Wednesday, 24th January 2018
Puzzles Solved Yesterday: 123
Home | Register | Login | Current Puzzle | Archives | Leaderboard | Forum | Tutorial | FAQ
Forum Index
 
ranking wrong solvers
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2006.09.25 15:42:59
Recently I got doubly addicted, to the regular puzzles and to the
wrongs. There is just one thing that bothers me about the ranking of
the wrongs (I agree, I am a nitpicker).

The principle of the ranking is that one more change is equally
`bad' as one minute of thinking. However if we compare the scores of
someone who made 20 changes in 57 seconds and another who made 19
changes in 62 seconds, in the current ranking the first wins,
because both `counts' are 20 and the first one is faster. I have the
opinion that the latter should win, since the fact that she made one
change less outweights her extra five seconds.

The ranking can be slightly adapted to let the second player win.
Just use the sum of the number of changes and the time. So in the
example the first person would score of 20+0'57''=20'57'' and the
second would score 19'+1'02''=20'02'', which is better.

I think that ranking in this way better comports with the principle
of 1 change = 1 minute. Furthermore, I don't like the discontinuity
of the current ranking.  It causes an unnecessary luck factor
(finishing just below or above a full minute).

Would anyone like to comment on this?
foilman
Kwon-Tom Admin
Puzzles: 1718
Best Total: 24m 8s
Posted - 2006.09.25 16:07:23
Hmmm... I quite like that idea!

And it would be fairly easy to go back through the database and fix all the existing scores to use the new method. Just a bit of work involved during which things might break a bit... I might think about doing it when I get some time.
foilman
Kwon-Tom Admin
Puzzles: 1718
Best Total: 24m 8s
Posted - 2006.09.28 07:56:25
Ok, I had a bit of time to look at this last night, and I've basically done it but it's not yet visible on the site for anyone other than myself.

How it works now is it takes your existing "change count" (for example, 20) and the seconds part of the elapsed time (for example, 57s) and combines them for a "wrong score" - for example 20 (+57s). So now it's basically the change counts that are used to rank people, with the number of seconds past the minute being used to rank people with the same number of changes. Because the minutes are already taken into account in the change count part (the count goes up for every elapsed minute) they're actually not necessary for the "time taken" part.

I might switch it on officially for everyone later but be aware that there are a few oddities in existing scores & rankings that might not make total sense until a full week has rolled by.
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2006.09.28 16:30:16
That is very nice, and quick. It is funny that on Tuesday as well as on Wednesday it depends on the method who is the winner. It might be the case by the way that people still like to know the number of changes and minutes their count is composed of.
foilman
Kwon-Tom Admin
Puzzles: 1718
Best Total: 24m 8s
Posted - 2006.09.29 09:12:48
Ok, it's gone live officially now... post any problems you see here, but as I'm away for a few days don't expect an immediate fix.
Para
Kwon-Tom Obsessive
Puzzles: 1911
Best Total: 19m 28s
Posted - 2006.09.29 15:31:22
Too bad though you don't see the number of changes the top scorer had when you finish your wrong. It only shows the combined score.
Just started here. Nice for me to compare my number of changes to the best. The speed will come later.
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2006.09.29 22:14:03
That is what I meant when I said that it might be interesting to know how scores are composed of minutes and numbers of changes. Just like before, this can be easily deduced by subtracting somebody's minutes from his score in the daily wrong results. There is no loss of information.

