On the contrary, you can know with absolute certainty what an Orc's CR is, in every case.
On the contrary to what? I never said anything about the relativity of an orc's CR.
EL is what's relative, which means that
the point is to generate a CR/EL that serves as a guideline for the majority of cases.
is problematic. The CR serves as a guideline for all cases; the EL serves the purposes of the campaign.
And the core designers are, in fact, very specific about what they consider a typical party. It's a very thoroughly tested design.
A very thoroughly tested design that doesn't count ability scores: an average is assumed; extremes are not excluded.
Any deviation from the typical party means that EL (which is the relationship between party CR and monster CR) will change, higher or lower.
Look; while calculating a monster's CR, yes, one assumes a 'typical' party--this maintains a standard that other CRs may be measured against. A CR is a CR, if I buy it from Wizards or from Necromancer (supposedly).
When it comes to playing against these things, however, it all becomes relative, on the one hand, because nobody sells PCs, on the other, because playing style (or campaign) determines typicality.
If people want to play super-PCs, you don't make adjustments to the EL: that would defeat the purpose of playing super-PCs.
If people want to play PCs perfectly balanced with CR and they happen to roll up a bunch of high-statted guys, then you may make an adjustment to EL.
It's a matter of maintaining the broadness of the field of options.
Though perhaps he doesn't realise it, that's the real implication of Anubis's argument. The point that counting ability scores may sometimes screw up the the pairing of PCs and challenges (in favour of the challenges) only reveals that this is a situational modifier; sometimes it's relevant and sometimes not--which means as a general rule it's not, in order to err on the side of the PCs.