A (Choose: Great/Pointless/Annoying) Scoring System Thread

Anything about the Diplomacy game in general.

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby GhostEcho » 28 Nov 2017, 06:55

I will sketch briefly the project/article I was working on, because I think I stumbled on something that's often overlooked by (or unknown to) players newer to the Hobby, but which ought to significantly affect the debate. Curiously, I don't know if I've ever seen an older player cite it, but then I'm not well-acquainted with the face-to-face Hobby and I don't know how many of those regulars play online. (I'm toying with the idea of working this into publishable form but I'm not sure the best place with the Pouch apparently down for the count.)

Over at the Diplomacy Archive, they maintain copies of old rule sets. A short game had been included as a possible variant in the rules for the majority of the game's existence. The board leader at a pre-set deadline wins. It's a little murkier than that: the first draft (pre-dating even the "Diplomacy" name) mandates the board leader wins even in a short game only if he's reached at least 13, while the first official edition omitted the short game variant - and since 2000, so have more recent rulesets, either by oversight or because "the first edition didn't", I don't know which. I have not been able to discover anything Calhamer himself wrote about this option, though with Google as my main research tool I may very well have overlooked something.

Simply counting wins/draws/losses is elegant (to me, obviously superior) in principle: but discovering these older rules was enough to convince me of one thing. For tournaments, where a result has to be reached on the evidence of only a few games, I am now strongly in favor of primarily size- or lead-based scoring within the context of a draw. It seems evident that on otherwise equivalent boards the player with 10 did better than the one with 8 (allowing possibly for country-based adjustments in theory?), and I'm inclined to agree (though this is more debatable) with those who say that the player with 10 and the lead also did better than the player with 10 running second to a 14er. The exact system of scoring chosen, within these constraints, I don't have strong opinions on - though I might if I investigated more closely.

However, the short game is still (at best) only an "officially endorsed" variant, and the time-constrained tournament reality an extrapolation from that. For games played without that time constraint, with the understanding that they all be played to the finish - a league with a significant number of games, or a site like PlayDip that maintains a player ranking - I think that the primary ruleset imposes additional conditions. The rules specify that all players share equally in the draw and always have - apart from a 1961 edition that left draws out completely, which given Calhamer's stance (as I understand it) has to be nothing but a monumental error.

Keeping in mind that the rules nowhere require or even suggest the complete removal of other players from the board, draw-size scoring is problematic because it creates incentives exterior to the game itself. But if draw-size scoring is to be avoided, and lead-based scoring is considered primarily as a convenience-based variant, this requires that the system adopted have a single score assigned to any player drawing. Representationally, this could be justified by saying the Diplomacy board sets a scenario with seven powers "at the table" and elimination marks a power's failure to keep his or her "seat": a binary result. (A solo overrides this, of course, by rule.) Such a system, then, would have a player perhaps earn 0 points if eliminated, 1 if surviving without allowing another to solo (and if so surviving therefore drawing by letter of the rules), and 7 (or other agreed value) for a solo. If in a situation (for instance, online) where players end up playing vastly different numbers of games, this might be then adjusted to a mean score, and/or a secondary system (anything from median center count to bonuses for board leads to another system entirely superimposed) used to break the virtually inevitable ties in a ranking.

This is clearly not a fixed-sum system: but then, I don't think Diplomacy is a "fixed-sum" game. Elimination is the "cardinal sin", at least as I see it, and any two games picked at random are likely to have differing numbers of players eliminated. While I would contend that it's the only "pure" rules-based approach to scoring, I will openly state, as above, that this long-term scoring system as I've laid it out would be utterly unworkable in a tournament setting - there are not enough games to shake out clear win/draw/lose rates, and frankly I don't see how a solo value could possibly be workably weighted under the time/board count constraints.
"When you absolutely don't know what to do any more, then it's time to panic." - Johann van der Wiel
"I'm not panicking, I'm watching you panic. It's more entertaining." - Elli Quinn
"[Diplomacy:] No dice or chance. Just calculated insincerity." - Counter Trap
User avatar
GhostEcho
 
Posts: 1668
Joined: 10 Aug 2008, 04:56
Location: PG County, MD, USA
Class: Ambassador
Standard rating: 1019
All-game rating: 985
Timezone: GMT-5

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby super_dipsy » 28 Nov 2017, 09:17

Just a quick question for people who may be interested in these things :)

I was wondering whether there would be any interest in seeing any statistics for games where there are no surrenders? It is something that can't really be done with simple database calls, because it is quote hard to sort out the games where noone has surrendered. However, it wouldn't take much to knock up a little program to gather the info and spit it out. I may do this anyway, because I admit I am now intrigued to see what the 'vanilla' statistics are based on our scoring system and draws approach. For example, how does DIAS affect things, and in particular how does it affect the different country performances if you strip out the surrenders.

