Subject: Never won winnables!
Date: Sat Mar 19 04:56:47 2022
User: Klepp
Message:16332-5 8X2 first of 34 Elo 2605
Date: Sat Mar 19 19:05:59 2022
User: Punster
Message:This wasn't a Winnable, but it's the first time in quite a while for me 12x0 21384-7 => record 1/1.
It was the first time the game was played, ELO was 1580.
Date: Sun Mar 20 03:47:32 2022
User: Patti_Scialfa
Message:Wowser. I was shocked at this statistic. A little bit tricky, but not a mind-bender.
8x2 28330-5
Game stats: 35/1/2.9%/2807
Player Elo: 1774.917 ▲ 7.981
Date: Sat Mar 26 09:55:52 2022
User: rbw--3
Message:7x5 27800-11 HotStreak 5:05 Won 20% 1849 1 5
Date: Thu Mar 31 14:36:23 2022
User: Klepp
Message:17944-5 11x0, first of 32, Elo 2224...sometimes I'm decent at this game...
Date: Sat Apr 9 13:46:19 2022
User: RottinJohn
Message:Another one of the 10X0 down
Game: 10x0 3998-5
Game stats: 25/1/4%
Date: Tue Apr 12 10:58:37 2022
User: Klepp
Message:32658-6 10x1 first of nine...
Date: Fri Apr 29 15:21:44 2022
User: TitanicTony
Message:You Win. ..... Game: 10x0 10857-5; ..... Game stats: 27/1/3.7%/2155; ..... Player Elo: 2296.181 ▲ 2.511.
That Elo increase is REALLY disappointing, 🙁!!!
Date: Fri Apr 29 17:13:46 2022
User: rbw--3
Message:This one seemed somewhat easy. was surprised it hasn't been won.
Game: W8x3 18300-8
Game stats: 5/1/20%/1597
Game ELOs in 8x3 always seem very low.
Date: Fri Apr 29 17:50:44 2022
User: Dr.Bombay
Message:TT wrote:
"You Win. ..... Game: 10x0 10857-5; ..... Game stats: 27/1/3.7%/2155; ..... Player Elo: 2296.181 ▲ 2.511. "
Well, if i won that game, I'd get a much bigger boost. Oh, and this is an argument against blindly setting unwon games to an Elo in the 4000 range.
Date: Fri Apr 29 18:50:14 2022
User: TitanicTony
Message:I never suggested "setting unwon games to an Elo in the 4000 range"!
What I suggested was that if anyone loses an unwon game, then the Elo loss should be set to ZERO!
Date: Sat Apr 30 01:33:43 2022
User: TNmountainman
Message:So, Tony..........what would you do with, for example, 7x6-9? Quite a few of those have never been played, and many of them have been played only once. So what if a player draws one that he/she loses, to make it 0/1, or 0/2? Even if it turns out not to be that hard of a game. Should that player not receive an appropriate loss penalty IF that game turns out to not only be winnable, but maybe even not all that hard? And that's just one example I pulled nearly at random. We all understand your point, which is the same one many of us have been making for a year or more. But if the 'will' is not there to filter the unknown ones, there's really just no reasonable way to fix that, short of letting big passages of time to do the work.
A better plan would be to check your "elo ego" (not being snarky; just factual here), and just enjoying playing the games you want to play.
Even all that notwithstanding....................you and your deck have been around the block enough times now that you know the risk/reward ratio.
Date: Sat Apr 30 14:50:08 2022
User: Klepp
Message:Tarot?
Link: Tony?
Date: Sat Apr 30 14:59:26 2022
User: Kaos
Message:Well, if we set ELO losses on Un-won games to zero, your bump on that 10x0-5 would have been something like 0.25….
Date: Sat Apr 30 15:13:06 2022
User: HopDiriDiriDattiriDittiriDom
Message:"So, Tony..........what would you do with, for example, 7x6-9? Quite a few of those have never been played, and many of them have been played only once. So what if a player draws one that he/she loses, to make it 0/1, or 0/2? Even if it turns out not to be that hard of a game. Should that player not receive an appropriate loss penalty IF that game turns out to not only be winnable, but maybe even not all that hard?"
