Just To Clear Things Up- For Sure!

[geots]

I have matches to post, so I don't have a lot of time here. I ran a few days ago a 50 game 40/3 repeating match with Komodo64 ag. Ivanhoe B46e. It is the Ivanhoe version that is stronger than any others on 64bit. The match (wins/losses) was 21-8 in favor of B46e- for an approx. 90 elo difference. It has been said that was a ridiculous distortion. Partly true. It was a distortion, but not ridiculous. These things happen all the time.

Let me be very clear. It happened. But a_n_y_o_n_e who gleans from it that B46e is 90elo stronger than Komodo64 is, I am sorry- an idiot. Along with that, anyone who thinks that I BELIEVE the same thing is- a bigger idiot.

There was a previous one that I had run between them a week or so before that- I had forgotten about it. In it, B46e under exact same controls had beaten Komodo64 by 5 games which says +35 elo. I am convinced that from also looking at all the games in both- B46e is stronger than Komodo64. I suspect possibly not by even the 35elo- more like 21 to 30 elo. I understand that 21elo for this number of games is less than the margin of error. I am merely giving you my opinion.

Be careful about confusing my opinions and the facts. Anyone who doesn't agree with my OPINION here- I can understand that. But anyone who doesn't agree with the FACTS is beginning the digging of a hole for themselves. And the facts are the match was run under ideal conditions with perfectly credible results. Page upon page of bar graphs and other private tests cannot change the fact it happened. The results were skewed- so what. Happens all the time.

In a 200 game match between Strelka 5.1 and Critter 1.4 in same conditions, except in 32bit on a different machine- Critter won the match by 103-97. It was run over 4 days with a 50-game set each day. The middle 2 sets Critter won by a combined score of 4 games out of 100. But the first 50 game set- Critter won by 8 games. And in the last 50 game set- Strelka won by 6 games! How could Critter win those 50 games by 8, and the next 100 games by only 4, and lose the last set by 6 games?! IT HAPPENS.

So the credibility of my match results is up to you to determine as well. I have given you what I know are accurate results for any particular 50 games I have run. That is all I can do. If you don't trust them by now- you never will.

Lastly- my intelligence was insulted when I was told to check and see if B46e had been running on "default" parameters. It had been- I had made the DEFAULT= 1CPU. It was not a saved parameter setting. I am not even sure that was understood.

To make it worse- a possibility was that I had inadvertantly been posting only results that made Ivanhoe look good. How exactly do I go about "inadvertantly" doing that. Simple- I don't. It would have to be done on purpose- no other possibility and he full well knows that. So I am left with not only my credibility as a tester being in question, but also my credibility in general- specifically my character. If he says in NO way did he mean I might "only post good Ivanhoe results on purpose"- he is a liar. That possibility is EXACTLY the seeds he wants to plant.

In no way- no time- no where, have my feelings been hurt. I got over that silly crap years ago. But if he has the right to defend his engine, I have the right to defend attacks on my character.

So that is it. For me it is over. I will read no more reply threads- and no more emails, PMs on the issue. It is simple. I have completely and 100% lost any respect I may have had for him. And since I have my doubts that he really ever had any respect for me at all- we are even. Some fences cannot be- and never will be- mended. This is one of them.


Best,

george

[geots]

Unfortunately for me- all this could have been avoided. I realize that now in hindsight. It changes nothing in my post, but if I had used good judgment- no posts would have been necessary.

I realize now I should NEVER HAVE used a commercial engine as a litmus test for Ivanhoe engines. At the very least- knowing full well there would be some losses and probably a distortion or 2- I should have contacted the program author and run it by him to get his ok. If he did not like the idea- trashing it would have been in order.

But I did not do that, and there is nothing I can do now to go back and change it. All I can do is apologize and say how very sorry I am.


gts