>I've been playing around with Urbit and it's pretty cool.
Thanks for the suggestion. I went and installed it and while it was booting (~10 minutes on a 2 core 4096MB VM), I poked around looking for more information and found some of the "knee-jerk hate in HN comments"[0] you mentioned, as well as a hit piece[1] on the founder/creator (whether the arguments therein are valid is something you need to decide for yourself) and a more positive take[2] on it as well.
While it's certainly an interesting bit of design and coding, with a laudable goal (decentralization of human interactions online), the functional model has all the hallmarks of a pyramid scheme, with everyone lower down paying rent to those higher up.
Despite the decentralized/p2p nature of Urbit, it's inherently hierarchical and seemingly designed to extract rents from those lower down the pyramid.
And since all the higher-up slots are already occupied, this seems more like Amway than a decentralized, egalitarian network.
I think you’re wrong, but I also understand your skepticism at first look (I asked similar questions).
The reason for a small cost associated with permanent user IDs (planets) is to combat spam and encourage reputation building without requiring real names if users don’t want to.
It’s a clever approach to this issue. One reason for the incentive to centralize on the existing net is to combat spam (since its zero cost to spin up millions of accounts to spam with, you need clever anti-spam which tends to cause centralized services to form).
The nodes higher up just route traffic, they don’t own any user data or do anything else. Users can “escape” to any of them for a provider so the “stars” (infrastructure nodes basically) are incentivized to provide good routing service for users in order to keep them.
The top of the hierarchy “galaxies” are basically governance nodes that allow changes to network policy based on a vote. If they became a problem stars could push back or jump off the network.
If you start with trying to come up with an incentive based design for a new network model that empowers users, but can actually work I think you start to see why these ideas are interesting.
On the existing net a lot of this stuff exists in an arguably worse form with less clarity (people route traffic, they also sell your traffic information, you have little control or choice over it).
The existing net also incentivizes centralized applications that collect user data. p2p open protocols don’t solve a lot of these issues around spam or basic usability (they’re DOA if a regular user has to run their own server).
Urbit’s design allows network updates to automatically get picked up across the entire network. Users control and own all of their data. It’s p2p by default and that complexity is invisible to users. It’s backwards compatible, runs on unix in a vm, but could run on its own custom hardware in the future.
It could allow people to have their own computing environment where they can send things like photos directly to others without a middleman like FB. The design means that software that runs now should run in ten years.
I think there’s a lot of potential, the tech is real (not vaporware), it’s open source, and you can play with it/talk to the community of people on it.
I have no idea if it’ll achieve what they’re trying to do, but if anything were to succeed in this space I think it’d have to be something like Urbit.
---
Re: The first article you linked, I find CY's politics/writing (from what I've read) to be contrarian and wrong in a similar to way to Peter Thiel's politics. That said, most people hold inconsistent views and people that think independently can be very wrong in one area and very correct in a different area (and people generally are wildly inconsistent in their views/accuracy about everything). Thiel is often contrarian and right about investing and building technology companies even though (I think) he's contrarian and wrong about politics.
While it can be useful to keep someone's political beliefs in mind when evaluating something just to be aware of potential motivated reasoning, I don't think that should allow you to dismiss everything else out of hand. Someone can hold both really good ideas and really bad ideas at the same time - similarly someone can hold true and false beliefs simultaneously.
When it comes to the author's example (Thiel and Palantir) - I find their framing to be misleading. If they're applying this kind of over-simplified analysis there then I expect they're applying it elsewhere. Their essay is mostly an example of their own cognitive bias - they already have a position and they are cherry picking evidence to support it. The reality is more complex and nuanced than what they suggest.
Thanks for your point of view. I really appreciate the time and effort to present your view of the Urbit ecosystem and the value it provides.
As I said, I really like the idea of decentralized network services[0].
I understand the motivation WRT a mechanism that will discourage spam and other garbage. And while I mostly focused on Urbit's similarities to a pyramid scheme, that's not really my primary concern. Rather it's the hierarchical nature of Urbit that seems more problematic to me, with the tiered rent-seeking is another, less important aspect (although still negative, despite the innocent claims of spam prevention) of it.
What's more, I'd want to use the technology for my own (admittedly narrow) purposes, without others having the power to shut me down or blackball me -- a possibility that a hierarchical model doesn't preclude.
