Doesn't the simple mathematics of why go is difficult make the 'faster
hardware' topic a small issue? At the starting move there is on the
scale of 361! = (roughly 10 ^ 300 ?) possible combinations having a
very simple algorithm with a hardware improvement of 10 ^ 9 can't
possibly outperform an algorithm with even minimal intelligent tree
pruning?


On Fri, Jun 3, 2011 at 6:18 AM,  <[email protected]> wrote:
> Send Computer-go mailing list submissions to
>        [email protected]
>
> To subscribe or unsubscribe via the World Wide Web, visit
>        http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
> or, via email, send a message with subject or body 'help' to
>        [email protected]
>
> You can reach the person managing the list at
>        [email protected]
>
> When replying, please edit your Subject line so it is more specific
> than "Re: Contents of Computer-go digest..."
>
>
> Today's Topics:
>
>   1. Direct DX11 and graphics cards for cheaper simulation
>      hardware? (Jacques Basald?a)
>   2. Re: Direct DX11 and graphics cards for cheaper simulation
>      hardware? (Don Dailey)
>   3. [Computer-go ]Congratulations to Zen! (Nick Wedd)
>   4. Re: [Computer-go ]Congratulations to Zen! (Andy)
>   5. Re: [Computer-go ]Congratulations to Zen! (Michael Williams)
>
>
> ----------------------------------------------------------------------
>
> Message: 1
> Date: Thu, 02 Jun 2011 20:48:16 +0100
> From: Jacques Basald?a <[email protected]>
> To: [email protected]
> Subject: [Computer-go] Direct DX11 and graphics cards for cheaper
>        simulation hardware?
> Message-ID: <[email protected]>
> Content-Type: text/plain; charset=us-ascii; format=flowed
>
> Don Dailey wrote:
>
>  > Are you trying to say that heavy playouts are better?
>  > Who is going to argue with that?    I agree completely.
>  > Are you trying to make the point that there are very simple
>  > to understand positions that computers cannot easily solve?
>  > I agree with that.   Are you trying to say that heavy playouts
>  > can solve many types of common positions orders or magnitude
>  > faster than light playouts? I agree with that.
>  > Are you trying to say uniformly random playouts suck?
>  > I agree with that.
>
> I do not pretend to argue. Just to clarify ideas and read what
> others have to say. And of course I agree on all that.
>
> In self play all MCTS programs scale. Everybody agrees and it
> has been tested empirically. Intuitively: If we admit that 2000
> sims is better than 1000, since nodes in the tree are trees
> themselves, it is clear that no matter how many million
> simulations we play, there will always be nodes with 1000 visits
> and they would be better evaluated if they had 2000. The entire
> tree relies on the correct evaluation at the nodes so the entire
> tree benefits of more sims.
>
> A different question is: Can a really weak program, say vanilla
> MCTS with uniform random playouts, just no eye filling (no RAVE,
> no progressive widening) reach the strength of, say Aya, with 2500
> sims (KGS 4 kyu) in 19x19 ?
>
> The answer is:
>
> Theoretically: Yes.
> In practice: No. Not with a trillion sims per move.
>
> You probably don't disagree since that is implicit in "heavy
> playouts can solve many types of common positions orders
> or magnitude faster than light playouts".
>
> Note that this question is equivalent to: Would the current
> version of Zen become a pro just with hardware evolution?
>
>
> Jacques.
>
>
>
> ------------------------------
>
> Message: 2
> Date: Thu, 2 Jun 2011 16:57:51 -0400
> From: Don Dailey <[email protected]>
> To: [email protected]
> Subject: Re: [Computer-go] Direct DX11 and graphics cards for cheaper
>        simulation hardware?
> Message-ID: <[email protected]>
> Content-Type: text/plain; charset="iso-8859-1"
>
> On Thu, Jun 2, 2011 at 3:48 PM, Jacques Basald?a <[email protected]> wrote:
>
>> Don Dailey wrote:
>>
>> > Are you trying to say that heavy playouts are better?
>> > Who is going to argue with that?    I agree completely.
>> > Are you trying to make the point that there are very simple
>> > to understand positions that computers cannot easily solve?
>> > I agree with that.   Are you trying to say that heavy playouts
>> > can solve many types of common positions orders or magnitude
>> > faster than light playouts? I agree with that.
>> > Are you trying to say uniformly random playouts suck?
>> > I agree with that.
>>
>> I do not pretend to argue. Just to clarify ideas and read what
>> others have to say. And of course I agree on all that.
