Comment by gonehome | Hacker News Reader

Comment by gonehome | Hacker News Reader

I think this is a good idea and I think making it easy for people to host their own services is valuable. [0]

It’s also been tried a few times with little success (sandstorm.io).

I think the problem is deeper, modern infrastructure is messy to run, it’s hard to configure, and every thing has to be specifically tweaked to work. Something that works for regular users will require extensive tooling to be useful and will need that tooling for every application and use case.

I think Urbit is probably the best long term bet (old blog post, but gets the idea across): https://urbit.org/blog/magic/

A more recent one: https://urbit.org/understanding-urbit/

Urbit abstracts away the p2p complexity so in the future users can used decentralized applications built on the platform without having to even know they’re decentralized. I think this is the only way this kind of thing can work at scale for regular users.

[0]: https://zalberico.com/essay/2020/07/14/the-serfs-of-facebook...

With respect, the blog post you pointed to is a load of self-referential blather, when I go to the top of urbit.org I can't see anything because my browser blocks content hosted on google, this project evidently doesn't even try to walk its talk.

You might find the explanation here more helpful: http://hyperstition.al/post/urbit-an-introduction/

The project website isn't trying to be decentralized, it's trying to share information easily with people using existing services.

Oh god no. Urbit is not less complex or easier. It's piling on all kinds of extra superfluous nonsense in service of an ideological project, not for technical reasons.

Urbit stack is ~30k lines of code, that's low enough to be grasped by a single person. What's hard is learning form scratch completely new platform, but that's deliberate and is not such a problem if it manages to deliver what it promises.

Somehow I feel that accusations of being ideological come from people, who themselves are ideologically oposed to Yarvins work and fail to consider Urbits network protocol purely for its properties. Idea of giving network addresses value, cryptographic ownership and creating a hierarchy of addresses resembling functionality on the network (users, ISPs, etc.) seems perfectly resonable.

From the last time this came up on HN: https://news.ycombinator.com/item?id=22859156

> the core of the whole system (the Nock language) has "loobeans" instead of booleans in which 1 represents false and 0 represents true at which point I nope'd out pretty hard.

>> Being contrarian is not the same as being clever, and this is very much not clever.

Do you think this convention would be that hard to get used to, so that the whole system should be disuqallified?

Yeah, no disagreement from me there. That bit is dumb and he admitted it was a mistake to do that.

It doesn’t mean that everything is bad.

I was skeptical initially too, but was curious enough to check it out. I was surprised when I did.

It's a small community and product is still in pre-widespread adoption stage (though far enough along to be useful for people to play with): https://urbit.org/understanding-urbit/roadmap/

My quick overview here: https://news.ycombinator.com/item?id=23397725

Also (not excusing the bad design) 0 is often the success error code in Unix for the shell and 1 or something else is often error.

So it’s not entirely without precedent, but it’s still an unnecessary and confusing inversion for true and false.

As long as a complex system has the complexity abstracted away it can still be simple for users.

The modern (non-Urbit) tech stack is extremely complex, but a lot of that (though not enough) is hidden from the end user.

In a lot of important ways the Urbit stack is a lot less complex than the existing modern tech stack, but it’s still early on and the abstraction for regular users is still in progress. Being less complex than the modern tech stack (by designing from the feet up by first principles) is one of the main reasons the project exists.

It’s just rarely done anymore in modern computing. It happened more in the 60s-70s when people would experiment.

I understand the skepticism - I think it’s a cool project, but there’s still work to do before it’ll win over average users.

Luckily the project is invested in long-term planning and design (with solid short term execution and feature shipping/iteration) so it’s improving quickly.

Urbit is extremely cryptic but I do like the idea as long as they don't include any 'blockchain' nonsense.

Bad news for you: https://urbit.org/blog/urbit-and-the-blockchain/

Urbit is now not a good alternative to decentralized applications.

The blockchain is only used for handling the IDs, but the core project functionality does not rely on a blockchain beyond that.

Reply on news.ycombinator.com