OK, here's an update on this subject.
The two questions that matter:
1. How strong is Pure Chess/Chess Ultra?
2. Which engine/programmer is responsible for it?
The answer to 2 above remains unclear and a bit controversial. The developer dodges questions. See this thread from 2016 (note mclane's presence):
http://www.talkchess.com/forum/viewtopi ... light=xbox
Item 1 above is also tricky, but at least I have some game data that gets me closer. For reference, reposting a portion of Sedar's rating reckoning (just the top four levels):
My strength is around 2000 elo and I can win against Master (2000) and less against World Master (2200) but I did.
Below UCH levels, and if Grandmaster has 2400 then:
Grandmaster 2400
World Master 2200
Master 2000
Professional 1800
I questioned those numbers earlier, but upon reflection, I think Sedar might be pretty close, especially when Pure/Ultra is "tuned" to its peak.
I'm limiting this discussion to the top four levels. Bear in mind that I think Ultra and Pure are the same engine. In Pure, the top four levels have these names:
10. Grandmaster (same in Ultra)
9. Master (World Master in Ultra)
8. Wizard (Master in Ultra)
7. Scholar (Professional in Ultra)
Critical point: The Pure/Ultra engine only shows its true strength with time controls. The unlimited setting, which I tried earlier, is weaker. The seconds/move setting is not good. The engine plays best on the only other option, the incremental gain setting. I've chosen G/15, with a 10s gain, pitted against dedicated at 30s/move.
To give a rough sense:
1. Super Connie (1800ish) is roughly even with Level 7 (Scholar/Professional), but it's very close, and I need more data to be sure (with Excellence and machines at similar levels). Level 7 strikes me as the marker between average and above average for the majority of chess players. I can beat Level 7. Level 8 will not yield to my feeble blows.
2. So far, Level 8 (Wizard/Professional) looks like it might be the biggest ELO jump. At the moment, it feels like it plays around 2170. I've been testing it against the Mephisto/Saitek Master, and the feel is close to even, but only if I tune the Master to its tournament book.
3. I ran some early games without the proper time controls on Levels 9 and 10 with stronger machines, but it's clear that won't do for a solid eval. More to come. I don't want to bore anyone with game examples just yet, but I'll get there.
- R.
"You have, let us say, a promising politician, a rising artist that you wish to destroy. Dagger or bomb are archaic and unreliable - but teach him, inoculate him with chess."
– H.G. Wells