[MacDV] how and why to benchmark

[Peter van der Linden]

Richard,

The comments I have made follow the rules of discourse and logic that 
are accepted among educated people everywhere.  They have been shaped 
and evolved and worked since the time of the ancient Greeks.  They form 
the basis of the scientific method worldwide.

If you or anyone wishes to make a claim, you have to produce evidence 
in support to those who challenge it.  If you would like some other 
process, then fine, but you cannot pretend that you have "won" the 
discussion, or that you have demonstrated anything at all, or that your 
other process shows anything except a lack of understanding about how 
factual matters are resolved objectively.

When I get around to writing the benchmarks, I shall do my utmost to 
evaluate the proposition carefully to reach an accurate answer.  [I 
know this is not the normal industry meaning of "benchmark"].  Why 
would I bother investing a minute of my time in anything less?  I am 
not interested in "winning" a debate (I have already done that because 
people have been unable to show the data on which they base their 
illogical beliefs).  I am interested in showing what the factors are, 
and allowing others to look at the evidence and make their own minds 
up.  And there is a very small, but non-zero, chance that I am wrong 
because of some factor that I have overlooked.  The benchmark will 
endeavor to find that.

Part of this involves making public the source code and benchmark 
process.  That will allow anyone to comment on it, to suggest 
improvements, and to reproduce it for themselves.  How amusing that you 
should think there is any point in doing otherwise.

      Peter