The gateways are indeed centralized (though there are several different public gateways). But anyone who installs IPFS on their computer can access the content directly. Also some browsers support IPFS natively, e.g. https://brave.com/ipfs-support/
>But IPFS is a protocol just like bittorrent or HTTP,
Yes, but it's a protocol with a centralized single group doing development who can change whatever they want without the users' consent. Take a look at what is happening to the Tor ecosystem on Oct 15th this year: all tor v2 routing support is being dropped from the main client and infrastructure (for security reasons). Entire communities built on onionland and other tor v2 features, as well as all URLs/links, search engine databases, etc, will just go poof when the devs drop support.
Unfortunately being a protocol isn't enough. It has to be a community protocol, not a proprietary one where everyone follows one group's code. HTTP and bittorrent are safe from these kinds of attacks. IPFS isn't (yet) and that's why their butt-covering anti-sci-hub/libgen stance is worrying.
That's being quite unfair to the developers. The entire process has been public and announced well in advance.
You are welcome to download the Tor source code and add v2 functionality back in, and you’ll be able to visit sites hosted by people who have done the same. No one is stopping you.
To very quickly summarize why we are deprecating, in one word: Safety. Onion
service v2 uses RSA1024 and 80 bit SHA1 (truncated) addresses [1]. It also
still uses the TAP [2] handshake which has been entirely removed from Tor for
many years now _except_ v2 services. Its simplistic directory system exposes
it to a variety of enumeration and location-prediction attacks that give HSDir
relays too much power to enumerate or even block v2 services. Finally, v2
services are not being developed nor maintained anymore. Only the most severe
security issues are being addressed.
That being said, the deprecation timeline is now quite simple because v3 has
reached a good maturity level:
* v3 has been the default since Tor 0.3.5.1-alpha.
* v3 is feature parity with v2.
* v3 now has Onion Balance support [3]
* Entire network supports v3 since the End-of-Life of 0.2.9.x series earlier
this year.
If somebody wants to make his project in own way, why should it be decentralized? How to control ways and quality?
Everything has its limits. Perfect decentralization of development lead to a libgen collapse which not many know about thinking that it's bigger servers are still a real libgen. They are not and are mimicking libgen after capturing its vital parts. This happened exactly due to idealistic views that development can be entrusted to different anonymous individuals without regard to centralized management.
Decentralization implies loss of control. While it's good for results, it's very destructive for development and team building.
From what I understand, the notion of sci-hub/libgen "not being welcome" was only about discussion on the official forums. See https://discuss.ipfs.io/t/mirror-of-sci-hub-in-ipfs/1613 and https://news.ycombinator.com/item?id=25209246. But IPFS is a protocol just like bittorrent or HTTP, and the software is open source; it doesn't and can't enforce copyright restrictions.