A couple of comments. The precision seems to be quite good. The particular protein, osteocalcin, may not be very informative or specific for short time-frames, such as 55,000 years in the case of the bison, but it will be informative for long time frames. Can you see now how that can be a useful way of extracting more information from ancient fossils?
Also, they only compared one, single protein! I wonder how many different proteins they will be able to recover from fossils.
Most proteins are recoverable for about 100,000 years (at cold temperatures). Osteocalcin is a bit unique, but they may find a few other proteins like that.