All 7x6 games are known to be winnable winnables. That variant can be excluded but it is a good idea for some other narrow variants. All unwon games can be assumed to be unwinnable and be given 5000 Elo (like in 10x0).
Date: Sat Apr 30 16:41:47 2022
User: TNmountainman
Message:Yes, any 7x6-9 would be winnable - but according to Tony's wishes.................anyone who plays one, and loses it, would receive no penalty, providing it hadn't been won by somebody else first. Even tho it's known to be winnable. And like I said, I just conjured that up at random. I was just picking one that had many games with few plays.
Date: Sat Apr 30 17:01:50 2022
User: HopDiriDiriDattiriDittiriDom
Message:There must be a way. So are you against the idea setting unwinnable games to a high Elo? And you were also against the idea that setting unwinable games to a high Elo whose numbers were obtained by using a solver.
So we accept it as is and lose unfair points to unwinnable games?
Date: Sat Apr 30 17:54:00 2022
User: TNmountainman
Message:I am NOT against the idea of setting unwinnable games to a high elo. I'm not sure I understand your second sentence. Maybe refresh my memory on that.
Since the solver stuff is "out of the bag", we might as well use that data to make the unwinnables have a high elo. But only if we do them all. And all I've 'heard' is that nobody that *can* do that is going to. Maybe I've misinterpreted that. It only penalizes some if only the 11x0s, or 12x0s, or 6x5s, or whatever get done. MrFixit keeps saying it's the same for everybody. I keep pointing out that it's not. It's the same ONLY IF everyone were playing the same variants and same climates. Which they're not. He has multiple times put forth the idea that it's ok to game the system by fishing for higher elo payoffs. (And maybe that's what he's thinking by saying it's the same for everybody.??) And yet I'm pretty sure the consensus here is that to be fair, it shouldn't matter AT ALL what variant one chooses to play. Even the playing field for all. We don't have that now, obviously, but at least everybody knows (or should) what they're getting into - at least generally.
So assuming I *have* 'heard' that correctly (and that *is* an assumption)..............we're stuck where we are, and it doesn't do any good to complain about it. Tony and his deck have been down these roads enough to know, to some extent, what's he's in for. And that's not even counting your formula thing, if he wanted to look at it from that perspective and take that time (and he obviously doesn't).
Date: Sat Apr 30 18:53:23 2022
User: TitanicTony
Message:I should clarify that the "zero points lost" doesn't apply to variants where ALL the games are known to be winnable. Most notably, the 10 "winnable" variants. But also some others.
Date: Sun May 1 00:54:33 2022
User: HopDiriDiriDattiriDittiriDom
Message:"I'm not sure I understand your second sentence. Maybe refresh my memory on that. "
TN, maybe I misunderstood the following:
"Hypothetical example:
Billy Bob is playing 10x0-5s. He's not really trying to find un-won 10x0-5s that *can* be won; he's just having fun playing them. By chance, he plays 15512-5 (from JackStraw's list above, and then apparently(?) then solved by Uberman). That game was 0/16 after JackStraw's 'miss'. Without looking it up, I'm assuming it's now 1/17, or something similar. So.........Billy Bob happens upon it, and through good play, beats it. But instead of the fun and surprise of achieving first assent, he's only second ascent, because this 'quest' has denied him that sublime notch in his belt. [And not that it matters for the main aspect of my argument, but he's also gonna get far less elo points.] Hopefully THAT (both points) make(s) sense. Again..........I don't expect to stop this process - but I see it as a negative to a nice aspect of the game. "
Date: Sun May 1 04:20:27 2022
User: TNmountainman
Message:Ah, two different things going on. Two different quests.
