home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!stanford.edu!agate!darkstar.UCSC.EDU!cats.ucsc.edu!jonathan
- From: jonathan@cats.ucsc.edu (Jonathan Cano)
- Newsgroups: rec.games.go
- Subject: Re: Internet Go Server Status.
- Date: 27 Jan 1993 03:47:25 GMT
- Organization: University of California, Santa Cruz
- Lines: 56
- Message-ID: <1k50kdINNd9p@darkstar.UCSC.EDU>
- References: <1993Jan21.152030.15620@fy.chalmers.se> <1jncquINN9h6@terminator.rs.itd.umich.edu> <1993Jan25.133008.18496@asns.tredydev.Unisys.COM> <1993Jan25.234742.8412@kpc.com>
- NNTP-Posting-Host: si.ucsc.edu
-
-
- hollasch@kpc.com (Steve Hollasch) writes:
-
- >rjs@citi.umich.edu (Robert J. Skrobola) writes:
- >> ... a resign win over someone 4 stones weaker (even game), should not be
- >> the same as a 5 pt win over someone 4 stones weaker (even game). The
- >> second case is something that should affect your rating negatively, and
- >> the opponents positively. The first, it isn't clear (to me)..
-
- >mdobbins@asns.tredydev.Unisys.COM (Michael Dobbins) writes:
- >| The rating algorithm does not consider how badly (number of points or
- >| resign) someone was beaten, just who won. A half point victory counts
- >| just as much as a 40 point victory.
-
- > At first this seemed to be a weakness in the ratings algorithm. After
- >all, if I give nine stones to an equally ranked opponent and lose by only 1
- >stone, it seems like this should bump my rank simply by virtue of huge
- >handicap and the fact that I "almost made it".
-
- > On reflection, though, it occurred to me that this is the sort of
- >information that a rating algorithm should NOT try to account for.
-
- Exactly! When you are playing a game the object is to get at *more*
- points than your opponent. Who is better: Ed who beat Fred by 40
- points or Fred who beat Ed by 1 point 40 times out of the last 41
- games? When a player is ahead, then the smart thing to do is 'coast'
- and play the safe moves.
-
- Is it smart to settle for a 1/2 point (or a 2 or 3 point) loss when by
- playing risky moves in a ko fight you have a chance of losing big or,
- possibly, winning?
-
- According to traditional rules, a 1pt loss == 100pt loss. To change
- this would be to create a new variation of Go. This is not bad, but
- I, personally, prefer the traditional values that say forty 1pt wins and
- one 40pt loss are better to have than one 40pt win and forty 1pt losses.
-
-
-
- > Fighter : 9.78
- > Dumpling Former: 1.11
- > Escaper : -7.43
- > Follower : 2.78
- > Peeper : 5.52
- > Resigner : ...
-
- > Heh heh heh. B^)
-
- I like these!
-
- --
- ============================================
- | Jonathan Cano | // _ |
- | jonathan@cats.ucsc.edu | \\ // (_) |
- | 10kyu (+- 2kyu) | \X/ |< |
- ============================================
-