train an AI to superhuman play and build a superhuman DB from that point forward. I'd guess an AlphaPente program would at least split TB sets with any human in about 8 hours or less. After that, run it against itself to build a DB and do retrograde analysis if necessary. Pente solved, at least weakly.
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Guys, it's a long post...I did point out at the start that J9 is the most prevalent response to J7 at ALL level of play, with examples from
ksackett dmitriking pente_gon
To be clear, I'll add that most of those people had access to the DB too. There's even an example from watsu.
It's the mainline...by thousands of games.
And until I dug in there, J9 had as good as record as anything, within a few percentage points of K11. Furthermore, my defeat of J9 makes use of actual opening theory via Tom Braunlich's concept of overseers, a concept that doesn't appear to have been discussed at pente.org ever before my post.
I've only had a few games in K11, it'll come if it hasn't already...I've had 7 losses in the main line that I've solved after all. And its "master level" results are 2-35 (both losses by dmitriking, with DB), but none of them faced my adopted innovation of L5.
That is an error in "opening practice" that I found, is it not? Yeah...finding a winning choice in a previously "written off by masters" line, as I have with L5, does actually prove the point that "opening practice" is neither complete nor necessarily correct.
That's especially true when it's the mainline. Not my fault the mainline isn't the best line, it's my point.
I'm not at all worried about M9. Maybe someday someone will champion it after K11 fizzles out. The bar will continue to shift with every novelty that I solve on my own, though. I know that and I think you do too.
So, until the K11 novelty stands up to ALL of the basic responses and even the oddball ones, and like karlw said, it probably can't as P1 has initiative to spare, I'll stick to the pragmatic over the dogmatic and explore K11 whenever it comes up.
Regardless, J9 is mainline, it was considered sound, now it isn't. You're welcome. If karlw saves K11, and he could, then it's possible that will become mainline...in TB play. And that will shift my playing style and choices some.
I'll wager that it'll never be mainline in OTB though...J9 and L11 are too natural.
I agree with you that AlphaPente could solve the game if it's solvable. And I wager it could take on all humans in at most the 8 hours you suggest.
I've read the idea that it should take 20 moves or less before...I'm not convinced yet, but could be. That's the reason I'm working on my own pet variation.
Remember though, in the early 80s, opening theory was "solved" with the hammer. Then came the wedge. And then came...less theory, really.
My point is that it's not solved for P1 yet, people just act like it is a given that it will be and moreover that the currently popular mode of play will be the solution to the game.
I agree on the less theory after the wedge. Wedge is unlikely to be optimal, probably Rollie introduced it simply to give himself lots of tricky chances for a P2 win after his opponents adopted it once they faced his Hammer. K10 L9 N10 N9 K11 likely wins in quicker, simpler and generally easier ways. As far as I know there was never any particular theory on that line. So, theory is lacking, I agree. Maybe you'll have some master level converts to K10 L9 J7. Perhaps you'll get to test it out further against a few more of the Pente_gon, Brf, Zoeyk, Happyjuggler, Nosovs crowd. If not, once our AI overlords arrive, it'll be easy enough to plug the haijinx opening into the computer to add it to the DB.
Oh, and by Pente theory being tested, I'm simply referring to stuff from Braunlich, BTW. Basic P1 opening shape stuff like: The L and the hat are the two strongest three stone formations P1 can make. Now, you can call that opening practice if you choose to do so, but I consider it to be tested theory at this point. One or the other of those two shapes will work against nearly anything, and then you've always got the fallback position of K11/N11 instead of playing into a hammer for the oddballs against which the hat and L fail.
Sure, the T is also strong and very winnable in many instances, as is the small diagonal T, which is what makes going offset work. However, you can make two big Ts and two diagonal small Ts as thirds, which gives you four options - versus 4 Ls and 4 hats, which is why the theory that staying online was stronger than going offset gained traction in the first place.
Some more modern players do favor going offset against certain firsts by P2. I'm on the fence as to whether it's either necessary or optimal to do that and haven't landed on one side or the other currently. I go back and forth on it. Sometimes, I try x _ _ x sometimes I go offset against particular P2 firsts. Message was edited by: watsu at Apr 17, 2019 3:30 PM
Message was edited by: watsu at Apr 17, 2019 4:51 PM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
Unless P2 plays K10 K9 N10 N9, invites wedge/hammer or creates certain pairs, Zoeyk's opening book is all hats and Ls against any K9 or L9 second. The hats and Ls (and the x _ _ x which enables their formation) are the prevalent theory against two of P2's strongest three first moves (according to Tesh). Are other P1 choices for seconds and thirds besides the ones zoeyk picked also winnable? Of course. A few, like the one I mentioned earlier are perhaps preferable if one desires a quicker or easier win. Some, like K10 K9 K7 and K10 L9 G10 don't go offline, but diverge from the main thoroughfare and yet are still likely to be sure P1 wins (my opinion, but others share it). If the goal is to find all sure wins or the easiest/shortest sure win path for P1, this is a more ambitious goal than to demonstrate a single sure win choice path for P1. Personally, if I were attempting the latter, I'd see whether or not anything can beat a suitably chosen x _ _ x and build from there. Yes, this is my (and a common) practice, but it's rooted in Braunlich's theory.