(1) As to the one above, Billy Bob is 'robbed' of the chance at that glorious first ascent by the collective efforts of all those trying to knock out wins in all the winnables in a particular variant and climate - THAT ARE ALREADY KNOWN (via cellmate's work). I personally think that removes a bit of 'magic' from the site - but I also think that's just a matter of opinion how important that is, regardless of (I think) not being able to refute my point. To me..........it's much, much cooler to win a game to make it 1/17 instead of 2/18. (And again, that's separate from the elo points awarded difference.) How important that is is a matter of opinion, I guess, but it clearly removes that aspect from that variant/climate subset.
(2) The *other* scenario, as described above by me (Sat Apr 30 17:54:00 2022), is the process of *discovering* the winnables and unwinnables - by computer solvers - for the purpose of removing them from penalty inflictions. That process shouldn't/wouldn't have any effect on the current status of any game, as I understand it. That 0/16 game would still be 0/16 until Billy Bob cracks it; so that thrill would still be available for him. Unlike scenario (1).
If'n I had my druthers...........and I expressed this at the time when cellmate first did it.................he wouldn't have "gone there", and these things would still be 'mysterious' (and at the same time vexing, true enough). But it's too late for that, so the question has become..........what's the best way to 'use' that data/tool? That's why I'm NOT against using that data, now that it's "out of the bag", for 'good' by removing the penalty for unwinnable games. NOT unwon games; only unwinnable ones, contrary to Tony's crazy idea. But the unwinnables should never ever ever be removed from play, because of the historical data set. All that stuff is my opinion, obviously, but I believe backed by sound logic.
(2) above is much more germane to everyone's complaints about the elo quest. (1) is more a matter of choice/preference for having a bit more mystery in the game. Is it more important to conquer game # 27xxx-6 (and all its brothers) in 8x3 "because it's there" --- or is it better to leave them be so that Billy Bob can get a kick out of beating it 'randomly'?
Date: Sun May 1 04:43:48 2022
User: HopDiriDiriDattiriDittiriDom
Message:So my suggestion (and Tony's) to set all the unwons (instead of unwinnables) to high Elos is in line with (2). And we don't have to use any data like in (1).
Date: Sun May 1 04:53:00 2022
User: TitanicTony
Message:"Zero" Elo points lost for losing an un-won game was meant to be an "easy", but temporary measure, until the "problem" of the unwinnable games was finally solved! The "problem" of the unwinnable games has been with us for way over a year now, 😕!!!
But, my Elo score went up 10.86 points yesterday, so today I'm a happy player!
Date: Sun May 1 12:35:01 2022
User: TNmountainman
Message:@ Hop.................... If I read your comment correctly, then, no, that's not what I mean, or agree with. Yet again...............if "the system" sets an unwon game to an artificially high elo number, and then later that game *IS* won, the "the system" was wrong in that assessment.
It's possible we're still thinking the same thing, but as I interpret your words, we're not.
Date: Sun May 1 14:48:24 2022
User: HopDiriDiriDattiriDittiriDom
Message:" Yet again...............if "the system" sets an unwon game to an artificially high elo number, and then later that game *IS* won, the "the system" was wrong in that assessment."
Yes I agree that "the system" is wrong in that assessment but when such a game is won, "the system" can change that high value just before the update takes place. We have an avarage rating for a variant and based on that "the system" calculates a more reasonable rating and the update takes place.
Date: Sun May 1 16:07:59 2022
User: TNmountainman
Message:All that I've been saying is based on two more, un-stated assumptions:
I've been led to believe, by ElGuapo, I think, that there were not going to be any retroactive changes of that type. Maybe I'm mistaken in that understanding. I further don't know the mechanism by which "the update" takes place - but don't think I need, or would even want, that kind of knowledge. I think we just trust the powers that be to correctly handle all that.
Further............I've been under the strong impression that there was no "will" (my term) to go through the entirety of all the known (as per cellmate) deals and somehow assign the unwinnables very high elos - which would be a mammoth undertaking. Thus this ongoing discussion seems, based on (my) current knowledge, to be moot. Again, that's just been my impression, so could be faulty.
