Usually, the discussions of the merits and demerits of the star system end up in the current Times critic thread. But it seemed silly to link Adam Platt's essay explaining New York/Grub Street's new replacement of stars with 0-100 numerical grades anywhere but here.
I've always found Platt's explanations of his star ratings particularly silly in their absurd apples-and-oranges overprecision. (Two stars because of a particular dish minus one star because there was a smudge on the wall?) I'm not sure I see how this numerical system will be better, though.
For one thing, 0-100 ranges always turn out to be illusory, because raters almost never go below 60.
For another, while I obviously applaud Platt's recognition of the problem of comparing unlike items (a pizza place and an expensive tasting menu place), I'm not sure that giving both of them a 90 is going to be any more sensical than giving them both three stars. I guess we have to see how the ratings play out.
One piece of great news is that the ratings in New York's restaurant archive will supposedly be at least occasionally updated from those in the published reviews.
As I've told you before, the five dimensional + many discrete dimension rating system of tabelog (cooking, service, value, atmosphere, drinks) solves nearly all of those issues (in context of Japanese dining) and could easily be replicated in nyc.