Message was edited by: watsu at Apr 17, 2019 7:02 PM
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
That's right boys and girls, it's time for a good old-fashioned
P E N T E S H O W D O W N
haijinx's J7 Vs. karlw's K11
The K11 killer is currently 2-0 vs. the J7 juggernaut, but J7 has all the confidence of P1's advantage, mixed with the bravura of youth.
What's this? k's K11 just walked onto the stage, full of swagger, and declared that he will win the first 5 matches without breaking a sweat! The audience erupts into a mixture of raucous cheers and boos. The spotlight turns to jinx and his J7...
It's no use going back to yesterday, because I was a different person then.
By the rating system, you should win 75% of our games on an even field...which doesn't exist in Pente...today...though we're both not exactly ratings stable.
In short, you beating me isn't actually proof except that your experience still trumps my intuition...it really doesn't prove anything about the variation though...especially if I have to take two times to knock down each one. Just saying...
Also, I'm older than you...just less experienced...you pente tramp... (just getting into the mood...nothing at all against tramps)
If Karl is a Pente tramp, that must mean I'm a Pente madam, based on my age and experience. I'm not sure how I feel about that... On the whole, I think I'll probably stick with my black belt in Pente T shirt when I want to advertise (it was a gift, btw, not something I'd have bought myself).
Retired from TB Pente, but still playing live games & exploring variants like D, poof and boat
"Playing as black and promising to play the same 2nd five times in a row is a pretty big deal! My owner played K9 against snapding, and lost three times in a row to his ****ing K7, playing L11 each time, and seeing a different 3rd in reply each time--and snapding is 2200-2300 (although you wouldn't believe it from looking at his K7)! This left my owner so shaken that he switched (successfully) from his beloved K9 to L9. If a 2200 can beat a 2700's Black three times in a row, then a 2500 should be able to as well! Sometimes, the opening matters more than the players!"
And for those of you out there who want to play K7 now 'cuz it's the "cool" move:
Don't.
It's no use going back to yesterday, because I was a different person then.
Posts:
2,233
Registered:
Mar 4, 2007
From:
San Francisco
Age:
45 Home page
Re: K10: The Supreme Overseer
Posted:
Apr 22, 2019, 12:38 AM
Haijinx, unless im misunderstanding, appears in the early stages of pente development, indicated by the lack of belief that in perfect play, in regular pente, P1 will always win. I mean no offense by this. Its a stage we have all been in. Its natural.
I assure you its solved. But, I understand that no one can understand this until they complete their own journey to come to this same realization that so many masters have come to terms with. it can not be explained to you. It is something you must discover on your own, in your own time
Scire hostis animum - Intelligere ludum - Nosce te ipsum - Prima moventur conciliat - Nolite errare
Re: K10: The Supreme Overseer
Posted:
Apr 22, 2019, 7:23 AM
I'd be careful with your language, zoey.
Does P1 have a large advantage? All the data suggests overwhelmingly yes.
Is it enough to guarantee a win with perfect play? By definition since a game of pente cannot result in a draw (let's ignore the possibility that the entire board is filled without a pente or 5 caps), the only possible options are that White or Black has a forced win. Again, all the evidence favors White.
Is it enough to make it impossible for Black to win consistently against a master player without using surprise and trickery? You betcha.
Is pente solved? Absolutely, 100% no. If we devoted one tenth as much time and resources to solving pente as we have chess, then pente would probably have been solved 20 years ago. But for good or bad (good IMO--leave pente alone!), it hasn't been. We still have not yet found the guaranteed winning line.
Also, your post comes off as uncessarily condescending to haijinx, who is clearly a very strong player trying to explore new ground rather than retreading stale database lines--as White at least; as Black he keeps trying to beat me with the wedge for some strange reason . I think that's noble and that we should encourage it, as it takes us closer to solving pente.
It's no use going back to yesterday, because I was a different person then.