Einzelnen Beitrag anzeigen
  #74  
Alt 17.10.2023, 13:07
Tibono Tibono ist offline
Mephisto Wundermaschine
 
Registriert seit: 22.05.2018
Ort: Frankreich
Alter: 62
Land:
Beiträge: 482
Abgegebene Danke: 2.266
Erhielt 1.167 Danke für 385 Beiträge
Aktivitäten Langlebigkeit
9/20 6/20
Heute Beiträge
0/3 ssssss482
Re: GK2000/2100 und Konsorten...

Hi Hans21,

time ago I shared elsewhere:
Code:
I can remember facing the H8 bug 3 or 4 times at most with my Mephisto Chess Challenger, 
out of 130 games played up to now. So, roughly 3% of games encountered the bug.

Now, how should this translate into Elo-strength?
Let's assume a same level opponent: bug-free, the Morsch program should achieve 50% points.
3% of games where the H8 bug happened should have granted 1,5% points to the overall Morsch score; 
we can assume these are totally lost due to the bug (as the bug is an obvious blunder and has a really bad impact).
So, the buggy Morsch program achieves only 48,5% points instead of 50%. This is a -10 Elo points performance...

The impact on strength is rather light. So, it is essentially a matter of frustration for the proud owner!
Of course, should you enter your computer in a single-elimination tournament, any occurrence of the bug would mean
a sudden death... but the bug can as well not surface at all!

To answer your question: the bug enables no additional value as a misprinted stamp would do; as a number of devices suffer from this common bug. On the other hand, according to the above quote, it does not deserve a throw into the bin trash.

And finally: a workaround can be used, but honestly limited to analysis purpose as it is rather cumbersome and time consuming. The idea is the bug only surfaces on the fly, playing the blunder instantly or so; while if you run the program in analysis mode one half move before (thinking as the opponent, so) it won't (at least from my experience so far).
Setup the position so that the opponent's move is either mandatory or obvious (such as a honey pot Queen to capture) and leads to the position where the H8 bug surfaced; let the program calculate the time you want (analysis level) and watch the PV2 display (PV1 should be the obvious or mandatory move you prepared). This is the workaround move.
Next steps are easy: let the program make the obvious expected move, restore the playing level you want, play the workaround move, take back the answered move and play instead (you can also use referee mode to play the workaround move, to avoid the takeback step).

Regards,
Tibono
Mit Zitat antworten