Annals of testing an elusive engine

[rainhaus]

I started testing the new Cyclones with version 4.0. It scored very good against stockfish1.4 and thinker 5.4Di and failed only with Rybka2.2n. But there was a mp bug, you couldn't use this derivative with more than two processors.
A few days later the xTreme series came out, but this one and all following updates had been decorated with another bug, the hash size couldn't be changed. No problem, "you have to break an egg to make an omelete". I reported the bug in this forum, but the thread went soon down the gurgler because you had suddenly to discuss about the author's malfeasances in the recent past and how to define an idiot and who is wasting his time with which engines. The banned author had read all this and fixed the bug on September 15 2009 08:00:52 . Satisfied, I started the tests again.
But from this time I clicked the Cyclone page more frequently and with suspicion. So number 01 of the new xTreme II series came not really unexpected. The author seemed to be in some state of creative frency or he followed the tricky principle: a new update make the tested one old. The tester's helplessness (it was me) bugged me incremental, I didn't want to get a sucker from someone's configuration orgies and I decided to abort. One or two days later I must have automatically loaded the site ....and I read something about 50 - 100 Elo more, I couldn't resist, the download ran at once.
You've seen the results and I wash my hands of it. I've taken the author at his word, but he had presented me probably his worst update of all. Bad luck for him, I don't test twice a whole session.
I'm not a programmer and I haven't analyzed the source and I haven't compared the various cfg-files. In such a way I don't know what may be the authors own parts of this derivative engine. The bandwidth of changing an engine reaches from the user's hobby tuning a la Chessmaster up to the programmers genuine working on the algos, the main, the subs, the functions, the loops and the variables of the source code. However, if pure playing strength is only the subject, nothing counts but the scores.
Will I test a Cyclone once again? Surely, when the author releases as reasonably as all the other programmers do.

Next: On the constructive side of Norman Schmidt.

Rainer Neuhäusler