I think the real fix is pretty obvious: verified ids + full disclosure + penalties
Verified IDs can be done a number of ways, most likely credit
card auth, or possibly via something like Google+. (Needless to
say, there is no way to do this that won't rub someone the wrong
way, because they all involve centralized agencies... you "web of
trust" guys, this is your queue, but please explain how you're
going to get something that hasn't taken-off for decades to work
now).
The "full disclosure" bit is also obvious: you have a TOS that
requires you explain (at a minimum) any financial connection that
might bias your opinions.
The "penalties" bit is trickier: we need the courts to start
recognizing intentionally deceptive behavior on the internet as a
form of fraud, and not just "guerrilla marketing" or "the way
politics works" or some such.
I've been saying this for years, now, anything that doesn't work
this way (and that's nearly everything, including slashdot) is
just a part of the THE_TOY_WEB.
The standard counter-argument is that if you impose any
user-restrictions at all, your traffic drops off by a factor of a
100 or more, but I fail to see how this is such an insurmountable
difficulty. Reducing the number of comments on, say, New York
Times columns by a factor of 100 wouldn't greatly disturb me, and
myself I'd prefer it if the political operatives were required to label
themselves.
I'm sometimes wonder if there might be something clever you can
do with privileges that ramp up if you're willing to pay a
nominal fee and not incidentally, verify who you are.