Adjustments to scores/ranking/records

This is where we inform about development updates.

Adjustments to scores/ranking/records

Postby super_dipsy » 08 Jan 2011, 18:02

Over the past month or so, a number of problems with the scoring/ranking archived records have been identified by some very helpful work carried out by a number of site members. Maucat and Sinnybee in particular deserve special mentions for their assistance. These issues have now been resolved. Some players may have already noticed these corrections starting to adjust their reccords.

The two major issues concern playing/scoring records for games where cheating was identified, and the game statistics for specific variants available from the Statistics tab.

The most important one for many people will be the cheating situation. When a cheat is unmasked, Rick reviews all the games in which that person has played, and if he determines that the game has been majorly influenced by cheating he then UNRANKS it. This is because if the game has been sufficiently tainted by the cheating, then it should not have any effect on anyone's scores because any associated points have been compromised. However, it has been discovered that in fact although the game was indeed being unranked, and showed as unranked when looking at game histories, nothing was being done to actually remove the associated scores from the records. This is why some cheats still seem to have high scores and rankings, although since we introduced the 3-month cut-off they do drop of the tables eventually. What we have done is to firstly fix the issue so that from now on unranking a game will delete all points records associated with it, but secondly to retrospectively delete all the points associated with any previously unranked game.

The effect of this on scores and ranking should be minor except for the cheats themselves. For innocent parties, it is quite likely that these unranked games reflect games that they lost (the cheats probably won) and therefore deleting the associated scores will simply remove the odd losses from innocent players' records, giving a small improvement in fading echoes terms. However, do not expect to see an instant change to your score/record. The fix is of the self-healing type, and your score will only be adjusted when you complete your next game. It is at this point that the fading echoes scores are recalculated and this is when the new changes will be picked up.

The second, much less serious problem was that while overall game records for each player were OK, when a player looked at their score and ranking for a particular variant (eg Fleet Rome, Chaos, etc) although the games were all listed with the correct scores, the variant total points and ranking did not include many of these games. So you might see that your total for fleet rome was 12, but in your fleet rome game list you would see that you might have 3 solos for 12 each so the total should have been 36. This has now been fixed too, but similar to the cheating case, the fix is self-healing. Therefore, you will see no change until you finish the next game of a particular variant. So in my example, if your fleet rome total/ranking is wrong, it will only be corrected when you finish your next fleet rome game. However, let me reiterate, this only affects scoring stats selected on a specific variant basis.
User avatar
super_dipsy
Premium Member
 
Posts: 12194
Joined: 04 Nov 2009, 17:43
Class: Ambassador
Standard rating: (1000)
All-game rating: (931)
Timezone: GMT

Re: Adjustments to scores/ranking/records

Postby sinnybee » 08 Jan 2011, 19:40

It sometimes can take a lot of work to fix a bug; good job super_dipsy.
Gold Classicist since 1-11-11
FT Asst GM of 35 player WWIV Aug 2011-Feb 2012
#1 ranked player of playdip early 2013
4th highest forum karma count at Apr 2013 ending (behind Craw, Dipsy, and Rick)
Tournament Director of the 31 game PDVT Feb-Dec 2014, the first playdip tourney with over 100 sign-ups
User avatar
sinnybee
 
Posts: 6097
Joined: 03 Sep 2010, 07:01
Class: Ambassador
Standard rating: (1332)
All-game rating: (1467)
Timezone: GMT-7

Re: Adjustments to scores/ranking/records

Postby dontejones » 09 Jan 2011, 18:47

I'm relatively new to this site, so forgive me if it's presumptuous to post a suggestion regarding the rankings system. But I believe that the present system is flawed because it rewards absolute point totals, without regard to the number of games that it took for the player in question to achieve such point totals. Suppose, for example, that one player has 250 points after 40 games, while another has 210 points after just 20 games. Clearly, the second player is playing at a much higher level than the first one. Indeed, he's averaging more than 10 points per result, whereas the first player is averaging only 6 points per result. It seems to me that a much more accurate way of ranking players is to calculate their total points in the same way as always (i.e., taking the weighted value of the results of their last 50 games, maximum), then DIVIDE the total by the number of games played to arrive at such totals. (To qualify for a ranking, a player should be required to have completed a minimum number of games--say, 5 or 10--to ensure a representative sample.) That will produce a weighted average per game, which is a far more telling statistic. All players should then be ranked according to their weighted averages, not according to their absolute point totals. At the very least, a ranking of weighted averages could be used as a supplement to the present ranking system. Just as in baseball there are rankings according to number of hits AND according to batting averages. (Batting Average is universally regarded as a more meaningful statistic than Hits, assuming a certain minimum number of at-bats.)