Addendum to my first point above: if I understand your proposal correctly............Billy Bob might play an unwon AND unwinnable game, lose it, and see that that game has an elo of only 2000, let's say. Then........if that gets adjusted to, say, 4000, and then Teresa Sue comes along and plays it, her penalty would be negligible. Which wouldn't be fair to Billy Bob. That may be a minor point, but it should be thought about. I could say more about this, but this post is already far longer than I wanted.
Date: Sun May 1 16:17:20 2022
User: TNmountainman
Message:P.S. @ Hop --- As to your last paragraph above, my response sorta addressed the converse (I think) of what situation you were examining. But the theory still holds. If Billy Bob lost a game that was 0/40, AND your and Tony's proposal had been adopted, he would lose zero elo points. But then Teresa Sue comes along and - lo and behold! - she beats it to make it 1/41, then not only was "the system" wrong, but those two players were evaluated (inequitably) by different standards. Would Teresa Sue get credit for beating a 4000-elo game?
Date: Sun May 1 16:31:15 2022
User: HopDiriDiriDattiriDittiriDom
Message:"Teresa Sue comes along and - lo and behold! - she beats it to make it 1/41, then not only was "the system" wrong, but those two players were evaluated (inequitably) by different standards. Would Teresa Sue get credit for beating a 4000-elo game?"
It's not a 4000-elo game actually right? It's only a way we try to deal with low rated unwinnables. Teresa Sue would only get the points based on the adjusted/reasonable rating. Until then the losses practically do not generate any exchange (since it's a high Elo game until won). Yes that's a comprimise we have to make to design a simple (but effective in my opinion) system. I like simple designs. I had provided a link about it and I will do it again below.
Link: KISS
Date: Sun May 1 21:20:05 2022
User: TNmountainman
Message:Most of us here in the U.S. are very familiar with that philosophy. And..........as we were getting this elo stuff rolling........Denny (and others?) were advocating "perfect is the enemy of good (enough)", a somewhat similar stance, altho not exactly the same. Perhaps(?) first stated by Voltaire, but we don't really know. And Churchill famously said something very similar.
But..............to us idealists, that's somewhat like "caving in". There are many, many, many cases where "good enough" is the correct approach. As is KISS. Not sure this is one of those times, because we're talking about a case where computing power, time, and philosophical and competitive distinctions are at stake - and/or being brought to bear. Its not like vacuuming a floor, building a house, or cooking grandma's spaghetti sauce. We have, I think, the capability to get this very close to "perfect", whatever we decide that is. And obviously, there will be disagreements about these things. I wonder (but not disparagingly) if cellmate doesn't fully appreciate the value of the database, as he's a relative newcomer, and more focused on elo stuff. He didn't have nearly as much experience as most of us utilizing that as the competitive proving ground.
Again, all just imho.
But again.............isn't this all a moot point, based on what we know now?
Date: Sun May 1 21:23:39 2022
User: TNmountainman
Message:".......utilizing [the historical database] as the historical proving ground."
Date: Tue May 3 14:28:31 2022
User: TNmountainman
Message:9x2 13327-6
7/1/14.3%
Surprising, but not utterly shocked.
Date: Fri May 6 00:41:27 2022
User: Klepp
Message:18490-7 10x1 first of six...
Date: Fri May 6 12:23:18 2022
User: redberet
Message:I'd just got my azz kicked on losing a winnable and "luckily" I got this next. Ha to the dealer
Game: 12x0 18828-11
Game stats: 20/1/5%/1897
Player Elo: 1914.297 ▲ 3.894
Elapsed: 9:27
Streak: 1
Date: Fri May 6 18:57:09 2022
User: Klepp
Message:19505-7 10x1 first of 2...but, worth over seven points people (looking for fattening Elo...
Date: Fri May 6 21:39:56 2022
User: joeygray
Message:I don’t get it. I been playing 10x1s all day, and I generally get about a quarter of a point for winning and minus 7.75 points for losing. I have never seen a 10x1 you could win 7 points with. The only ones rated that high are unwinnable ones, you caint get nuthin from that turnip.
Sometimes if they really are tough, you can get 1.8. Or maybe even over 2. But 7?
