ActivityPub isn’t great, but anything is going to look bad when compared to a utopian pipe dream.
Im looking forward to a better comparison in Q4 or Q1 2021 when Urbit is more usable for regular people though. As of now, I don’t suspect that anyone will pay $10 for a VPS and $5 for (reliable) star service.
It should be noted that when I critiqued ActivityPub here a couple months ago, a main developer of the spec was very supportive and already on top of the improvements that I was hoping for (such as being less reliant on servers, more true decentralization, etc)
It's not a pipe dream. I'm using Urbit on a daily basis, and this year I'm having a blast with it - in particular because of high barrier to entry, the conversations there are really great with the most geeky and motivated people.
Lots of features are there, it got faster, more stable.
There's no need to pay for anything. Urbit planets are usable from desktop, and don't really need to be turned on all the time. Urbit should have no problem eventually running from a phone, which is like 99% online device for me anyway.
Federation of services running on Unix machines is just not going to work. That's how Internet used to work, and it was not practical so it degenerated into what we have right now.
Full disclosure I have been on Urbit (probably talked to people for 3 hours in total and spent 12 hours total setting it up and tinkering with Hoon).
Great conversations with interesting people has exactly zero bearing on the quality of the project. I have had great conversations on IRC.
You do need to pay for a planet of course (because of the “skin in the game” concept that urbit people are obsessed with). But I’m speaking of paying for Star service though. Yes, stars are free now, but the whole point of Stars is that they will not be free. Otherwise they will serve you ads or do some other things to be sustainable. And in my experience, stars should be a paid service even today, because the connections are so unreliable that it is just beyond frustrating.
If Urbit is a toy, it’s a really good one. If it’s a tool for communication, it’s terrible and you’re better off with literally anything else. I mean, it doesn’t even have notifications so idk how it’s useful for a communication today.
That’s why I’m saying to just wait til Q4 or Q1 and see where it’s at.
Why are you poo pooing federation when Urbit is a federation of Stars? What are you even saying? Every decentralized mesh network today still requires a federation of relay servers. Unless you’re talking about going around the whole networking stack as it is today. But in that case, nobody would need federation and Urbit wouldn’t be the only solution for that either.
My comment is overly long already, but I feel compelled to also say that Urbit is bad for privacy today as well.
> Why are you poo pooing federation when Urbit is a federation of Stars?
Because I own my own identity, and can move it to another star quite seamlessly without losing my ID and all connections I've made. And I'm not a second class citizen in the system because of that.
With Federated protocols I can create my own instance, sure. Then I do own my identity, at the cost of all the administrative overhead and being a second class participant. Spam and abuse are an unsolved problem at the core of the legacy Internet and all these federated protocols can't do anything about it.
If I own an instance I have to deal with it (spam and other abuse) myself, and constantly fight the fact that other people think I might be a potential abuser myself, which puts people on minor-instances at a disadvantage.
So natural economic and social forces constantly push the system to centralize and that's exactly what we're seeing with everything on the Internet.
Decentralized Identifiers / accounts are being added to every federated spec that I know of. So I think everyone is aware of the issue there.
Regarding spam and abuse, what deters that on Urbit? The cost of a planet? Planets are only as valuable as the Social capital that they’ve built up. If a planet has no mutual “friends” with me, I’m not trusting that planet, period.
The planet starts with a value you paid for it. Which makes the owners at least start with some skin in the game.
I'm looking forward to see how decentralized identifiers will work in practice. It's a step in the right direction, though I don't think it will be enough on its own.
The real spam prevention will always go back to reputation and web of trust. Even people writing for urbit understand this. The idea that $10 solves everything is silly. Ultimately the $10 thing (and even the limit to the number of planets) is impotent in the face of natural reputation.
> By convention, booted addresses are expected to have some existing reputation outside of their name alone, since they’ve been used on the network. Reputation, good and bad, comes in many forms. Did the address operate any useful infrastructure? Did it get placed on any blacklists for spam or abuse? Did it simply send and receive messages? The ability to programatically track reputation is still in its infancy, but we expect the tooling to develop as urbit grows.
As somebody who went to all the trouble of:
1) Converting cash to Bitcoin
2) Transferring the Bitcoin to my browser plug-in wallet that was the only approved way to pay for the “planet” or whatever
2) Purchasing the planet and somehow linking my ID with their registrar/account info
3) Downloading and trying to installing the software despite several dependency issues with the stock Ubuntu (or whatever they recommend) VM.
4) Finally getting it working, being greeted by a truly bizarre user interface, I figured out how to get into chat. Joining the main channel, my screen was flooded with out of order chat messages.
In a project that’s been around since 2012.
I was really excited to experience the new wave in computing. I found smoke and mirrors. Cool idea though.
Im looking forward to a better comparison in Q4 or Q1 2021 when Urbit is more usable for regular people though. As of now, I don’t suspect that anyone will pay $10 for a VPS and $5 for (reliable) star service.
It should be noted that when I critiqued ActivityPub here a couple months ago, a main developer of the spec was very supportive and already on top of the improvements that I was hoping for (such as being less reliant on servers, more true decentralization, etc)