My main concern though is some of the data is going to be skewed because of the selection criteria. When we look at games with no surrenders, we are not looking at games where surrenders are not possible. Therefore we would be looking at games where players chose not to surrender. So for example, in games where Austria is quickly stomped, the likelihood is in at least a portion of games the Austrian player will just surrender. If the game proceeds without surrenders, it may mean that the selection criteria has filtered out some of the 'poor start' games and only includes games where Austria starts well, possible skewing country performance statistics.

Anyway, do people think it would be interesting to see statistics for surrender-free games?
User avatar
super_dipsy
Site Admin
Site Admin
 
Posts: 10892
Joined: 04 Nov 2009, 17:43
Class: Ambassador
Standard rating: (1000)
All-game rating: (956)
Timezone: GMT

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby DQ » 28 Nov 2017, 18:16

super_dipsy wrote:Just a quick question for people who may be interested in these things :)

I was wondering whether there would be any interest in seeing any statistics for games where there are no surrenders? It is something that can't really be done with simple database calls, because it is quote hard to sort out the games where noone has surrendered. However, it wouldn't take much to knock up a little program to gather the info and spit it out. I may do this anyway, because I admit I am now intrigued to see what the 'vanilla' statistics are based on our scoring system and draws approach. For example, how does DIAS affect things, and in particular how does it affect the different country performances if you strip out the surrenders.

My main concern though is some of the data is going to be skewed because of the selection criteria. When we look at games with no surrenders, we are not looking at games where surrenders are not possible. Therefore we would be looking at games where players chose not to surrender. So for example, in games where Austria is quickly stomped, the likelihood is in at least a portion of games the Austrian player will just surrender. If the game proceeds without surrenders, it may mean that the selection criteria has filtered out some of the 'poor start' games and only includes games where Austria starts well, possible skewing country performance statistics.

Anyway, do people think it would be interesting to see statistics for surrender-free games?


I would like to see those numbers. But I have a question - does "Surrender" mean "Go into Civil Disorder"? It sounds like it ought to mean "concede to a solo" but I gather from your context above that this is not the case?
Stab you soon!
User avatar
DQ
 
Posts: 127
Joined: 11 Sep 2008, 14:29
Class: Diplomat
All-game rating: (1000)
Timezone: GMT-5

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby jay65536 » 28 Nov 2017, 22:08

Yes, "surrender" means civil disorder.

I would hesitate to draw too many conclusions from such a sample because, as dipsy points out, the non-surrender games are not a random sample, and we don't really know how to control for the specific ways it is non-random.

I would also say, to the previous poster who said "elimination is the cardinal sin", that I don't think this view is accepted at all by top players. The real question, and one that every scoring system tries to answer in its own way, is: how many eliminations should you be willing to take if you can get a solo as well?

For example, in a pure Calhamer point system, if you have a choice between a bunch of 3way draws and a solo followed by a bunch of eliminations, how many eliminations is worth it to take for the one solo? Calhamer points would say that for each solo, you could take 2 eliminations and you'd be equal to where you'd be if you 3way'ed every game.

On the other hand, in the Carnage system, the number of eliminations you should take in exchange for a solo is "infinitely many"! And it doesn't matter what other string of non-solo results you are comparing that to.
jay65536
 
Posts: 55
Joined: 10 Sep 2016, 18:13
Class: Ambassador
Standard rating: 1090
All-game rating: 1096
Timezone: GMT-5

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby super_dipsy » 29 Nov 2017, 08:27

At the risk of confusing,
jay65536 wrote:Yes, "surrender" means civil disorder.
this is not quite accurate.

On Playdip, Surrendering means the player leaving the game. That player may be replaced by someone else joining as that country of course, and in fact we have quite a high rate of pickup of surrendered positions. Civil disorder technically just means the units have not received orders and therefore hold. So missing a turn means your units are in civil disorder, but noone has surrendered. Running stats for games with no surrenders would still include games where units were in civil disorder because of a missed turn.

However, the more I think about it, the more I suspect Jay is right and the sample would be too skewed. Think I'll just go back to getting the mobile orders interface finished ;)
User avatar
super_dipsy
Site Admin
Site Admin
 
Posts: 10892
Joined: 04 Nov 2009, 17:43
Class: Ambassador
Standard rating: (1000)
All-game rating: (956)
Timezone: GMT

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby DQ » 29 Nov 2017, 11:47

super_dipsy wrote:At the risk of confusing,
jay65536 wrote:Yes, "surrender" means civil disorder.
this is not quite accurate.

On Playdip, Surrendering means the player leaving the game.



