Chessmetrics uses a method to compensate for rating inflation, but it is seriously flawed. Neither Anderssen, Bird, nor Paulson played at any such level. Play over their games - read the notes which have been tested over more than a century and see the sort of mistakes they made. Compare to players of that rating today - it doesn't work. Today's 2744 players don't make nearly the dumb mistakes Anderssen did. Maybe he was 2544 at his peak, which was probably London 1851.
Seriously, look at the Chessmetrics ratings for the lesser master through the late 19th and early 20th centuries, you will find them all laughably overrated.
Morphy's strength cannot be approximated with any scientific accuracy because he was too much stronger than the players he faced. Remember, his chess career was a way for him to kill time - having passed the bar exam at 18, he had to be 21 to practice in Louisiana, so he played chess. He never accepted money for playing, except when Anderssen insisted on the traditional stakes and Morphy quietly returned the purse to Anderssen's wife.
The problem with using a method to correct for rating inflation is that according to Regan's quantitative analysis inflation does not objectively exist. Eg, the objective accuracy of play of a 2700 now is slightly higher than that of a 2700 from 30 years ago. As a result, any correction for "inflation" is really a correction for the general improvement in the standard of play (according to computer metrics). It is a perfectly reasonable viewpoint to say that players in the old days were handicapped by the primitive state of chess knowledge, and their ability to find good moves was hindered by their lack of databases and computers to prepare for matches with, as well as having far less theory to learn from and to be inspired by. But the bottom line is that they made more and bigger mistakes. If Morphy could have been born in the 20th century and somehow the environmental factors that made his chess talent flourish preserved but also he be given access to modern resources and competition to hone his skills, he might have been a stronger player. As it was he was a very strong player for his time, but lacked the brutal precision of a Kasparov. And surely so would Kasparov, it he had been born in the 19th century.
My theory is that Paul Morphy's rating in todays standards, if he happened to come from the grave and was playing as well as he was in his prime from the 1800s his fide rating would be about 2638; while not world champion level, he could still be considered a strong grandmaster. I came to this conclusion by looking at Morphy's match/tournament scores of his games that were not played at odds or other handicaps. Also I only mentioned opponents that Morphy played a substantial amount of games against (at least 12) in order to increase the accuracy of my conclusion (this also factors in the idea that chess players often have 'bad days', and with all the traveling players did they often were sick for the first few games in a match).
His record against Eugene Rousseau in match play was 45/50; which would make his rating at least 358 points higher than Rousseau.
Against Louis Paulsen in tournament and match play his record was 9.5/12; making his rating at least 226 points higher than Paulsen.
Against George Hammond in a match Mophy scored 15/16; making his rating at least 412 points higher than Hammond.
Against John William Schulten in a match Morphy scored 23/24; making his rating at least 457 points higher than Schulten
Against Henry Edward Bird in a match Morphy scored 10.5/12; making his rating at least 320 points higher than Bird.
and finally Morphy scored 14/17 in matches against Adolf Anderson; making Morphy at least 257 points higher than Anderson.
This astounding record is 117/131; 89%.
The rating points are based on his win percentages against those mentioned players, also I only choosed players that Morphy played a substantial amount of games against in order to get accurate results, and all of the players I listed were pretty good.
I estimated that this lot of masters today would have an average fide rating of todays standards of at the 'very least' 2300, so 2300+338=2638 (338 is average amount of points Morphy is higher than his "rivals.") Also when i say 'very least' it should be noted Adolf Anderson in particular was 'estimated' to be over 2600 by Aprad Elo (the inventor of the rating system in the first place). One should also note that Aprad Elo himself believed Morphy was 2690.
On top of this evidence it is said that Morphy often played 8 blindfold games at a time while still managing to win the great majority of the games. I asked one of my freinds who is a Grandmaster (2496 fide atm) how many blindfold games he could play while still keeping most of his strength, and he replied that he was sure he could play 3 at the same time and maybe a 4th without over exerting himself. Morphy being able to easily play twice as many blindfold games further proves that he was not just any 'weak grandmaster, strong international master level', and that he could in fact be a strong 2600 rated Grandmaster.
I came to the idea of how much rating points Morphy was stronger than an opponent by plugging in his win percentage into this chart that corresponds with elo ratings http://www.chessville.com/Reference_Center/WinPercentageExpectancies.htm.
which are summarized in http://en.wikipedia.org/wiki/Paul_Morphy
If what you say is this (I have bolded and colored red), then Morphy would be more around 2857 is what it seem like what you are actually saying.