> I don't know whether most people do prefer to have only one source for their information; I'd like to see more data before forming any strong view on that one.I admit I have no scientific sources. This is just a gut feeling birthed from my numerous interactions with people during my whole life. I also happen to believe that not everything can be measured scientifically and there are a lot of things "all people know but almost nobody will admit in an official survey" and that "almost everyone prefers buying only one newspaper" is one of them -- but that would go wildly off-topic and I'll stop right here.
> Is that a problem?
Of course it is a problem. If we assume there are malevolent people who would want to suppress certain kinds of news (and with Trump, Kim Jong-un et. al. in power I don't think anyone can doubt the existence of these people anymore) it's much easier for them to bribe or coerce a single entity to censor things. For a true free speech, we need a fully anonymous but attack-resilient immutable decentralized network (sorry to keep repeating myself but... like IPFS). Good luck DDoS-ing or bribing/coercing that. It's made to be resistant against planting fake data, man-in-the-middle attacks, and DDoS -- right from the get go.
Admittedly IPFS isn't still there, for example it lacks automatic replication and the fact that there are already organizations whom you must pay so they "pin" (replicate) your content is telling me IPFS might eventually lose its credibility as well :( again though, off-topic. Sorry, you caught me in a very chatty mood.
>Is it really so far-fetched that we could cut out the middle-man in some of these cases and move back to a more peer-to-peer, decentralised system with neutral infrastructure?
In technical terms, it is not far-fetched at all. We're actually very close to it. Man, I'd love to work on that, only if that was my main source of income. I would pour a lot of energy and heart into such a work.
In economical and general reality terms however, it's almost impossible. As I mentioned, I am convinced there will be a lot of resistance from agents who would be negatively impacted in the pockets or their data collection. But you know what, if I am wrong, I'll be the happiest little panda.