Aha - back in MY day, kids, we called this "an abandonment." And, ok, too skewed maybe - but obviously any game with a surrender are going to be tainted when it comes to getting a solo.

Speaking again to the question of ratios of solos/non-solo results - I think the key to remember for most top players is that you are bound by the parameters of the event to make that decision. Two eliminations and a solo is going to win 95% of all f2f tournaments - that other 5% being ones where someone else has a solo as well. That being said, three "Top Results" - Small draws, big board tops, what have you - will win 80% of f2f tournaments, while winning the tournament with a combination of non-solo results that include an elimination is very, very hard. Some tournaments have a "drop round" which complicates things, but basically, once you've had an elimination you are on the "solo or bust" train if you want to win.

IME not enough players factor this into their play in later rounds, but that's another topic.
Stab you soon!
User avatar
DQ
 
Posts: 127
Joined: 11 Sep 2008, 14:29
Class: Diplomat
All-game rating: (1000)
Timezone: GMT-5

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby jay65536 » 01 Dec 2017, 22:17

I think I would have to disagree with your assertion that "a solo and two eliminations will win you 95% of all FtF tournaments." It will win some, but not nearly 95%.

In the Dixiecon system, 3 good draws will beat a solo and 2 losses almost for sure, I think. Right?

Also, in any tournament that uses sum of squares, having a score greater than 100 after 3 rounds, with no solo, is uncommon but not that rare. (If you rack up something like 40+/30+/30+, you'll have beaten a solo and 2 eliminations.)

Obviously in a system like Carnage, you win 100% of the time when you have the only solo.

In the GenCon system, if you solo an early round you could easily still lose on the top board.

What other scoring systems are in regular use in North America? (Or around the world, I guess...)
jay65536
 
Posts: 55
Joined: 10 Sep 2016, 18:13
Class: Ambassador
Standard rating: 1090
All-game rating: 1096
Timezone: GMT-5

Re: A (Choose: Great/Pointless/Annoying) Scoring System Thre

Postby DQ » 03 Dec 2017, 18:03

jay65536 wrote:I think I would have to disagree with your assertion that "a solo and two eliminations will win you 95% of all FtF tournaments." It will win some, but not nearly 95%.

In the Dixiecon system, 3 good draws will beat a solo and 2 losses almost for sure, I think. Right?

Also, in any tournament that uses sum of squares, having a score greater than 100 after 3 rounds, with no solo, is uncommon but not that rare. (If you rack up something like 40+/30+/30+, you'll have beaten a solo and 2 eliminations.)

Obviously in a system like Carnage, you win 100% of the time when you have the only solo.

In the GenCon system, if you solo an early round you could easily still lose on the top board.

What other scoring systems are in regular use in North America? (Or around the world, I guess...)


Hmm, looks like you are correct in theory: A Dixie Solo is worth 342 points, a 2WD is 170+(4*SC), a 3WD is 130+(4*SC) = so, heck, two 2WD is enough to beat a solo and two eliminations. I've just never seen it happen. (And there is something wonky going on with Dixie Scoring in WDD) I beat Jon Quarto von Tividar for the World Championship in 1998 by 2 dots with a solo, a 2WD, and an elimination to his Solo, 3WD + elimination. His 3WD was at 17 centers I think, and my 2WD 11? It was close.

I suppose the best example recently of a solo not being good enough was WDC in Chicago in 16, when John Gramilla's solo and three eliminations put him in 8th place in the tournament, and not on the top table. (Through 4 rounds, one dropped)

Peter Yeargin 25.633 52.16 51.042 39.286 168.121
Adam Silverman 41.322 0 43.478 54.144 138.944
David Maletsky 40.878 0 50.35 42.308 133.536
John Gramila 100 0 0 22.378 122.378

I guess the question that we are all asking is "Do we agree that this solo + one other result was not as good as the other results?"

And that's where it becomes what I call "religious" and the thread title calls pointless/annoying, because there aren't any objective standards to point back towards. I'd argue that Peter's two 50+ point results are NOT as good as a solo and an elimination. But what do I base that on? I think its about the "win" - accomplishing the goal set out in the rules. Solo victories can be easy or hard, and so I don't think degree of difficulty can play into it. My line falls somewhere around "The aggregate results of the three players above John do demonstrate a higher level of skill when considering playing the game four times" But ... I find myself also sympathetic to the view that a solo win should bump a player into another "tier" where you are considered to have a higher raw score than any aggregation of non-solo results.
Stab you soon!
User avatar
DQ
 
Posts: 127
Joined: 11 Sep 2008, 14:29
Class: Diplomat
All-game rating: (1000)
Timezone: GMT-5

Previous

Return to Diplomacy Lore

Who is online

Users browsing this forum: No registered users and 1 guest