As for the politics/philosophy of Urbit's creator, that's not so important to me as long as I can use the technology the way I wish.
That said, there are aspects which seem troubling, not least of which is that the founder, despite his apparent departure from the scene, still owns a significant portion of the hierarchy's top level, potentially giving him significant power over the governance of the Urbit universe. Which may or may not be an issue, but a flat, fully peered model avoids that issue completely.
>Urbit’s design allows network updates to automatically get picked up across the entire network. Users control and own all of their data. It’s p2p by default and that complexity is invisible to users. It’s backwards compatible, runs on unix in a vm, but could run on its own custom hardware in the future.
>It could allow people to have their own computing environment where they can send things like photos directly to others without a middleman like FB. The design means that software that runs now should run in ten years.
Aside from automatic network updates (a useful feature indeed), I wonder what value Urbit has over a platform such as Diaspora[1], which, assuming I run my own pod (a similar situation to Urbit) provides me with full control over my data, as well as federation services and strong controls over the content I allow into my environment.
The Diaspora model is completely free (both libre and gratis), doesn't have a hierarchical structure and provides a pretty good UX.
Please understand that I'm not rejecting Urbit, I just don't really see the value of it over other platforms that provide similar services without financial entanglements or potential issues with those "higher up" in a hierarchy.
As a technical person, the Urbit technology itself is pretty cool, but given its implementation and structure, it's difficult to see it gaining wide acceptance.
Whereas (using my previous example) if/when Diaspora is packaged in a way that most folks can easily install/configure it, it's likely to see much broader acceptance.
I may well play with Urbit a bit more, but AFAICT, its utility is limited using it as a "comet" rather than purchasing an ID.
Thanks for the questions - I'm not an expert on this stuff either, it's just something I've been messing with for fun during covid.
> Whereas (using my previous example) if/when Diaspora is packaged in a way that most folks can easily install/configure it, it's likely to see much broader acceptance.
I've basically come to the conclusion that this is impossible to do successfully at scale, or at least impossible to do while keeping the original p2p intent alive on the modern stack. Attempts to do this fail either outright or by reverting back to being centralized (at best they retain a small core of highly technical users). The context that causes these attempts to fail is what Urbit is trying to fix with its design. It remains an open question whether this will work, but I think there's more of a path for it with Urbit. I think things like the ability to push updates across the fleet is one example of a critical feature that fixes a common issue with versioning in federated systems, but there are some others.
> Aside from automatic network updates (a useful feature indeed), I wonder what value Urbit has over a platform such as Diaspora[1], which, assuming I run my own pod (a similar situation to Urbit) provides me with full control over my data, as well as federation services and strong controls over the content I allow into my environment.
Urbit is more of a platform ("Overlay OS") than a more narrow open social media protocol (diaspora, mastodon, etc.) - you can build applications on top of it that take advantage of its ability to route encrypted data between users. Standardizing the stack makes it easier to reason about and easier to build/run these applications for all users that want them. You can't really do this on the modern tech stack without armies of people keeping things up to date (which creates a strong incentive to centralize). Urbit's design allows for decentralized applications (and Urbit itself) to actually work and stay working.
"The state of your Urbit OS is a pure function of its event history. It’s auditable, inspectable, repeatable. You can actually trust it. Writing decentralized apps becomes vastly simpler than in the old world, since every node computes exactly the same way. The entire Urbit OS stack, from programming language to applications, is upgradeable over the network. For ordinary users, this makes for almost no system administration."
Urbit's light hierarchy I think is necessary for this to work and solves most of the hard problems around decentralization in a way I think is clever and pretty light-touch. I do think they could be better about the ownership and governance transparency (how much is owned by any individual), but I think they're working towards this: https://urbit.org/blog/governance-of-urbit/
> What's more, I'd want to use the technology for my own (admittedly narrow) purposes, without others having the power to shut me down or blackball me -- a possibility that a hierarchical model doesn't preclude.
You're able to escape to different Stars if you need to so in practice this shouldn't be an issue (and Stars are incentivized to keep their users happy). It'd be more comparable to your ISP blocking access - they could do it, but they're not likely to.
- https://tlon.io
- https://urbit.org
It tends to get a lot of knee-jerk hate in HN comments, but I think it's a from first principles approach that could actually work.
At a minimum it's a pretty cool/ambitious project and it's been fun to play with.