The real debate is, who gets to decide what the filter is? We have to have one, nobody can handle the full flood of data, so it's not a matter of whether we have one, it's who gets to control it.

Do you let the user control it? That's great philosophically, but I think in practice it will be hardly any different than search engines just intelligently twiddling the settings based on your actions. On the other hand, pretty much any answer other than the user creates even larger problems with conflicts of interest.

Eli Pariser did a great job of selling the idea that we should be exposed to true diversity, but I still fear that in practice the truly gut, visceral reaction that people have to this is that people can end up in $NOT_MY_IDEOLOGY bubbles, and what they really want is going to be less about diversity and more about making sure that people's bubbles put more of $MY_IDEOLOGY into everybody's bubble (as $MY_IDEOLOGY is obviously the one true correct ideology and therefore deserving of special treatment). There are a lot more people claiming to value diversity than there are people who truly want people to be exposed to the best arguments against $MY_IDEOLOGY there is.