How is this different from many other approaches like Sandstorm, or any other effort to "decentralize the web" by giving people their own servers?
I'm asking in terms of actual benefit it provides, not the technical part. I think at this point it's pretty clear that these efforts have failed due to lack of traction in that department (not enough people care about "owning their own server" or "controlling their own destiny")
It provides an application framework and reinvention of the world to have decentralized apps work, instead of just providing packaging apps like Sandstorm.
Urbit has an identity system with limited space to prevent Sybil attacks. The network pretends to expose exactly-once networking that Just Werks so that apps don't have to worry about it. All apps are purely functional state machines to make it easier to handle them and transparently upgrade them from code updates. Sending RPC messages so your own apps I exactly the same as sending them to remote hosts, the same with reading files, with peer discovery happening for you by the kernel.
Sandstorm does a whole lot more than "just packaging apps", and compared to most other things in the space is ridiculously ambitious... but Urbit does indeed go far further.
This doesn't answer my question. In fact, it's the opposite of what i asked. I said:
> I'm asking in terms of actual benefit it provides, not the technical part. I think at this point it's pretty clear that these efforts have failed due to lack of traction in that department (not enough people care about "owning their own server" or "controlling their own destiny")
I'm not really sure what else to say, then. Can any project with similar ideas answer anything other than "no, nothing different"? The idea for all the different projects is to make self-hosting sufficiently easy and usable, just different ways to come about it. You could ask the same thing about IPFS or even Bitcoin, and the disregard it when you don't get a yes.
Bitcoin does have a use case--you can buy drugs with it. I'm not going to judge whether this is moral or not, but the point is it's serving a previously unmet need. There have been tons of attempts at cryptocurrency before Bitcoin, but Bitcoin came and it just worked to serve this purpose. And it enabled whole new type of transactions online. Currently it's definitely not mainstream currency but at least it has a clear set of customers from which they can expand.
Which means, it's not like Bitcoin invented a non existing demand. It already existed, and Bitcoin came and served the previously unmet demand.
My question was whether this urbit technology is something like Bitcoin where it serves a previously unmet (AND significant) demand. All previous attempts at this approach failed NOT because the technology was bad but because not enough people used them, which is different from Bitcoin.
Urbit is a network where you can, by default, trust the other party you're communicating with. It does this by making identities hierarchically controlled; if you behave badly, the rest of the network will blacklist you and your master. Thus, your master has a strong incentive to police your behavior.
tl;dr; it should be a spam free network.
Note that this is already a mechanism used in the public stock markets; your broker polices your behavior because if you screw up, the broker is on the hook for your losses, and the clearinghouse is on the line for the broker's losses.
I think the answer, the unmet need Urbit is trying to meet, is "I wish I could install apps on a server as easy as I do on my phone." On your phone, the security comes from a crack team of engineers paid by Google or Apple or MS to make sure no bad apps get in, and on Urbit it would come from Urbit's architecture.
But to be clear, none of those apps exist yet; at this point it's just barely functional enough to start developing real apps.
Honestly? Probably not then. You can make social networks that can't be censored, like GNU Social, or other decentralized apps easily. But then you have the problem you mentioned of "but who actually wants that". All the things that Urbit can do is already possible, just not easy, and each product in the same sphere has to reinvent certain subsystems to do it. Nothing is stopping you from writing Facebook-on-BitTorrent. The identity on it wouldn't be able to talk easily with Twitter-on-IPFS, but that can be fixed with elbow grease too. It's just easier to have them both powered by frameworks that provide stuff from them both, and preferably easier for your grandmother to join.
Sorry, I kinda lost the plot, didn't I? I'm not actually a believer that Urbit will do any of this, or that it's answer to the question of how to get people to switch to a decentralized platform. I just think it's a neat project to play with. But, fundementally, it doesn't do anything new, just connect several things together.
> I think at this point it's pretty clear that these efforts have failed due to lack of traction in that department (not enough people care about "owning their own server" or "controlling their own destiny")
Er, the whole effort has just started. What's happening right now is that people are experimenting with various UX'es. Many are failing (as expected) but I am confident a winner will arise. Just give it some time.
I'm asking in terms of actual benefit it provides, not the technical part. I think at this point it's pretty clear that these efforts have failed due to lack of traction in that department (not enough people care about "owning their own server" or "controlling their own destiny")