Chele Nica wrote:

I don't know that it would significantly change the game, but it would give people who are not winning more of a reason to play well till the end. Is that worse than having despondent players give up or prolong the game? 

As for the other two points:

I don't usually try to attack top players early in the game, as that is a good way to get eliminated early on, or at least significantly weakening yourself from retaliatory attacks. And I remember reading in these forums that some people attempt to use this strategy in the current set-up because they want to avoid that person from getting the winner points. 

I'm talking about multiple players using the strategy in concert.  It's one thing to take out the top player to increase your chances of winning, and with the current system there's a logic that suggests that it's good strategy to not be the player wasting resources in that effort, but with the prospect of everyone benefitting, in effect playing as a team against the top player, there would less of a reason for it not to become SOP, with no need to formally request an alliance in certain circumstances.  Of course, I could be wrong ..or right for the wrong reasons, but the meta-point is that in any system where in the course of play, it becomes logical/correct to evaluate the 'external' game and then alter game-play to accommodate those considerations, I believe it is not healthy for the site.

it matters how good you are relative to other players, because you "live" to play another game tomorrow.

..by 'live' you are talking about that 'bigger' game.  I don't want to play that game.  As it is, unranked play is not supported on this site, so in a sense the the problem is unavoidable.  As it is, I am managing to avoid ranked play by playing my own boards privately and in Dev, but everyone else on this site is subject to the pressure and complexities of ranked play.  Introducing added incentives for players to play differently because of external considerations is precisely the road I would like to see the site avoid.  Its just my personal preference, but consider that a large majority of contributors to these forums support the idea of supporting unranked games.

I support unranked games as well, it'd be easier to get comfortable with a board before (or instead of) trying to compete for ranking.

As for ranking everyone, I grew up playing like that. My friends would consult me about the rankings that i was keeping track of after the games, but everyone would still play to win, with the same kinds of strategies that i see on this site, except of course that back then people were negotiating alliances in public, not through private messages. In general, players were not making alliances against a single player, but were trying to win. It's simply very rarely a realistic scenario for everone to ally against one player, and even more so playing fogged multiplayer games, which is the way I play 95% of multiplayer games these days. Playing unfogged multiplayer games leads to ganging up on a player, but generally not the top ranked, just whoever is strongest in that particular game.

In a way, I see this idea similar to what xrayjay wrote on a similar thread, he said the current problem with the CP system as it is, is that someone could be ranked 11 on a hundred boards and not have a CP ranking point. Well, right now, you can have someone who's finished 2nd place on ten different 15 player boards, and s/he might have a negative ranking, whereas another player who won a single 15 player board might be higher ranked, even if that player finished close to the bottom on every other game they played. I would argue that the first player is the stronger player, despite their not coming number 1 on any game so far.

In any case, it doesn't seem realistic to implement retroactive rankings for players who finished 2nd or below, but I would be supportive of that idea if it became feasible down the line.

If we were to do something like that I don't think it should be standard, but it could be a fun option for a board designer to choose for a particular board/scenario.

@Ozyman - How is it coming along?  Any progress?

@anyone else - What are the legal considerations?  On the one hand, Microsoft seems to hold a patent and not allow it on 'commercial' sites, but on the other hand, it looks to be easy to find code in Python out there.  Is this code independently developed ..from scratch .is it freeware? 

M57 wrote:

Awesome!! Can this be set up on a spread sheet with something like 10 players? ..or is it all code and has to be run in a dedicated program?

Looking at the numbers above is fascinating.. after winning just 2 games in a row, #2 is rated higher.  Also, #3 seems to have hit a temporary equilibrium (possibly because confidence in #1 is wavering?).  What's disconcerting is that after 5 games, no one has a rating above 1000.  This is ok if the site only has one board because everything is relative, but..??

I'd be curious to see if the numbers spread out more if you add a fourth player and input the same results.. 1,1,1,2,2,2.  

Yeah - I also thought it was weird at one point everyone is below 1000.  I'm guessing that is an artifact of it being so few games?

