[tptacek]

Right, but you could do that with any P2P overlay network with any embeddable VM. Why use the very unorthodox one that Urbit came up with? I really think that's the question people are asking when they say they don't understand Urbit. Not, "why would we want to do what this system claims to let us do", but rather, "why would we want to do it that way?"

[chc4]

Do you have something in mind? Using GNU Social as a convient whipping-post, it's built on PHP and MySQL and Salmon and PubSubHubbub and...

The point of Urbit is to be a cobherent, self-contained platform. Every other decentralized app I've seen had to reinvent several wheels and can't interact with each other because of it.

Urbit has a typed versioned filesystem exposed as a global namespace, typed RPC with a diff/patch system, an identitt system, forces apps (and the e tire platform) to be purely functional for easy saving and crash tolerance, and other fun things. It's like shitty OSX - it has a /vision/, even if some parts don't make sense.

[tptacek]

The problem I think a lot of people have with this system is that they agree with you: the underlying concept of an overlay network with abstracted addressing and an embedded programming language VM is straightforward, and may indeed be useful. Since all the core technological ideas in this system are simple enough to be undergraduate projects, why not just wait until someone builds one out of conventional components?

You get the irony of complaining about how every competing design has had to "reinvent several wheels", right? The system you're advocating reinvented ASCII.

[chc4]

My dude, how can you read code out loud and not understand the draw of having single-syllabul names for all the weird characters in ASCII. I don't have many of them, but this is a hill I will die on even if Urbit falls off the internet tomorrow. Luckily, you don't /actually/ have to memorize those. They aren't used if you aren't saying runes out loud.

There's a difference between reinventing ten wheels ten different ways for ten apps, and reinventing one really big wheel. It has to be self-contained enough that other apps can use it without having to roll their own anything, so all of them has the same API.

One of the problems with "just use a different P2P VM" is that it probably wouldn't solve the problem of having a server your mother could use. You'd probably still have to install and setup Apache and MySQL and ElasticSearch and whatever, even if someone implements the perfect transport protocol and platform. For Urbit you just run it, which can be hidden behind a pretty "click to start" button that pings our to AWS. Hence, one big wheel.

If you do see an Urbit-like project that has the same goals, I would be ecstatic to hear about it though. I've basically had "urbit but different" on my TODO list since I first saw about it.

[munin]

I think urbit started out as a pomo critique of modern academia and tech, with heavy jargon and wheel rebuilding. Then, some other people started taking it seriously. I’m not sure if that is the best or worst outcome for this kind of joke.

[jolux]

Moldbug hates pomo though. Pretty sure the dude doesn't read anything that was written after the year 1900.

[munin]

Don't you know, the Moldbug persona is separate from the persona that develops Urbit!

[jolux]

What the heck is that supposed to mean? "Persona?" Fine, Curtis Yarvin has said that he only reads old books.

[Karunamon]

Sorry, Pomo? Can’t get a good definition via the googles.

[grzm]

Post modernism

[jacobush]

That is so pomo.

[wmf]

Coherent, self-contained platforms seem to always lose in the market, so it's not even clear that's a good goal to aim for. That's why I prefer Sandstorm; it provides an evolutionary path from normal Linux.

Getting back to your question, Blockstack is claiming to provide virtually the same personal cloud features as Urbit but it's only boiling half the ocean.

[defen]

Theory: only a system as esoteric as Urbit could generate the necessary asabiyyah to actually pull this off. It's not just a tech problem, it's also a social engineering problem. As in, how can you self-select for only the smartest, most dedicated people to build your system? Well, allow me to introduce you to Hoon https://urbit.org/docs/hoon/

[likeclockwork]

Well, there's a question of who I want engineering my social reality in addition.

I think there's quite a lot of selection going on and the criteria are much broader than just just "the smartest, most dedicated" engineers.

[ineptech]

> Right, but you could do that with any P2P overlay network with any embeddable VM. Why use the very unorthodox one that Urbit came up with?

Urbit claims to be uncrackable. Not "secure when well-configured and patched regularly" uncrackable, but "Put one on an EC2 instance and tell the world there's 50 BTC inside it and come back in five years, or fifty, and your money will still be there" uncrackable.

That's its selling point - that you could run server apps on it (in some hypothetical future where someone has written some useful apps) and not have to be a part-time sysadmin like you would if you ran any sort of *nix server.

I'm not saying that's true, I'm just answering your question. More details on why they claim it to be that way here: https://urbit.org/blog/2017.5-frozen/

[sanxiyn]

I am very interested to hear about other competing P2P overlay network with embeddable VM of similar maturity.

[tptacek]

"Of similar maturity"? What does that mean? What is one mainstream application --- let's define "mainstream" here as having more than 10,000 active users --- that is built directly on top of the Urbit stack?

[sanxiyn]

By maturity I meant something like "continuous opeartion for a year with cryptographically signed live update (including cryptography themselves, implemented in VM)", which Urbit did. But okay, I am also interested in competing projects of any maturity.

[tptacek]

Because implementing cryptography in a bespoke VM is an important feature? Why? That's the kind of thing cryptography engineers try to avoid.

[sanxiyn]

You haven't named even one competing project. I think it is a good justification for interest in Urbit.