A Coder’s Diary 3 (How I Tried IPFS)

Wednesday, March 6, 2019

Dear Diary,

This morning I decided to try out IPFS once again. It is such an intriguing project with a seemingly limitless potential.

It is broken!

I do not know how many articles on it have been written, quiet some of them on Medium. But there is such a gap between whats going on in your visions and dreams, and what is happening in reality. I doubt there have been many people who write sophisticated stuff about IPFS and at the same time try to do something really useful with it. I was definitely one of those who got stuck in theoretical speculations, not doing hands on stuff.

IPFS is supposed to be a peer-to-peer filesystem, serverless, distributing fragments of files over the net and setting their pieces together whenever and wherever it is needed. Each file is supposed to be uniquely identified by its hash.

I tried it out with a small picture. Its hash was

QmWmEZ3FiNB2KcniUHDFQPTeo6XTzFeje41AcN4xE3Mqq2

The installation of go-ipfs went smooth on two different servers. All the commands from the Installation-Dokumentation worked as expected. I started the daemons. No error messages. The whole infrastructure seemed nerdy, but well structured, well documented, solid and reliable.

The simplest of all use cases should be to add a file on one node, and read it out on the second one. That is where I got stuck. No error message, nothing strange at all — but the file shows up only locally, not on the second node, and not here.

Of course, there could be many reasons, why it did not work for me! Most likely there is just a configuration error caused by me. Or I just used the wrong commands. Both these explanations do not seem likely to me, but I do not want to exclude them. It could be my fault! But that cannot be the complete truth! For whatever reason my simple test failed, it falls back on IPFS and it is justified to say that IPFS is broken.

Getting a file from one location to the other is just an everday’s task. People are doing it everyday, and normally they do not need to read any docs for it, or do any additional installations. OK, the way IPFS does the copying is something completely new and could revolutionize the web.

I have no doubt that it works in principle, has been thoroughly tested and works even today for other people but me.

But even with such a brand new technology, it is not good if ordinary people fail to do ordinary tasks with it. That can only be excused, if you value vision higher than reality. But the visions of IPFS are not here since yesterday. The project has started in 2015. It is ambitious. Very ambitious. It is fitting well into the hype of peer-to-peer networks, that we experience today. Because of the analogy to conventional filesystems, the goals of the project seem to be understandable for many people. And when they have understood, they wish it were true, and the wish soon becomes stronger then the will to test it in reality.

Maybe, a reality check should be done more often and with many more of the promising projects.

Sometimes, being a believer is just not enough. You have to get it working. I did not get it working. Shame on me!