hyperpape wrote:
Joaz: the cat's out of the bag here, and I think you need to say a bit more than "there are better ways of dealing with this." Either "shut up and ignore it" (bad solution), "hash it out amongst yourselves" (better, IMHO) or "be quiet while the admins decide what to do" (could be good or bad).
Maybe this is the only time in the history of L19 that someone is accused of being a sock puppet. But sock puppets are bad for the community, and if there's reason to believe there's one, people should be able to say it.
And, sorry Mike, but it's not like it's a crazy idea. If someone shows up on the forums and their only interest seems to be promoting a particular piece of commercial software, the idea has to cross through your head. Surely there are more than me, Deja and one or two others who are wondering........
a) Perhaps you ought to check back on your own postings in this thread? I was not the one who initially raised MFOG but you did (in effect) by making a statement about what these programs could or could not do (and most of the MCTS programs ARE "free software" even if not being made easily available).
Why did you interpret that "no longer true" as a promotion of MFOG? I have indicated my interest in "what is the state of the art" and I did say "as of 12.021" which is less than two months old. If one of the free MCTS programs also offered the feature I would have told about that too. See note below*.
b) You have apparently failed to note that I did not suggest to Stefany that she buy this program precisely because might be too pricey for a Bulgarian youngster to afford. My failure to suggest any of the "free" alternatives has to do with her current strength and where she apparently wants to go, not any opposition to free software. If she had told us that she was 12-15k instead of already 9k I would have suggested she get gnugo. OK, this is just my opinion, but to effectively use computer go playing programs they should be at least a half dozen stones stronger than you are and then you'll need a stronger program when down to three stones. Precisely to prevent learning "bad habits" from the machine.
c) The focus of many of the free go projects hasn't been create a competitive "production ready" release. WHY? (besides many of them being academic projects). For some of us the use of free software is a fetish, we feel that there is something
wrong about people creating software for pay (as opposed to in addition using monopoly powers to charge excessive amounts and only offer oppressive/unfair licenses).
Sorry, I lost my set of the "correspondence" in a house fire so I can't quote from some of the original "free software" papers. But might I humbly suggest that few if any of us involved imagined that we were suggesting something that would make it impossible for us to make our livings from creating software. Thought we were promoting something that would oppose the abusive conditions under which commercial software was being sold (not opposing the existence of all commercial software). Keep a couple things in mind. This was before the internet exisited as we know it. There were no "free downloads" of anything so it was planned that "free" software (source code) would be provided on standard medium for the "reasonable and customary charge" for that. Computers were using different architecures and operating systems so we assumed that users of "free software" would be paying to have it compiled for them, etc.
The point is that I am not "political" on the subject of free software. I am not promoting commercial alternatives when I say (if true) "at the present time this or that commercial progam is better than what is available with free programs". And in reverse, I am not promoting free software if I say "the terms under which this or that commercial alternative is available are unacceptable" (if true -- obscenely high price and/or unfair/oppressive license).
A couple of years ago there was more than one commercial product superior to what was available free so writing then I would not be talking about just one of them. But I am writing as of the current reality. Don't confuse "not opposing" with "promoting". When I say MFOG is offered with a fair license and not
too high a price that is just an opinion about the price -- not saying it would be worth it to you or anybody else and of course based upon my idea of what compensation we programmer/analysts should get for our time.
d) I am
not David Fotland (I'm about a decade older). Except for when still in high school when I and a friend built a tiny entirely solid state "computer" (in '61 that was a big deal) and briefly while in college I haven't done "hardware" and my degree was in Physics, not computing (University of Pennsylvania '66 -- so you can look up if I'm a real person). I did other things for a while, but about the same time Fotland graduated from college I began working for MassMutual Life Insurance and did software for them until retiring early after Y2K was over (and then a few stints back there consulting). Didn't do any "scientific" programming until after retiring.
But my "Monte Carlo" programming has nothing to do with playing go. Instead it's about genetics. Trying to breed American Chestnut resistant to the blight fungus, modeling what to expect in our populations of partially resistant trees over time (simulations for best numbers per line and how many lines to have in orchards with respect to things like inbreeding).
Michael D Novack, FLMI (that's an insurance industry designation)
* I suggested that MFOG could add this feature because of the way it built its tree starting with what its AI go knowledge engine suggested (I believe this is true of MFOG but I haven't looked at the code). After more thought I have realized that this isn't necessary and all that is required is the presence of such an engine (pure MCTS could select the move and AFTER ask the AI "look at this move and compare to your set of possible moves and if it matches one of them show the reasons"). Since the gnugo project does have an AI engine the branch trying to use MCTS would be able to add "reasons for move" even if not using the engine to build a plausible move set. Similarly Mick Reiss would be able to do that for go++8 if he ever gets that working competively to MFOG (which I doubt -- think he's too far behind to catch up)
PS --- Going to now acuse me of being Mick? Or because I mention gnugo, Trevor Morris who did some work on that (and who is top player at my local go club). Why should you find it strange that somebody who has created a few hundred thousand lines of code in my day would be interested in keeping current about "computer go"?