>>
>
> I'm not really directing this to any specific individual, sorry it came
> across that way.
>
>
>
>> In self play all MCTS programs scale. Everybody agrees and it
>> has been tested empirically. Intuitively: If we admit that 2000
>> sims is better than 1000, since nodes in the tree are trees
>> themselves, it is clear that no matter how many million
>> simulations we play, there will always be nodes with 1000 visits
>> and they would be better evaluated if they had 2000. The entire
>> tree relies on the correct evaluation at the nodes so the entire
>> tree benefits of more sims.
>>
>> A different question is: Can a really weak program, say vanilla
>> MCTS with uniform random playouts, just no eye filling (no RAVE,
>> no progressive widening) reach the strength of, say Aya, with 2500
>> sims (KGS 4 kyu) in 19x19 ?
>>
>
> That's not an interesting question.   You can just run a program like this
> and get your answer as you have mostly specified the level and the
> algorithm.     I don't know the answer, but 4 kyu seems pretty strong to me
> for a program that only uses uniform random playouts and no tricks.
>
>
>> The answer is:
>>
>> Theoretically: Yes.
>> In practice: No. Not with a trillion sims per move.
>>
>
> Uniformly random sims will often send the program down an incorrect pathway,
>  but the program eventually discovers it's error (assuming there is SOME
> exploration) and will find better moves.    A trillion sims is a LOT of sims
> and I believe it is enough to get above 4 kyu, which is not a very high
> level.    But I don't really have a good way of estimating this level.
>
> I do believe that with the quality of the sims there has to be some
> adjustment to the tree search algorithms.   If the sims are horrible you
> cannot depend on them as much to direct the shape of the tree and visa
> versa.     So I would say that if such a program is allowed to be properly
> tuned and gets to do a whopping trillion playouts per move,  it's going to
> be stronger than 4 kyu.      I believe that even an alpha/beta search
> (perhaps like Aya was)  is doing a trillion nodes per move, it is going to
> play a good game.
>
> To be sure, you would still be able to find some simple position that it
> screws up big time.     But isn't that also true of the 3 and 4 Dan programs
> we now have?     So why can't we have a 4 kyu program that plays stupid
> moves too?
>
>
>
>>
>> You probably don't disagree since that is implicit in "heavy
>> playouts can solve many types of common positions orders
>> or magnitude faster than light playouts".
>>
>
> Heavy playouts is like turbo charging the program.   The difference between
> a uniform simulation and heavy playouts is not just enormous, but grows with
> each doubling.   In practical terms,  you cannot have a strong program
> without heavy playouts and in 10 years the majority of progress will be with
> the playouts,  not the search.       The quality of the software (if it's
> like in chess) will grow at least as fast and probably faster in GO as the
> hardware.   Of course if we are seeing the limits of Moores law,  that is
> even more true,  but I don't think we are (they said that 10 years ago,
> didn't they?)
>
>
>>
>> Note that this question is equivalent to: Would the current
>> version of Zen become a pro just with hardware evolution?
>>
>
> When you ask questions like this you open things up to confusion and
> argument because you didn't ask a well formed question.   You did not
> specify a time period or any kind of rate of hardware evolution.   Are you
> just asking for a prognostication?
>
> With infinite hardware evolution (assuming it will always reach some
> arbitrary speed if you wait enough years or centuries)  then the answer is
> yes.     But at some point the limits of physics must rear it's ugly head
> and the hardware progress with be much slower.  So it's not possible to give
> any kind of estimate that is attached to reality.
>
> I can only guess about this,  but I think we are going to be surprised by
> how much a 100x faster computer gives us.   If you had asked this exact
> question 35 years ago with respect to computer chess,   most reasonable
> people would have said that having a computer 10,000 times faster would
> probably only give you a hundred to two additional ELO.   I'm not joking,
>  this is how it was.      This happens because when we see a really big
> problem,  we imagine that it's even bigger than what it is.    It was very
> common back then to show people problems from real games that no computer
> imaginable would ever be able to solve.   And yet now they are solved
> trivially.   I don't mean to be disparaging,  but it shows how stupid and
> naive and shortsighted we all tend to be at times.
>
> The Zen question won't happen because the software (in my opinion) will
> evolve much faster than the hardware.   But they will both be very
> important.    We will get a pro playing Go program decades sooner if we
> don't wait for hardware to run Zen on,  but instead we make constant
> advances.
>
> Don
>
>
>
>
>>
>> Jacques.
