FIBS: "non-match play" please
David Lloyd Mcadams
I would appreciate if a new feature could be added to FIBS. Rather than
enter into a match with a pre-specified limit, I would prefer to play with
my opponent until we chose to stop. Then, FIBS could adjust our ratings
by taking into account the number of games won by each player before we
stopped.
We all know how lame the ends of matches with pre-specified limits can be.
In a 3-pointer, for instance, if my opponent is ahead 2-1, then I will
double immediately no matter what. The cube loses its magic. The
"non-match play" that I propose would be more akin to the sorts of play
that I enjoy with friends at home and with strangers on the street. I'm
sure others feel the same way.
To illustrate how the ratings adjustment would work, consider an example.
Suppose that hotrod and I have decided to call it quits after a few games
and the point total is
david_mcadams 2
hotrod 4
At that point, our ratings would be adjusted as if we had played six
one-pointers, out of which I won two.
David
PS If some feature along these lines could be added, I think FIBS would be
even more of a draw to more players. For example, in my own experience,
sometimes during a short lunch break I don't log in precisely because such
a feature is missing. At such times, I would like to play just a single
game but with an active doubling cube. And the only way to get an active
cube is to play a match that will take too long.
__________________________________
David McAdams
mca...@leland.stanford.edu
http://www-leland.stanford.edu/~mcadams
"My name in print ... things are gonna start happening to me now ..."
- The Jerk
David Lloyd Mcadams
On Sun, 13 Apr 1997, cruelshoes wrote:
> I'm not totally sure I understand this 'non-match play' term.
> If you just want to play for the fun of it, there's always
> unlimited matches.
> The score won't affect your rating.
>
I wish I could find people willing to play unlimited!! The problem for me
is that most people seem to care about their rating (or else seem to care
about a sense of closure to their games). An unlimited game would work
with an opponent who I know well and who I play frequently. Over FIBS,
however, I don't have any regular opponents, I'm always inviting
strangers. Believe me, I've tried to convince people to play unlimited,
but it doesn't fly (at least not for me).
The point of my request is not that there is any glaring weakness in the
current features of FIBS. My point is just that I believe there is a
sizable (and potential) constituency who would prefer a new feature that
would allow them to play exactly as long as they'd like without having to
play unlimited. My problem with unlimited is that no one wants to play
unlimited. And the general problem with unlimited (at least as far as I
can tell from the comments of all the people who have declined to join me
in an unlimited game) is that it is so long.
Let me explain. Unlimited play seems an excellent way to allow regular
opponents to play without worrying about limits and, over the course of a
99 point match, to discover who is really better. There are other people,
however, who don't like this sense of length or who don't like the idea of
knowing so for certain whether another player is better than they are.
They won't play unlimited, but I suspect they would accept what I call
"non-match play". Here, we play until we choose to stop and that's it.
There is an end in sight (after any game, we can quit) and we don't play
long enough for any irrevocable conclusions to be reached.
Given then that we agree that some new feature (call it "non-match play")
would be worthwhile to draw in new players to FIBS, some new feature that
allows them to play a match as long as they please without limits but not
an imposing "unlimited match"; and given that these players will often be
concerned with their FIBS rating, the question arises "How to evaluate the
rating?" Honestly I don't care -- whatever system is devised, I feel
confident that enough people will be willing to play under it that I'll
enjoy myself more on FIBS. I anticipate, however, that my particular
suggestion for adjusting ratings will be (and should be) criticized.
I proposed that if you and I decide to quit playing at a score
cruelshoes 4
david_mcadams 2
then our ratings would change as if we had played six one-pointers, out of
which I won two. Now, that system puts a decided disadvantage on the side
of the more experienced player. For as it stands on FIBS, the winner of a
k-point match gets credit for having won k points and having lost none.
If a very good player defeated a very poor player but only narrowly, then
under my suggestion the good player's rating would actually go down!!
The alternative, however, seems awkward. If the person who is currently
winning when we stop gets all the points, then I will never feel
comfortable stopping when I'm ahead (it wouldn't be polite). Another
alternative would be to adjust ratings according to the difference in the
scores when players stop, but this too adds distortions to the situation.
I would say, "Use any system you like. The people who will use this new
feature, anyway, we don't care. We're just looking for a way to play a
single game with an active doubling cube. We're just looking for a way to
sneak a little bit of our favorite game into a tight lunch break. We just
want to play more backgammon, at every opportunity."
-david
Richard McIntosh
David Lloyd Mcadams wrote:
>
[... elision ...]
> For example, in my own experience,
> sometimes during a short lunch break I don't log in precisely because such
> a feature is missing. At such times, I would like to play just a single
> game but with an active doubling cube. And the only way to get an active
> cube is to play a match that will take too long.
Or you could play unlimited, then resign the appropriate number
of points (if you're really that concerned about your rating).
You could just play the game unlimited, without worrying about
your rating -- lots of us do just that.
Regards,
Richard
Greycat Sharpclaw
Meow, all...
There is an allegation that David Lloyd Mcadams
<mca...@leland.Stanford.EDU> wrote:
snip...
>The point of my request is not that there is any glaring weakness in the
>current features of FIBS. My point is just that I believe there is a
>sizable (and potential) constituency who would prefer a new feature that
>would allow them to play exactly as long as they'd like without having to
>play unlimited.
snip...
>Given then that we agree that some new feature (call it "non-match play")
>would be worthwhile to draw in new players to FIBS, some new feature that
>allows them to play a match as long as they please without limits but not
>an imposing "unlimited match"; and given that these players will often be
>concerned with their FIBS rating, the question arises "How to evaluate the
>rating?"
snip...
Mt two cents worth about the rating.
