Re: [pbmserv] Palago notation?

[John Farrell <friendless.farrell@gmail.com>]

I've been thinking about such a system as part of my vague dreams of
writing a Palago bot. Ideally a solution would:

(a) not change the description of the location of a particular piece
just because other pieces were added later

but that means that you can't have

(b) equivalent board positions have identical descriptions.

I guess to describe the progress of a game you need (a), but to index
a library of game positions you need (b). My gut instinct is that we
would need a system which describes the position of a piece as a
series of steps from the start position, e.g. if the start position is
defined as X, then the locations in the six directions from there are
x0, x1, x2, x3, x4, x5. From then on you tack on another digit, but
positions can have multiple names, e.g. x01 is the same as x10, and
x03 is the same as x. You then need to canonicalise the location
descriptions. That would give you (a).

To achieve (b) you need some algorithm to canonicalise whole board
descriptions. Choosing x is a bit tricky - maybe you'd need to
calculate some sort of centre of gravity to identify a start location,
and somehow figure out a rotation from which the board is to be
viewed, e.g. the direction in which the board extends furthest.

This is the most thinking I've done in this direction, but if anyone
wants to proceed I'm keen.


John

On Sat, Sep 19, 2009 at 8:53 AM, David J Bush <twixt@cstone.net> wrote:
> | > Could someone please define for me the space numbering
> | > scheme used in Palago here on Gamerz? Thanks.
> |
> | Essentially random. ?The possible spaces are renumbered each turn.
> | ?You can't tell anything by looking at the notation.
> | ?It's the saem with all of Cameron's boardless games.
> | ?Would you like to suggest a better system?
>
> Well I doubt Cameron made use of any random value generator in
> his program. Renumbering the spaces is fine, as long as this is
> done in a consistent manner, otherwise, as you point out, the
> whole purpose of a notation system is defeated. Cameron's
> algorithm is deterministic, but it's not easy for the user to
> determine what it is. So, I would suggest a scheme which IS easy
> to understand. For example, label the vacant spaces in the
> leftmost column top to bottom, then the next column to the right
> top to bottom, etc. until you reach the bottom space of the
> rightmost column.
>
>
> To unsubscribe, send a message to esquire@gamerz.net with
> ? ? ? ?unsubscribe pbmserv-users@gamerz.net
> as the BODY of the message. ?The SUBJECT is ignored.
>
>