There is often a difference between theory and market. Sometimes market practice can explained from theoretical point sometimes not.For example, if we talk about Libor as a borrowing rate I think the theoretical rate calculations should be similar to what one was outlined in http://ssrn.com/abstract=1971373
, FX Basic Notions and Randomization. The similar idea I read somewhere for the derivation probably Interest Rate Parity. It is easy to calculate implied Libor using available data. Market practice is different and it is a law. One time quite recently it was a short news about Panel manipulations. After 2007-08 market was changed. It was observed deviations between primary rates. From my point of view there are economical background. Primary financial institutions that were closed to default were pumped by money ready for borrowing. Subjectively, these institutions were need new rates for practice. Such one rate is OIS though it is only intuition and there can be other reasons for using OIS. OIS as any other can be represented in the discounted form. This rate can be primarily lower than correspondent treasury rate that I think remains for T-bond prices calculations and should be interpreted as the primary discount rate in USA. If in future it will clear that OIS rate does not good to use they might return to old practice of discounting swaps. The existing theoretical contradictions treating OIS as a new discount rate rather practically negligible as it still does not discussed.