If this is implemented (and I expect thats a *big* if), do _not_
integrate the results into the current ratings. Rather, produce a
second rating for "point" play. Using the same rating system for both
would not be good; what you're measuring is too dissimilar.
Many game servers have two ratings -- typically for slow and "blitz"
(extremely fast) games in those strategy games that use a clock. So
multiple ratings is not inherently bad.
Of course, I don't know how hard it would be to retrofit this into
FIBS, not being familiar with the code. If the modulization of the
code is right, it should be practical. If the change cannot be
isolated to a limited number of modules, then it may not be.
Oh, and in this system, we don't need "matches" per see at all. Just
get a sequence started, and play until one player doesn't join.
Greycat
Gre...@tribeca.ios.com
Does anyone have any spare tunafish??
bob koca
David Lloyd Mcadams <mca...@leland.Stanford.EDU> wrote:
>To illustrate how the ratings adjustment would work, consider an example.
>Suppose that hotrod and I have decided to call it quits after a few games
>and the point total is
> david_mcadams 2
> hotrod 4
>At that point, our ratings would be adjusted as if we had played six
>one-pointers, out of which I won two.
>David
Suppose one is interested in the rating. (Of course one could play
unlimited games if they desire money games and weren't interested in
having it affect ratings). Your scoring system has at least two very
big problems.
The first problem is that the strategies which are ratings points
optimal are not the same as regular money play backgammon.
Here is one example: Suppose I am playing an opponent who is say
rated 200 points lower. I reach a position in which I have a checker
on my 5 point and my 2 point, my opponent has a checker on his ace
point. Being the last roll of the game it is an obvious money double
take as I am the favorite(19/36.) However, under your system it is
not a double for me if I worry about my rating. The reason is I need
to win more than 19/36 of hte points in order to break even.
The second problem is that one game could have too big an effect
on the rating. For the ratings system to work it is important that
players are near their "true ratings" most of the time. Suppose
someone loses a 64 cube and their rating drops from say 1700 to 1400.
Obviously there is not much merit in concluding so much on the basis
of one game. There is now a either a player who is way unnderrated
(bad) or possibly a player who about to start over at 1500 with 0
experience. Bad also since then there is not as much to risk in the
big cube game.
Is there any system which could work? I doubt it. Suppose my first
condition holds: Strategy to maximize rating is same as money play.
I think the only one which works is that winner gains multiple times
points won and loser loses a multiple times points lost. Note that
this precludes the use of handicapping based on ratings difference
and still has the problem of big swings.
To those who prefer money play to tournament play, this is
available to you. If you need something to make it more exciting
play for $. Don't expect to ever see ratings adjustments though.
,Bob Koca
bobk on FIBS
David Lloyd Mcadams
>
> Oh, and in this system, we don't need "matches" per see at all. Just
> get a sequence started, and play until one player doesn't join.
>
Yes, this is exactly the idea which I tried to get across (apparently not
so clearly because I've gotten several emails from individuals who were
confused by my request).
And to the other common response, "why not unlimited play?", I have a
simple response, simply the desire behind my request. I'd like to
be able to play a very fast game during my work breaks with an active
cube. It takes too long for me to find someone willing to play an
unlimited match (or a very long match) and also willing to quit after
just a single game.
I think some feature that allows me to take advantage of FIBS in this way
will also allow a number of others to do the same. I am a little
surprised that most of the feedback to me has been of the form "Why don't
you just play unlimited?" or (even) "Why not just join a long match and
resign all the remaining points after your first game?" and not
immediately positive. Most probably (I figure) the regular readers of
this newsgroup don't need the sort of change that I propose. The regular
readers here (I figure) are insiders of a sort who know each other well
and can find and play the sorts of matches they want more conveniently
than I can. But many players on FIBS are like me, fairly new and
irregular players who can use any possible features which would make
matching-up more convenient.
(Now, I'll illustrate the sort of frustration that an irregular FIBS
player can feel when he just wants to play a fast game or two with an
active doubling cube. Forgive me if I sound peevish, but it's this sort
of frustration -- that I've felt several times -- that has led me to
invest all of this time trying to spread the desire for a new FIBS
feature in the first place)
Do you know who I am? Would you accept an unlimited match from me if I
invited you to one out of the blue? I make 95% of my invitations to
people I don't know, and no one has ever accepted an unlimited match with
me out of the blue. You need to get to know me first. So we play a
3-pointer or a 5-er, you realize that I'm a worthwhile opponent, and
invite me to join you in an unlimited game next time. But then I never
see you again (because I'm a very irregular player).
If it's a small trouble to allow players to join in these "sequences" that
sharpclaw describes (exactly what I meant by "non-match play"), players
like me would benefit greatly. FIBS would get more visits from me, at the
least.
Patti Beadles
In article <Pine.GSO.3.95.970413...@cardinal1.Stanford.EDU>,
David Lloyd Mcadams <mca...@leland.Stanford.EDU> wrote:
>I proposed that if you and I decide to quit playing at a score
>
> cruelshoes 4
> david_mcadams 2
>
>then our ratings would change as if we had played six one-pointers, out of
>which I won two.
I'm not going to comment on the rest of the proposal, but I want to
address this one. Treating a rating change as a series of
one-pointers seems very unreasonable, and would skew the rating
system pretty strangely.
Why? Rating values aren't linear based on match length. If two
equally-rated players play a 1-pointer, they'll trade two points. If
they play a 3-pointer the value is 3.5 (and not 6, as would happen
with three one-pointers.) For a 9-point match the value is six,
although under your proposal you could win or lose anywhere from two
to eighteen points.
-Patti
--
Patti Beadles |
pat...@netcom.com/pat...@gammon.com |
http://www.gammon.com/ | "I trust you. It's just
or just yell, "Hey, Patti!" | that I'm scared of you."