The Arpanet was designed to be decentralized
The tensions between the USA and the USSR after the Second Worldkrieg War prompted both sides to engage in an extraordinary arms race that was to last for decades. However, this arms race was not only reflected in growing quantities of nuclear weapons and men on the moon, but also in significant developments in the field of information technology: the important research units and military entities were to be networked, resistant to attacks in a possible Krieg. The Arpanet, the forerunner of today's Internet, emerged from these efforts in 1969. It was remarkably conceived: Packets were to be routed and delivered within a grid-like topology.
Today, this sounds like old hat. Peer to peer, mesh topology, TCP/IP: what was high tech back then is commonplace today. But the Internet, which was originally very similar to the Arpanet, has changed rapidly. Although today every Internet user is also a "pro-sumer" - whether they are surfing the bloggosphere in a cool way or simply want to show a picture of their dog - the Internet has become centralized. You can search on Google, find things too, socialize on Facebook or Twitter, and if you want to know something, you turn to Wikipedia. Entertainment is available on YouTube and Netflix, and IMDB tells you how good the movie you've just seen was.
There is also a trend towards centralization in the technical sector
This trend has also been noticeable on the technical side for years. Hosting is shifting to the cloud to large providers such as Google or Amazon, who conveniently also co-develop the most popular tools and technologies. Many are actively opposing this development and, mindful of privacy and security, are once again advocating decentralization. The boom around blockchains was certainly an important point in this development. With the popularization and new tools and platforms, it was suddenly possible for anyone to store information in a cryptographically secure, decentralized manner. And it wasn't long before everything really was "powered by blockchain technology": logistics, gambling and even trading cards.
The Inter Planatary File System
Meanwhile, a number of interesting projects have emerged, including the Inter Planetary File System. IPFS is a peer-to-peer protocol that is intended to replace HTTP and decentralize the network again. It has been under development since 2014 and has already gathered a remarkable fan base.
The basic principles are simple:
- Each participant in the network only saves the data that is relevant to them.
- If you want to download data, you do so from participants who have saved it.
- All data is hashed so that it can be identified.
- The network helps participants to find the data belonging to certain hashes.
IPFS has already proven itself in the fight for freedom of information:
After Wikipedia was blocked in Turkey in 2017, the IPFS team set about making a snapshot of the site available. This is stillavailable today[via a gateway].
Hosting via IPFS is also relatively simple. Once you have created your repository and started the daemon, you can add your website to IPFS and pin it. Pinning ensures that the site is permanently stored and can be accessed by others. The site can then be accessed from anywhere via a gateway:[ipfs.surreal.is].
Challenges for IPFS
But what's stopping people from dropping everything and moving all hosting to IPFS? IPFS has a few fundamental problems:
- Because it functions more like a file system, the Interplanetary File System currently only allows hosting of static sites. This means that the majority of services on the Internet would simply not work via IPFS alone.
- The access times within the file system are very high. This is mainly due to the fact that the code base is not performant, at the same time the network suffers from the growing size -- the current implementation was not designed for thousands and thousands of users.
- It is not yet user-friendly. IPFS must be installed on the computer for proper use, and operation is still via the command line - although there is already a browser interface. If you want to access a resource, you either have to work with hashes or request it via a gateway; if you are looking for something, you will hardly find it. Addressing via names using IPNS does work, but is cumbersome and not beginner-friendly.
IPFS does not currently allow any statements to be made about the size of the network or the number of peers, but indirect metrics suggest that popularity is stagnating. Most of the award-winning blog articles were written quite a while ago, and IPFS is still operating in its filter bubble of a tech-savvy audience.
A dying project?
This development seems strangely familiar: Wasn't the decentralized social network Mastodon the "hot new Twitter"? Wasn't Graphite Docs once a dangerous competitor to Google Docs? Will IPFS disappear from the market as quickly as it appeared? Probably not. The community around IPFS is very active and is actively addressing the problems that users are still experiencing. Many relevant market players are behind IPFS - Cloudflare, for example, has made a gateway available to the public and supports the use of readable domains and HTTPS.
However, the success of developments such as IPFS or Mastodon may not depend on individuals: even if decentralization is praised on a small scale, its basic idea is so disruptive for existing structures that it is questionable whether society will ever embrace change on this scale - especially if the effective benefit for the individual is small.