I think of IPFS as a cross between HTTP and BitTorrent. Like BitTorrent it can seek a given dataset by some kind of hash, no matter where it's hosted; like HTTP it's more suitable for collections of small files.
In practice, it takes upwards of a whole minute to locate a file it's never seen before, so it's not terribly useful. It's better than nothing, but it's not terribly useful.
It's still cool that someone tried. IPFS is one in a long line of ideas that didn't really work. Occasionally some of these ideas have massive success, like the Internet, and Bitcoin.
In practice, it takes upwards of a whole minute to locate a file it's never seen before, so it's not terribly useful. It's better than nothing, but it's not terribly useful.
It's still cool that someone tried. IPFS is one in a long line of ideas that didn't really work. Occasionally some of these ideas have massive success, like the Internet, and Bitcoin.