>>
>> _______________________________________________
>> Computer-go mailing list
>> [email protected]
>> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: 
> <http://dvandva.org/pipermail/computer-go/attachments/20110602/efb664e5/attachment-0001.html>
>
> ------------------------------
>
> Message: 3
> Date: Thu, 02 Jun 2011 22:17:29 +0100
> From: Nick Wedd <[email protected]>
> To: [email protected]
> Subject: [Computer-go] [Computer-go ]Congratulations to Zen!
> Message-ID: <[email protected]>
> Content-Type: text/plain; charset=ISO-8859-1; format=flowed
>
> Congratulations to Zen19S, winner of last week's slow bot tournamwent,
> four wins ahead of its nearest rival!
>
> My report is at http://www.weddslist.com/kgs/past/S11.2/index.html
> As usual, I will welcome your comments and corrections.
>
> Nick
> --
> Nick Wedd
> [email protected]
>
>
> ------------------------------
>
> Message: 4
> Date: Thu, 2 Jun 2011 16:48:37 -0500
> From: Andy <[email protected]>
> To: Nick Wedd <[email protected]>, [email protected]
> Subject: Re: [Computer-go] [Computer-go ]Congratulations to Zen!
> Message-ID: <[email protected]>
> Content-Type: text/plain; charset="iso-8859-1"
>
> "Also in round 9, there was an interesting game between AyaMC and
> ManyFaces1, involving a semeai at the lower left sideof the board, as shown
> to the right. If move 215 had been answered by a move at 218 (or one point
> below, at A7), the result would have been seki. Move 216 is a blunder,
> allowing Black to kill with 217. White 218 achieves nothing."
>
> I think at that point white can't do anything:
> http://eidogo.com/#HFxgu5j:0,1,0
>
> It is a big-eye kills small-eye situation.
>
>
>
> On Thu, Jun 2, 2011 at 4:17 PM, Nick Wedd <[email protected]> wrote:
>
>> Congratulations to Zen19S, winner of last week's slow bot tournamwent, four
>> wins ahead of its nearest rival!
>>
>> My report is at http://www.weddslist.com/kgs/past/S11.2/index.html
>> As usual, I will welcome your comments and corrections.
>>
>> Nick
>> --
>> Nick Wedd
>> [email protected]
>> _______________________________________________
>> Computer-go mailing list
>> [email protected]
>> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: 
> <http://dvandva.org/pipermail/computer-go/attachments/20110602/227bee15/attachment-0001.html>
>
> ------------------------------
>
> Message: 5
> Date: Thu, 2 Jun 2011 18:18:00 -0400
> From: Michael Williams <[email protected]>
> To: [email protected]
> Subject: Re: [Computer-go] [Computer-go ]Congratulations to Zen!
> Message-ID: <[email protected]>
> Content-Type: text/plain; charset="iso-8859-1"
>
> Why doesn't CrazyStone compete?  I'm guessing Remi does not have the time.
> Perhaps one of his trusted students could run it.
>
>
>
> On Thu, Jun 2, 2011 at 5:48 PM, Andy <[email protected]> wrote:
>
>> "Also in round 9, there was an interesting game between AyaMC and
>> ManyFaces1, involving a semeai at the lower left sideof the board, as shown
>> to the right. If move 215 had been answered by a move at 218 (or one point
>> below, at A7), the result would have been seki. Move 216 is a blunder,
>> allowing Black to kill with 217. White 218 achieves nothing."
>>
>> I think at that point white can't do anything:
>> http://eidogo.com/#HFxgu5j:0,1,0
>>
>> It is a big-eye kills small-eye situation.
>>
>>
>>
>> On Thu, Jun 2, 2011 at 4:17 PM, Nick Wedd <[email protected]> wrote:
>>
>>> Congratulations to Zen19S, winner of last week's slow bot tournamwent,
>>> four wins ahead of its nearest rival!
>>>
>>> My report is at http://www.weddslist.com/kgs/past/S11.2/index.html
>>> As usual, I will welcome your comments and corrections.
>>>
>>> Nick
>>> --
>>> Nick Wedd
>>> [email protected]
>>> _______________________________________________
>>> Computer-go mailing list
>>> [email protected]
>>> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>>>
>>
>>
>> _______________________________________________
>> Computer-go mailing list
>> [email protected]
>> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>>
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: 
> <http://dvandva.org/pipermail/computer-go/attachments/20110602/8c41e8f9/attachment.html>
>
> ------------------------------
>
> _______________________________________________
> Computer-go mailing list
> [email protected]
> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
>
> End of Computer-go Digest, Vol 17, Issue 11
> *******************************************
>
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to