The performance score is always relative. Its performance for its class. Not shear performance... as otherwise say a TR2 beast would be a perfect score and every consumer CPU would be like a half that at best. We dont think that would be all that fair or all that helpful (someone looking at a 2bill processor doesnt care about what a 1K CPU can do... they care about other 2bill CPUs). Instead we try and keep things apples to apples. Roughly speaking we usually break things into a few groups: entry, mainstream, enthusiast, enterprise/workstation. If there is as subset / edge case then we look at its direction competition to see its relative performance.
Value score is also always relative and while performance does play a large role in its overall value we look at other things such as included accessories etc to build a value score.
With these two scores it makes finding the optimal choice easier for your needs (we are generalizing things and encourage you to make your own decisions based on what is most important to you). EG something could be a beast for its class but way overpriced for say the mainstream market. Great example of this is the 8086K. It offered very good performance (not best in class as it only has 6 cores) for a mainstream CPU BUT was wayyy overpriced... BUT if you wanted great performance and did not care about 'value' you could ignore our score and skew YOUR final score to fit your needs best.
IMHO more data the better when it comes to make a judgement on what to buy.
Hope that helps!