"There are certain positions where the equities of, or equity differences between, various moves is zero or extremely close to zero. In such cases, a bot may make a seemingly "random" move, or a bizarre move."
This is true, understanding that the code executed is static. It may chose the first or last evaluated of all equal moves, which may "seemingly randomly"; or, although very unlikely, it may in fact deliberately pick one of the moves "truly randomly".
"You might or might not choose to resign,"
This is false, assuming that the "YOU" here means the "BOT", because this is not a case of choosing one among several moves but one of only two decisions. Thus, given a condition, the bot will either always resign or never resign.
Just like a human, a bot should make that decision at a certain point during its turn (i.e. right after computing its equities and before making a cube decision and then a checker decision).
It would be poor programming to make a resignation decision in more than one location in the code and worse yet using differing pieces of code but if this were the case with one software, it wouldn't mean anything by itself.
However, if you have more than one software that share such characteristics, then it would at least cause some people to question whether and what it may mean.
"your resignation might or might not be technically correct (depending on whether the position is truly gin or not)."
This is a different subject and whether intentionally or not, it clouds and distracts from the real issue at hand.
"This argument is not quite as absurd as some of Murat's other arguments"
Why do you feel a need to make such a wholesale dismissing argument, instead of simply focusing on this "not quite as absurd argument of Murat"?
"However, all he has done is to collect some examples of such "random coincidences," without any effort to systematically calculate the probabilities..... collecting all instances of agreements and disagreements between the two bots, in order to arrive at some kind of statistically significant result."
As a strange coincidence to his credit ;) the subject header of Murat's original post in RGB was: "XG / GNUBG resigning behavior coincidence or indication of GPL violation?"
The data used to make the decisions may vary and be flawed, etc. to produce inconsistent results however the code that uses the data to make the decisions is static. Therefore there is no need for large statistical data or probability calculations, let alone collecting all instances, etc....
Since you all stubbornly trying to ignore/omit/hide the "both bots resigning wrong amount of points" issue, (which is indeed more indicative than failing to resign), let's look at one of Murat's examples:
                       |
| ||||
| XGID=aBDCBBa-----b---bc-f------:1:1:-1:65:24:10:0:25:10 | |||||
| White to play 65 | |||||
Gammons and backgammons don't matter. As soon as XG enters, it knows it lost all chances of winning the game (and the match) and it resigns.
The strange thing is that it resigns a gammon when a regular game is enough.
You may argue that in order to save time, it doesn't bother to resign the "proper or minimally sufficient" number of points but then why doesn't it resign a backgammon, especially even after rolling 65 and at the moment of making the resignation decision, it still has a man in its opponent's home board??
If we were only talking about XG, it would be "so what? just buggy software" but when we notice that GNUBG also resigns not 2 or 6 points but 4 points (a gammon) we can't help but become suspicious... :)
GNUBG resigns one and a half turn later than XG (because it doesn't check for "gin positions" again after rolling), but while it also still has a man in its opponent's home board and it's quite clear that both bots decide how many points to resign based on the same code.
BTW: XG's checking for "gin-positions" both before and after rolling the dice flies in the face of some people's pathetic argument that XG was failing to resign properly because it didn't want to waste time by checking for "gin-positions". Now, there! Twice!!