I was delighted to see that you'd commented on my Edo historical chess ratings in your blog of
Feb.6,2005. I can try to give some perspective on the odd-looking things you noticed.
First, it's worth noting that the Edo, Glicko and Spino ratings you give (all of which came from my site)
use a very similar data set for this period, mainly the collection of tournament and match results
compiled by Jeremy Spinrad. That's partly why the same players occur in similar places in these lists.
The Chessmetrics ratings are independent and use a different data set, namely Chessbase,
which according to Spinrad has biases at this early period (Staunton was fonder of recording
his wins than his losses, for example).
Why does Morphy go back to 1849? Well, my rating lists and Spinrad's use all available match and
tournament results, including casual and odds matches, for this early period where results are
scarce. The idea is that casual matches do give valuable information about playing strength, and
to ignore anything but unequivocally official matches would leave too little to go on. We do
not include partial match results, or results of individual games out of a match, unless the final
match score is known. This is to avoid the biases that can arise from selective recording of games.
As you describe on your Morphy's Childhood site, Morphy won a short match against Lowenthal in
1850, scoring 2/2 or 2.5/3 depending on which source you read. In 1849 he won what we can
consider a one-game match against Rousseau. These were casual matches, of course, but surely they
do give some indication of Morphy's strength at the time. Rousseau and Lowenthal were part of the
main group of active players of the time, so these matches allow the rating systems to assign
Morphy a rating as early as 1849, though a very uncertain one. Note that rating uncertainties (or
deviations) are also given on my site for the Edo and Glicko ratings.
So, the early ratings for Morphy have to be taken with a grain of salt, but point out what
the results of those early matches suggest - that Morphy could have been the strongest player in the
world at age 12 or 13. As I mention on my background page for the Edo system (way down near
the bottom), we could not have expected any player in the world in 1850 to do better than a
2.5/3 score against Lowenthal. In fact, if we look at results of other players against Lowenthal
in 1850 and 1851, nobody else had a percentage that high against him (for example, Anderssen had
8.5/15 against him, Buckle had 4.5/8, Kieseritzky had 9/17, Williams had 9/19).
Finally, you mention that arcane players turn up surprisingly high in the rating lists. But isn't
this part of the fun of rating systems? On the plus side, I could say that the rating systems
attempt to be objective, and therefore make an interesting comparison to commonly held but
subjective beliefs. On the minus side, however, keep in mind that some of these ratings are very
tentative and based on few actual game results - but the uncertainties are built into the systems
(Edo and Glicko anyway) so you can see where they are less certain. If you have the patience to
look through my Background page, you'll see all kinds of discussion of this sort of thing,
including some of the surprising cases. And by the way, an update to my site is coming soon.
Hope all this is of interest to you...