(Of course I know it depends what your own elo is, but Klepp yours is higher than mine. So how?)
(Btw, losing too many today. 😦)
Date: Sat May 7 13:16:20 2022
User: Klepp
Message:I can't fully explain why a seventh level 10x1 that had been lost to solely once (unbeaten) would have an Elo as high as 2200...gold nuggets still to be found...I blew the seven I got on it on the very next game, anyway though...game giveth, game taketh...
Date: Tue May 10 12:54:42 2022
User: Klepp
Message:30567-5 8x2...5+ Elo pts....first of 49...sweeeet...tough....
Date: Tue May 10 12:57:15 2022
User: TNmountainman
Message:Nice; but gosh, what does it take to get the full complement of the k value. 1/500?
Date: Tue May 10 14:03:52 2022
User: Klepp
Message:I blew the points on a (tough to me) ensuing game that was solved around half of its attempts...eight by two is a brutally unforgiving bitch...fearfully tough...
Date: Thu May 12 06:50:38 2022
User: Gulbis
Message:"I don’t get it. I been playing 10x1s all day, and I generally get about a quarter of a point for winning and minus 7.75 points for losing. I have never seen a 10x1 you could win 7 points with. The only ones rated that high are unwinnable ones, you caint get nuthin from that turnip.
Sometimes if they really are tough, you can get 1.8. Or maybe even over 2. But 7?"
There are some level 10 games that have only been played once and it was a loss. Due to the nature of the 10x1 variant ElGuapo's algorithm thinks it's quite likely that the game might be unwinnable so it gets a relatively high Elo, around 2300 if I'm not mistaken. If you happen to win such a game you'll (unfairly) get a lot of Elo points. That's why I'm happy that ElGuapo is finally using cellmate's list of unwinnable games and adjusting the ratings.
Date: Thu May 12 07:03:23 2022
User: HopDiriDiriDattiriDittiriDom
Message:"I have never seen a 10x1 you could win 7 points with."
You can use an Elo calculator to experiment with various ratings. As for my rating, a game with a rating about 400 points above mine give me more than 7 points. I didn't check the list of 10x1 games. if there is a game with a rating about 400 points above yours in that list, it should be the one.
Date: Thu May 12 07:40:57 2022
User: joeygray
Message:Yeah. I daresay there might be one, or even 30, such 10x1 games; just cause I haven't seen one don't mean nuffin. But I took Klepp to mean you could routinely raise your elo that way. On a thousand to one shot, nah.
Date: Thu May 12 13:12:01 2022
User: Klepp
Message:I think I'll return to them, though; I recall being nicely/fairly rewarded (opinion) after the due diligence necessary to ascend to random games...
Date: Fri Jun 3 15:20:40 2022
User: Dr.Bombay
Message:Here's an example of why unwon games shouldn't get Elo ratings of 4000:
Game: 7x3 3720-5
Game stats: 51/1/2%/2168
Player Elo: 1805.398 ▲ 7.184
Elapsed: 2:39
Streak: 1
Date: Fri Jun 3 17:27:58 2022
User: Kumquat-of-Conciliation
Message:Exactly. But it should likely be well more than 2168 - altho you probably got near the maximum points anyway. But seems to me that that kind of win deserves at least 7.7 or so.
Date: Fri Jun 3 17:53:25 2022
User: Dr.Bombay
Message:Yes, I thought the same thing. I'd be peeved if my Elo rating was 2200 and I saw I was expected to win a game that was previously 0-50.
Date: Mon Jun 6 06:35:59 2022
User: TitanicTony
Message:You Win. ..... Game: 11x0 19976-5; ..... Game stats: 69/1/1.4%!!!
Date: Tue Jun 7 05:14:33 2022
User: TitanicTony
Message:Btw, the original purpose of this thread was to post unwon games, so that others could try them.
That game above, 11x0 19976-5, should have been posted here a long time ago.
Date: Wed Jun 8 18:45:44 2022
User: Klepp
Message:19413-7 8x3-7
I can brag on this one...second of 174...first try...5+ Elo...
Post follow-up