Here's the results for 4 player games:

INITIAL SETUP:
Player 1: 1000.0 w/333.0
Player 2: 1000.0 w/333.0
Player 3: 1000.0 w/333.0
Player 4: 1000.0 w/333.0
P1 Wins
Player 1: 1217.253694 w/259.895297034
Player 2: 927.329439025 w/216.640496603
Player 3: 927.632699713 w/216.590806461
Player 4: 927.784167257 w/216.780115676
P1 Wins
Player 1: 1293.09234278 w/225.178502399
Player 2: 909.80507011 w/162.395989427
Player 3: 910.057427404 w/162.303154043
Player 4: 910.170672828 w/162.512979737
P1 Wins
Player 1: 1332.13869665 w/204.76187867
Player 2: 903.073876225 w/132.47191779
Player 3: 903.279967769 w/132.358899303
Player 4: 903.365491611 w/132.56419478
P1 Wins
Player 1: 1356.63692876 w/191.026685959
Player 2: 899.682734021 w/113.658169529
Player 3: 899.856459485 w/113.536910391
Player 4: 899.924360137 w/113.732727579
P2 Wins
Player 1: 1114.27840956 w/140.402494341
Player 2: 965.804421446 w/103.452426338
Player 3: 909.743741406 w/100.181808846
Player 4: 909.74262383 w/100.372057416
P2 Wins
Player 1: 1031.32825882 w/115.986746757
Player 2: 1007.68094648 w/96.1156241203
Player 3: 911.239178626 w/90.3385719578
Player 4: 911.231009927 w/90.5230365401
P2 Wins
Player 1: 990.73771949 w/100.96018007
Player 2: 1037.38100383 w/90.5455323053
Player 3: 910.431608002 w/82.8066715602
Player 4: 910.429298267 w/82.9848366332

btw - results are exactly the same whether I have all the losers tied for 2nd or tied for 3rd.

Another weird thing - why are the various players scores not exactly equal.  For example after game 1, should 2,3,4 all have exactly the same score - not just mostly the same?

tom wrote:

Some good points about impact on CP etc - it could be parallel run so we can model the impact on ranking tables, CP etc.

If it helps Ozy I can pass on some ranking data - as the dataset can get pretty large this could be a subset of a more popular board or the entirety of a less popular board's games.

Definitely would need to run in parallel to compare.

Sure! Send me some data.  XML, CSV, json?  Something like that?  Send it to the email address I'm signed up with, or let me know if you need me to PM you that.

Thanks for the interesting discussion M57 & Chele Nica.  It's a tough call, I think I'm slightly more convinced by Chele Nica's argument, but maybe best would be as Korrun states, and let it be a board (or host option).

One other thing to think about - TS also has functionality to compute the 'fairness' of a match.  It's something we've talked about before, but I'd like the option to mark a game as 'fair', and only players that are similar in skill to me could play it.   For me this would be almost as good as unranked games, because when I'm learning a board, even though it would count, at least it would be other newbs or not so great players that I'm learning against.

Rank by elimination order? Nah. You'd get pernicious impacts in the end-game as the losers fight to be the last loser. Make the 3-man endgame worse than it already is.

Ozyman wrote:

btw - results are exactly the same whether I have all the losers tied for 2nd or tied for 3rd.

Another weird thing - why are the various players scores not exactly equal.  For example after game 1, should 2,3,4 all have exactly the same score - not just mostly the same?

Yeah I notice that in your first example. I'll bet it has something to do with order of execution.  More importantly, it appears to take care of itself over time, note how the differential looks to vacillate in a fashion that causes alternation between players 3 and 4, settling down over time.

3 - 4 = rounding

-0.15
-0.11
-0.07
-0.08
+0.00
+0.08
+0.00

M57 wrote:

@anyone else - What are the legal considerations?  On the one hand, Microsoft seems to hold a patent and not allow it on 'commercial' sites, but on the other hand, it looks to be easy to find code in Python out there.  Is this code independently developed ..from scratch .is it freeware? 

