A good but flawed suggestion and an on point retort.
I do think it is actually dooable and have made similar suggestions in the past.
There is a more serviceable means of doing this though.
In essence, SE could record players raw potency per second across the entirety of the run and build up an anonymous percentiles stack much akin to logs. After a week or so of building up the initial data (Which could likely just be pulled from the QA team), you could simply rank players against their class averages and give a semi vague score or alphabetical rank from there.
Most importantly, once coded in it'd be self maintaining. As players improve, the difficulty of scoring highly increases, not to mention it wouldn't need to be adjusted for new content as it's simply building a table of averages from a pool of results.
Going off potency negates the majority of group comp disparities as well as removing gear from the equation giving about as pure an apples to apples comparison as you're ever likely to see in this game.