Any thoughts on this?

Many thanks to everybody who has had a role in developing this site. I've only played a few games here, but it strikes me as very well-run, and it's been a pleasure to join in the fun.
dontejones
 
Posts: 97
Joined: 13 Nov 2010, 19:23
Class: Diplomat
All-game rating: (1000)
Timezone: GMT

Re: Adjustments to scores/ranking/records

Postby Waterice man » 09 Jan 2011, 19:57

dontejones wrote:I'm relatively new to this site, so forgive me if it's presumptuous to post a suggestion regarding the rankings system. But I believe that the present system is flawed because it rewards absolute point totals, without regard to the number of games that it took for the player in question to achieve such point totals. Suppose, for example, that one player has 250 points after 40 games, while another has 210 points after just 20 games. Clearly, the second player is playing at a much higher level than the first one. Indeed, he's averaging more than 10 points per result, whereas the first player is averaging only 6 points per result. It seems to me that a much more accurate way of ranking players is to calculate their total points in the same way as always (i.e., taking the weighted value of the results of their last 50 games, maximum), then DIVIDE the total by the number of games played to arrive at such totals. (To qualify for a ranking, a player should be required to have completed a minimum number of games--say, 5 or 10--to ensure a representative sample.) That will produce a weighted average per game, which is a far more telling statistic. All players should then be ranked according to their weighted averages, not according to their absolute point totals. At the very least, a ranking of weighted averages could be used as a supplement to the present ranking system. Just as in baseball there are rankings according to number of hits AND according to batting averages. (Batting Average is universally regarded as a more meaningful statistic than Hits, assuming a certain minimum number of at-bats.)

Any thoughts on this?

Many thanks to everybody who has had a role in developing this site. I've only played a few games here, but it strikes me as very well-run, and it's been a pleasure to join in the fun.


That's what fading echoes does. The problem with your system is that if someone plays one game in which they solo, they will go directly to the top of the leaderboard, and there may well be cases of people openning an account just to play one game with which they are aiming for a solo. It would work if everybody were honest and played several games, but in reality, neither of those are true.
Why did the chicken cross the Möbius strip?

The word 'surrender' derives from old French
User avatar
Waterice man
 
Posts: 4677
Joined: 31 Dec 2008, 18:36
Location: Britain. Chances are, you used to be in our empire
Class: Diplomat
All-game rating: (1000)
Timezone: GMT

Re: Adjustments to scores/ranking/records

Postby dontejones » 09 Jan 2011, 20:34

That's precisely why I said that to qualify for a ranking, a player would have to have played a minimum of, say, 10 games. Or 15, or whatever. (How dare you suggest that my system had a "flaw"!)

Fading echoes is nice but it's really just a half-measure. Or maybe even a quarter measure. The truth is that the system places a huge amount of weight on the sheer number of games played. The only way to have a chance to really compete on the leader board is to play 45 or 50 games. But I don't see why someone who has played, say, 25 games should effectively be excluded from consideration.

Regardless, I certainly see no harm in having two rankings--one on the basis of average (again--with a minimum number of games required for somebody to qualify) and the other on the basis of absolute points.
dontejones
 
Posts: 97
Joined: 13 Nov 2010, 19:23
Class: Diplomat
All-game rating: (1000)
Timezone: GMT

Re: Adjustments to scores/ranking/records

Postby rick.leeds » 09 Jan 2011, 23:13

Uh-oh... there are enough debates in Suggestions about the rankings without opening one here. But, it isn't actually true that you need to play lots of games to move up the rankings; my highest ranking was with MUCH fewer games.

Anyway this is an announcement thread not a place to debate it, other than perhaps to ask questions about it.
World Diplomacy Forum.
Online Resources editor at the Diplomatic Pouch.
Don't let the stepladder get you. Watch where you're stepping. ANY step could be a doozy.
User avatar
rick.leeds
 
Posts: 8360
Joined: 11 Jan 2009, 04:40
Location: Wherever I am, I'm scratching my head.
Class: Diplomat
Standard rating: (1158)
All-game rating: (1070)
Timezone: GMT


Return to Developments

Who is online

Users browsing this forum: No registered users and 4 guests