Just make sure we don't use any rounded rectangles. Then we'll be in heaps of trouble.

Ozyman wrote:

Thanks for the interesting discussion M57 & Chele Nica.  It's a tough call, I think I'm slightly more convinced by Chele Nica's argument, but maybe best would be as Korrun states, and let it be a board (or host option).

One other thing to think about - TS also has functionality to compute the 'fairness' of a match.  It's something we've talked about before, but I'd like the option to mark a game as 'fair', and only players that are similar in skill to me could play it.   For me this would be almost as good as unranked games, because when I'm learning a board, even though it would count, at least it would be other newbs or not so great players that I'm learning against.

There's nothing as good as unranked games :) I don't want to be ranked on boards until and unless I'm good and ready ..and sober, etc.., and until that time I don't want CPs or ICPs or whatever they are hanging over my head.   For me, it's about playing for fun, and the accessibility of the site.

Anyway, it looks as if, because players who have won games can so easily fall in the ranks just by losing a few more recent games, perhaps a number of games (10?) need to be played before a player's score is 'posted,' otherwise a player can come in, win one game, and be ahead of the crowd.  It just kind of flies in the face of the current GR paradigm.

For the record, the ranking i used as a kid went like this:

Two player game: winner gets 1 point, loser 0 points

Three playee game: winner gets 2 points, 2nd place 1 point, 3rd place 0 points

Etc.

You had to play 10 games to be part of official rankings

Your ranking relative to other players was based on actual points divided by possible points times 100. So if you won a five player game (4 points out of 4) and lost a 2 player game (0 points out of 1), you had 4 of 5 possible points, or .80 * 100 = 80

smoke wrote:

Rank by elimination order? Nah. You'd get pernicious impacts in the end-game as the losers fight to be the last loser. Make the 3-man endgame worse than it already is.

+1

That 3-man endgame almost always comes into play in no-fog games, and that's one big reason why I avoid them. Rank by elimination would not change any of that. If you want to avoid stalemates, go for medium to total fog

itsnotatumor wrote:

smoke wrote:

Rank by elimination order? Nah. You'd get pernicious impacts in the end-game as the losers fight to be the last loser. Make the 3-man endgame worse than it already is.

+1

+1

tom wrote:

itsnotatumor wrote:

smoke wrote:

Rank by elimination order? Nah. You'd get pernicious impacts in the end-game as the losers fight to be the last loser. Make the 3-man endgame worse than it already is.

+1

+1

Not fair ;) Tom gets 1,000,000 votes. Not that I'll complain in this case.  Actually, it would probably do a lot to end crab games, but for all the wrong reasons.

Well, I'm definitely outvoted now :)

Time for an MB style too much booze post:

ARGGHHHHHHHH.......Did we not agree on the other thread "ICPs" were the way to go?!?!?! Recently there was an offensive thread (nicely quickly obliterated) that insinuated that all of us either had darker skin pigmentation or perhaps were interested in other gentlemen. You know how I know that wasn't accurate? Black dudes and gay fellows are way too cool to spend 2 years debating a scoring system for online Risk! Lets implement something!

Xrayjay wrote:

Time for an MB style too much booze post:

ARGGHHHHHHHH.......Did we not agree on the other thread "ICPs" were the way to go?!?!?! Recently there was an offensive thread (nicely quickly obliterated) that insinuated that all of us either had darker skin pigmentation or perhaps were interested in other gentlemen. You know how I know that wasn't accurate? Black dudes and gay fellows are way too cool to spend 2 years debating a scoring system for online Risk! Lets implement something!

Given the current GR arrangement, there's consensus around ICPs.  But the CP system sits on top of GR, and there's also consensus that GR is lacking.  TrueSkill, if implemented, would replace GR, which in turn would have a significant impact on ANY CP system that's in place.  Better to resolve the GR/Trueskill issue first.  BTW, if a TS-like system is brought on board, the current CP system will surely make no sense at all, and will need immediate attention, so really, upgrading GR to TS will necessarily entail a simultaneous upgrade of the CP system.