By the way, you've got an excellent number of puzzles/best total ratio for someone who is not interested in speed yet!
Last edited by chairman - 2006.09.29 22:52:21
Para
Kwon-Tom Obsessive
Puzzles: 1911
Best Total: 19m 28s
Posted - 2006.09.30 08:21:38
Yes it is possible for the new puzzles. Don't have a problem with those. But not if you try and solve puzzles from the archives. There you only get the score (full minutes plus moves) with the added seconds. That is where my problem lies.
I try as fast as possible. But i am not completely unformiliar with these puzzles, but i like it better to solve puzzles on paper.
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2006.09.30 10:22:00
Ah.. I now see what you mean. Indeed, it is interesting to know the pure change count of the winner for archived puzzles.
rg
Kwon-Tom Addict
Puzzles: 256
Best Total: 50m 51s
Posted - 2006.10.02 15:08:32
Another suggestion for the "wrong" puzzles. Can you highlight the changed number? I must have spent two minutes looking for the number and 20 seconds solving the puzzle today. Even if it was just shaded or something, that would help.
foilman
Kwon-Tom Admin
Puzzles: 1718
Best Total: 24m 8s
Posted - 2006.10.02 15:11:26
Quote:
Originally Posted by rg
Another suggestion for the "wrong" puzzles. Can you highlight the changed number? I must have spent two minutes looking for the number and 20 seconds solving the puzzle today. Even if it was just shaded or something, that would help.
Think of solving a wrong as a two-part puzzle - part one is finding the change, and part two is fixing it...
Jankonyex
Kwon-Tom Obsessive
Puzzles: 3729
Best Total: 9m 35s
Posted - 2006.10.05 14:29:35
Kwon-Tom Wrong Results for Wednesday
1.  PuzzleLover  2m 28s  48 (+28s)
2.  gadget1903  2m 46s  48 (+46s)
3.  Jankonyex  1m 50s  48 (+50s)
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2006.10.05 14:52:17
Quote:
Originally Posted by jankonyex
Kwon-Tom Wrong Results for Wednesday
1.  PuzzleLover  2m 28s  48 (+28s)
2.  gadget1903  2m 46s  48 (+46s)
3.  Jankonyex  1m 50s  48 (+50s)

This is the third time that the new method changes the winner and the first time (as far as I can remember) that the winner is (likely to be) PuzzleLover. Congratulations!
Last edited by chairman - 2006.10.05 14:52:45
chairman
Kwon-Tom Obsessive
Puzzles: 1395
Best Total: 17m 32s
Posted - 2006.10.05 20:20:02
Quote:
Originally Posted by para
Too bad though you don't see the number of changes the top scorer had when you finish your wrong.

Settled! Thanks,
Para
Kwon-Tom Obsessive
Puzzles: 1911
Best Total: 19m 28s
Posted - 2006.10.05 21:57:23
Quote:
Originally Posted by chairman
Settled! Thanks,

Still keeping the calculating in i see
Jankonyex
Kwon-Tom Obsessive
Puzzles: 3729
Best Total: 9m 35s
Posted - 2006.10.07 08:03:20
Quote:
Originally Posted by jankonyex
Kwon-Tom Wrong Results for Wednesday
1.  PuzzleLover  2m 28s  48 (+28s)
2.  gadget1903  2m 46s  48 (+46s)
3.  Jankonyex  1m 50s  48 (+50s)
It means if two people use different time but same counts to finish the wrong,
e.g.
"A" 1m 59s 36 (+59s)
"B" 2m 00s 36 (+0s)
The winner will become "B" but not "A" even though "A" is smarter?
Last edited by Jankonyex - 2006.10.07 08:03:46
foilman
Kwon-Tom Admin
Puzzles: 1718
Best Total: 24m 8s
Posted - 2006.10.07 08:34:30
Quote:
Originally Posted by jankonyex
It means if two people use different time but same counts to finish the wrong,
e.g.
"A" 1m 59s 36 (+59s)
"B" 2m 00s 36 (+0s)
The winner will become "B" but not "A" even though "A" is smarter?
Ah, but you're forgetting that your count goes up by one for every minute that you take, so in your example even though "B" had a score of 36, he took over 2 minutes to achieve it, which means he made only 34 changes, whereas "A" must have made 35 changes. So "B" is the smarter one, just one second slower...
prj
Kwon-Tom Obsessive
Puzzles: 2356
Best Total: 18m 20s
Posted - 2006.10.07 18:49:59
I think the ranking would be less confusing if it showed the time, the pure actual change count, and then the combined score which is used for ranking.

Forum Index