Back in the early 'oughts, the advent of peer-to-peer filesharing systems inspired a number of projects and companies to be founded with the goal of applying P2P technologies to more general applications.
At the time, P2P architecture offered 2 primary advantages:
1. Since the infrastructure of a P2P system was provided by its users, it would be self-hosting and self-scaling.
2. Having no central point of control, P2P systems are difficult for law enforcement to censor or shut down.
Unfortunately, it turned out that P2P had one big downside, which was complexity of implementation. In theory everything would smoothly scale to millions of users, but it required some serious expertise and investment to achieve that in practice. As hosting prices dropped and AdWords was launched, the conventional client-server website approach reasserted itself as the most sensible way to build networked applications.
Thus, P2P's only remaining reason to exist was for the evasion of the authorities. And even this had its caveats- law enforcement may not be able to shut down a decentralized system, but it certainly can shut down the company(s) that pay for the system's ongoing development and maintenance.
While decentralized technologies still have a lot going against them, I think the thing that has changed this time around is the presence of cryptocurrencies. Cryptocurrencies offer the possibility to fund the development of decentralized software using micropayments, but considering the amount of work required to design the architecture of such software before even being able to think about usability concerns, I am skeptical as to whether micropayments or even ICOs will result in the investment necessary to make decentralized systems appealing to the general public.