This was one of a number of recurring themes on the conference I took part in last week. At least when it comes to Creditmodelling. I am not able to tell whether this is a more widely held view in the finance community but it should be stressed that a lot of the top brass in the credit world attended the conference.
Personally i have written an article in Journal of Fixed Income that suggests a different way of modelling credit risk dependencies: Instantaneous Credit Risk Correlation, The Journal of Fixed Income 17 (2) Fall 2007, pp. 5-12. It remains to be seen if my approach goes the same way as the Copula (regardless of where it is going...)
IF YOU ARE NOT FAMILIAR WITH COPULAS: copulas, particularly Gaussian ones, are used to model default correlatons/dependencies. They are widely used in pricing CDOs etc and I think i dare say that they are at the center of the current credit debacle.