But, do realize that there most certainly is value being contributed by journals who are making you pay. For example, the editorial work, the typesetting, the selection, the whole system, etc. So they do need a paycheck from somewhere.
So, I think this is actually like the common software/music/movies piracy situation... but a lot better!
My (simplified) feelings about piracy are: do it if you're dirt-poor, don't do it if you can easily afford it. By this account, Kanye was a d-bag for pirating.. whatever software he pirated, because he, of all people, should be able to afford it. A starving artist with lots of debts, in my opinion, can't be accused of doing a great moral wrong when he's stealing a thing which has an effective zero marginal cost (arguably). The biggest problem here is that... there do exist people like Kanye who really should pay for things but end up not doing so.
What's different in this scenario is that... I am almost 100% certain that all institutions that should be paying... will be paying! I mean, can you imagine some Harvard or MIT lab skipping payment on the journals? No, that's just not going to happen. The Harvards and MITs of the world will keep paying... the rest -- the public, or schools in suffering areas and nations, who can't afford it as easily, will be able to get it using scihub. It works out wonderfully in this way.