Roger, a typical game does not have 490 moves.  such a game would have a 
probability of 10^(very large negative number) of ever being played, thus it's 
significance is ~0. This is a dampening factor that N(number of moves) tends to 
infinity, the probability of playing such a game reduces dramatically.  Your 
brief assay would be like summing the cumulative brightness of all stars in the 
universe as if they were the distance of our own sun away.

I don't have my more detailed calcs at hand, but it's fairly simple to 
work out the approximation - but you have to cheat by assuming the bag is 
endless for X moves, for the purposes of simplicity.

1. Assume all words of eight letters or more are playable on each turn(114575). 
 
2.  Assume the average moves per person per game is X=15.
3. N^X=114575^30=~10^151
The true answer is significantly less than (3).  200 words or less are playable 
per turn (play a few games of quackle and ask it to generate all moves to 
calculate this for yourself) 

200^30=10^69

I think in my original calcs i looked at my own stats of 14 plays per game 
average 200^28=2*10^64 and you can see how the original figure is tending to 
the answer Graham correctly gave you. 

This in itself is a rough overestimation, as on each turn available there are 
less than 50 (and in a lot of cases less than ten) "sane" options.  
unfortunately, there is no mathematical function to describe "saneness" when 
talking about near-infinite branching integrals.  If we implement "saneness" 
(<50 moves/turn) the number of total games drops to around 10^40, if we 
implement "expert saneness" (average 5 moves playable) you are looking at a 
figure around 10^21.  Still, this is a number so large we could play a game 
every hour for the age of the universe and have a never repeat.  I think that 
when considering games of go and chess, the saneness and expert-saneness 
protocols reduce the number of games playable far more than scrabble is 
restricted.

As it stands, a definite approximation is impossible to achieve, but we are 
absolutely sure it is more than 10^21 and less than 10^151

Regards.

--- On Fri, 2/27/09, Roger <[email protected]> wrote:

> From: Roger <[email protected]>
> Subject: [quackle] Re: Scrabble Complexity
> To: [email protected]
> Date: Friday, February 27, 2009, 1:17 AM
> I think that 10^60 is a gross underestimation of the size of
> the game 
> tree of Scrabble. If we account for exchanging tiles and
> passes, then 
> each player has 128 or 2^7 options of what to do a turn
> without even 
> playing a tile on the board. We can alternate doing this
> for 5 turns 
> each time until one player plays a tile (2 on the first
> move of the 
> game) and then repeat. Because there are 100 tiles in the
> bag, there 
> will be ~98 or so repetitions of this and we'll end up
> with ~490 or so 
> moves. Each moves has a 2^7 (assuming all tiles are
> different, which 
> is not the case, but for our purposes let's assume
> this). (2^7)^490 = 
> 2^3430 ~= 10^1000 (which is considerably larger than
> 10^60). yes, we 
> are making the assumption that the tiles on the racks are
> unique to 
> come up with the 2^7 number, but notice that we haven't
> even taken 
> into consideration the geometry of the board at all.
> 
> I believe that Scrabble has a game tree complexity
> significantly 
> greater than 10^60.
> 
> --- In [email protected], Graham Toal
> <gt...@...> wrote:
> >
> > The *exact* number is impossible to calculate; it
> would have to be
> > enumerated - and it's so large that that isn't
> possible either!
> > 
> > However an estimate would be around 10^60.
> > 
> > Russell Honeybun is the expert in this area and should
> be able to 
> give
> > you a derivation of the above estimate.  If he's
> not on this list, 
> try
> > wordgame-programmers which is where I saw him last.
> > 
> > 
> > Graham
> > 
> > 
> > On Thu, Feb 26, 2009 at 7:14 AM, allstrugglewithin
> <pookp...@...> 
> wrote:
> > > I just search wikipedia and it seems like
> 'Go' has the most total
> > > number of possible games that can be played.
> > >
> > > But wikipedia doesn't say anything about
> Scrabble, so if someone 
> knows
> > > the exact number, please tell me.
> >
> 
> 
> 
> 
> 
> ------------------------------------
> 
> Yahoo! Groups Links
> 
> 
> 

      

Reply via email to