Tim Berners-Lee is rightly famous as the originator of HTTP, a fundamental of the World Wide Web as we know it.
But according to some, HTTP is old hat. It has helped create a web full of dead links, that is increasingly centralised, open to control by governments etc and prone to failure.
HTTP depends on IP, a device-specific method of addressing.
IPFS, on the other hand, relies on content addressing. In other words, each stored item has its own unique identifier, an immutable hash created for it alone.
This allows data to be stored anywhere, and those that request it can access it from the nearest location, or from many locations.
In fact, IPFS has characteristics somewhat similar to BitTorrent, the peer-to-peer protocol beloved of illegal movie downloaders and used by the likes of PirateBay.
Instead of downloading one file from one place a so-called Torrent swarm allows the user to download many shards of a file from many locations simultaneously.
In IPFS, as in a Torrent swarm, this is organised by a DHT, a distributed hash table.
And, with blockchain technology, you can record the hashes of data held – but not the data itself – with an immutable timestamp that also allows searching.
It’s early days yet. There are some efforts ongoing to marry blockchain technology to distributed storage in a payment model, such as Storj and SIA.
And IPFS’s related project the Interplanetary Database, which sought to create an internet-scale blockchain database, gave up the ghost earlier this year.
But, the idea of being able to distribute storage, to hold data anywhere and access it from anywhere, in a system uncontrolled by large entities, is surely and attractive one?
Well, perhaps for many, but the potential is possibly there to undermine the current economic models of big storage, in the datacentre and even, as some now call it, the “legacy cloud”.