Rating History?

[spincycle]

Is there a way to get the history of my matches, and specifically my before-and-after rating, directly from the FIBS server? I know my client saves them as MAT files but that doesn't include my rating.

The reason is because I logged in today and my rating was not what I expected it to be, based on my (sometimes faulty) memory. I'd like to find out if the issue is my memory or if there is a bug that affected my rating.

Thanks for any suggestions.

[diane]

The most likely explanation is that you had a saved match with a bot which was adjudicated.  You would lose points for that, as the bots will always resume, and make themselves available for it.  If it doesn't happen - that is down to you.

You also lose a disproportionate amount of points, to deter players from dropping bots.

No is the answer to your question.

If you use javafibs, that will log a graph of your rating, as well as log matches, I am not aware of any other interface doing that.

[spincycle]

Thanks for your response, Diane, that's probably what it was.

Now for more general discussion on this: Adjudication of bot matches won't deter players from dropping if they don't know about it. It's not in the FAQ and I don't see it mentioned anywhere on the site or in the forums - I found one tweet from Patti back in October.

I'm all in favor of automatic adjudication of saved games, but why just for bots?  If it were a consistent policy across all matches it would essentially eliminate droppers from FIBS - that would be fantastic. I'm a software engineer by trade and I would be happy to help Patti or whoever do the grunt work involved in setting it up.

[sixty_something]

there is also the personal ratings graph in JavaFIBS .. while it doesn't provide the detailed tracking you are suggesting,, you can often infer match by match changes when looking at the last week or fortnight .. other time periods include month, year, five year, and decade .. this feature is available in JavaFIBS from the Tools/Statistics menu and the Rating tab

[diane]

Bots are the only single case where it can be assumed with real confidence that the human player was at fault.
The bots dont drop, they make themselves available for resuming matches, and they invite players to resume as a priority over starting new matches. This can never be guaranteed so easily with human v human matches.

The grunt work you mention would involve changing fibs code I think, and there is a reluctance to do that, as it may create other unforeseen problems.


As for notification about bot match adjudication, yes it is hard to find, and it would be better if it were more publicised.

[diane]

Adjudication of bot matches won't deter players from dropping if they don't know about it.

Even with poor publicity, players soon find out about this...by one of the many realtively painless routes...and ultimately by the very painful one - and it is showing significant improvement with regard to bot dropping. It can actually be measured that they have fewer saved games and higher ratings 

[dorbel]

Saved matches with bots aren't adjudicated in any sense of the word. What happens is that every so often Patti runs the botdropper script. If you have a saved match with a bot, it will disappear. You will be "fined" a number of rating points for each saved that you have. I seem to remember that it is 15 rating points. This is crude but effective and yes it could be better publicised.
I think you will find that Patti also resets the bots' ratings now and again, adding an arbitrary amount to them to compensate for all the times that they have been dropped. I don't know how much she adds, but it looks like 150 points or so.
Adjudication is a different animal. How it works is that you view the match at the point at which it was terminated and settle the match. If one side is 100% to win, they get 100% of the points. If it's 90%, they get 90% and so on. You use an analysis program for this, Gnu or Snowie for example. It is time consuming for the admin of course, for precisely nil return, so it isn't going to happen. It doesnt even add much to the pleasure of playing either. It will probably make dropping even more likely and it works against the more skilful player, as the weaker player will be credited by the bot with equal playing ability for the purposes of settling the match.
The way to protect your rating from droppers is to avoid playing them in the first place. All the tools are in place to do this and learning how to do it is a rite of passage for newbies. It isn't very hard and once the skill is mastered, you will hardly be dropped once a year. Happy gammoning.

[Patti]

The matches are adjudicated in favor of the bot, and the human player gets a penalty.  As of right now, the bot gains either three or four points (I don't remember which) per match, and the human loses 15 points per match.

This was done to address the problem of people thinking that bots were free points.  People would play bots and only finish when they were going to win, then wait for the matches to expire and do the same thing.

[spincycle]

Adjudication is a different animal. How it works is that you view the match at the point at which it was terminated and settle the match. If one side is 100% to win, they get 100% of the points. If it's 90%, they get 90% and so on. You use an analysis program for this, Gnu or Snowie for example. It is time consuming for the admin of course, for precisely nil return, so it isn't going to happen.

Yes, what you're describing is what I meant by adjudication (play65 does this at the instant of the drop) but I wasn't suggesting it be a manual process. I was suggesting some sort of automated thing that happens after N attempts to resume, or at the time the saved games disappear, whatever is fair ... but as Diane says this would be a change to the code so if that's not an option then my idea is a non-starter and I'll go back to playing backgammon. Thanks for everyone's comments.

[Patti]

Actually, adjudication is just judgement -- the word doesn't imply any particular mechanism for passing that judgment.