@Ghost- perhaps it has changed, but as I understand it (and I think the site does at large) is based off your post about it on the GR site, namely that each player puts up some fraction of their GR (2/37, maybe?), and then this pot is distributed accordingly at the end according to whether it is WTA or PPSC. So in this sense, if you lose, your rating goes down by 2/37 whether you lost to the 6 best players on the site or the 6 worst, which is what I think rdrivera is referring to.
I don't think there is anything fundamentally wrong with this, especially because in general GR lag skill level, so it's fundamental "ethos" is for one to prove oneself so it makes sense that the differentiation is on positive results rather than negative ones.
I think what you wanted someone to say (maybe?) is that games against weaker players are not fundamentally broken, i.e/ that if GR were an unbiased predictor of skill then the EV for each player going into the game would be zero. This is true, but I think it's also worth mentioning that there is no doubt that since effectively all (good) players are underrated, playing against weak players is effective GR farming.