First, PLoS fees are lower and as a nonprofit only have to support operation (no profit). Also as a nonprofit there is some transparency as to where those fees go, so I assume that publishing there really does cost something like $1600.
Second PLoS's fee rules include the sentence "Authors' ability to pay publication fees will never be a consideration in the decision whether to publish."
Elsevier's "unfair advantage" (something we're supposed to look for on HN) or IP is their ownership of journals that acquired renown over decades. Publication in those journals as opposed to others has, depending on your field, tremendous effect on tenure and grant opportunities. It takes a long time to transfer that repetitional benefit to other institutions like PLoS.
That's why physics (with a more fluid tradition of journal bona fides and a wider use of preprints and passing papers around) was the pioneer field for open access and the like while, say, medicinal chemistry has been a laggard in this area. Doesn't mean the researchers want it that way, but those cultural traditions take a while to change.
If UCB and some other high profile institutions (I'm looking at you, MIT!) can declare a policy away from these journals it will provide cover to allow others to do the same.