Hacker News new | ask | show | jobs
by antpls 2845 days ago
When we think about decentralization, we sometime think about uncontrolled extreme decentralization. That's the ideal : no single point of failure, with everyone hosting their data where they want, and deciding who access what about them.

Practically, as pointed in the article, there are laws to comply with and those are IMHO the biggest lock to decentralization.

The fair middle ground between extreme centralization à la Facebook/Twitter and total network anarchy is something based on federation, like emails and Mastodon. With federations, there are several providers for the same end-user application, with native data exchange and interoperability. The idea is to give the power to anyone with hosting capabilities to compete with the Giants, even if only a few domains will actually survive (like Gmail, Hotmail, etc because of network effects and funds, probably).

What we need is a framework, or a backbone, that allows people to easily create new federated-native apps ("dapps") without thinking about consensus issues, protocols versioning, and with native laws compliance.

4 comments

So, you're essentially advocating rewinding the Internet to pre-Facebook time and continuing from there? I'm all for that. Damn, every time I stumble upon the Google Talk page, with its explanation how participating in an open decentralised world is good for everyone involved, I feel tremendously sad for how things worked out.

Unfortunately, the world decided to go the centralised way. At some point I had to re-work my outstanding papers, because any mention of peer-to-peer or even decentralisation meant immediate rejection. Internet service providers went more greedy, so if you don't build your own global backbone to have some leverage, you need to pay someone who does or you're hosed. Even the laws in place start to strongly reflect an expectation of overpowered centralised platform beneath any communication.

Then, finally, what we ultimately need is to figure out the money flow. People want polished products and that costs money. The centralised platforms we have today have succeeded because they figured some funding. Achieving that in a decentralised world is the main problem we should be looking at. I'm afraid "just slap blockchain on it" is a highly detrimental approach, but I haven't seen anything more serious (not that I looked seriously).

Dislaimer: I'm in Google now, but this comment actually reflects my personal post-INRIA sentiments.

> What we need is a framework, or a backbone, that allows people to easily create new federated-native apps ("dapps") without thinking about consensus issues, protocols versioning, and with native laws compliance.

I agree that this is probably the way forward. The only downside is how your identity is tied to the service provider you choose. It was a PITA when Lavabit went down and I lost that email address.

> The only downside is how your identity is tied to the service provider you choose.

Fully agree with this. The link to identity is not often brought up. I run a university lab focussed on re-decentralisation of the Internet as a day job. We focus on identity & p2p + trust.

Beaker browser is impressive early work focussed on the raw bit transport. It re-uses DNS for global discovery, its hard to do everything decentralised at once. How to do global search on a decentralised Twitter or spam control?

The hard issue we need to solve in the coming decade is the governance of such systems. Ideally it would rule itself. Definition of a self-governance system as: a distributed system in which autonomous individuals can collectively exercise all of the necessary functions of power without intervention from any authority which they cannot themselves alter.

Would be nice if all devices that wanted to be routable could get an ipv6 address including mobile, and we not have to worry about turn and stun...
Practically laws are forced to contend with the challenges of decentralization when it emerges, as in the case of nunerous laws that became harder to enforce due to the internet.

Federation is what we have now and it has a tendency toward centralization, as we see with the WWW and the mega sites where users aggregate.

>What we need is a framework, or a backbone, that allows people to easily create new federated-native apps ("dapps") without thinking about consensus issues, protocols versioning, and with native laws compliance.

This is really needed. There are endless great tools for centralized apps that make it trivial. I could build a usable forum website in rails in a day. I have no idea how to do that in a decentralized and secure way.