Dedicated nemesis

[Reinfeld]

In sports and in chess, it's common to speak of a team or a grandmaster feasting on one particular opponent. Maybe it's psychological - maybe it's a function of particular stylistic clashes suiting one particular opponent. For instance, Tal never had much luck with Korchnoi.

I'm wondering if this holds true for the giants of dedicated programming. We know who's on top in the aggregate - but how much of that reflects one particular machine beating the crap out of another particular machine, in effect fattening its rating while playing everyone else fairly evenly?

Before delving too deeply, here's the short list of names:

Lang
de Koning
Morsch
Kittinger
Spracklen
Schroeder

(I suppose you could throw Nelson in here, but the machines for which he's given sole credit simply don't rise to the level of the names above, even if we consider the current debate over Nelson/Spracklen hybrids).

The difficulty in figuring this out honestly is the ever-changing hardware. It seems reasonable to suggest that programmer vs programmer tests should include machines that are relatively even in terms of rating.

I'm sure the information is out there in one form or another, but it's not easy to sift. I'm looking at the SSDF rating list as I write this. Consider the Star Sapphire/Star Diamond (Kittinger).

This is the strongest Kittinger machine by every measure. By Wiki-Elo and by SSDF (and my own average of averages) it ranks below the following top-rated machines (plastic only)

1. Berlin Pro (and variants) - Lang
2. RISC 2500 and Montreux - de Koning
3. Atlanta - Morsch

The Star Diamond/Sapphire ranks above the best efforts of the Spracklens and (I think) Schroeder.

In SSDF, however, Sapphire/Diamond fights Montreux (9-11) and Atlanta (9.5-10.5) to near draws, while losing rather handily to RISC 2500 (6-14).

The SSDF numbers for Atlanta are also interesting.

vs Montreux: 16-24
vs RISC 2500: 25.5-14.5
vs. Star Sapphire/Diamond: 10.5-9.5

The hardware factor in that last set makes me wonder again about the differences between RISC 2500 models (SSDF seems to be using the weakest version with the 128K chip).

Again, I'm wondering which particular programmers give other programmers fits. It seems that the matches of dedicateds haven't created a sufficiently large sample size to reach clarity.

- R.

[IvenGO]

I might be not understand 100% correctly what's meant here, but if we would like to compare programmers skills we must do it using same hardware - we have to take two PCs and start some matches kinda WChess - Genius, Gideon - TascBase etc.

For DCCs themeselves there are some stats of CPU's perfomance (in dhrystones):

ARM2 14Mhz = 14701
SH7034 20Mhz = 14618
H8S/2000 25Mhz ~10000
68020 24Mhz = 5676

So in this case its not good idea to compare, for exaample, Lang vs Koenig using DCC's matches coz hardware in BerlinPro and Montreux has very different perfomance...

[Fernando]

As much as programs are different s it is only a necessary result that sometimes an specific code can be specially lethal to another specific code.
Even a weaker -in general terms- program can be fatal to a stronger one in general terms too.
For the same reason some programs are more lethal to human than plays them. A supposedly "just" 2300 elo program is more lethal to me that one that is 2700.
¿Why?
The style, the tricks, etc. Thousand factors you do not even imagine.

Fern