The first Urbit apps were meant to mimic the most useful services on the modern web: Discord, DMs, video chat, a friends list, etc. I'm not against building these apps, but the reality is that these are just less reliable versions of their Web2 counterparts. Urbit doesn’t enhance my experience of Discord or DMs - if anything, it tends to get in the way and make it worse. Instead I propose that we should use Urbit as it was intended: as a personal server.
Most Urbiters understand that Urbit needs to be self-contained, but they draw the wrong conclusions from this principle. Urbit is a personal server, not a login: there is no reason why anyone needs a planet just to read content; you only need a server if you want to serve content. This should be obvious, but everyone is building apps as if Urbit is just another login to another Web2 service - a planet is required to read everything.
This has killed Urbit's retention because no one is willing to muck around with their VPS just to use a discord clone. If Urbit were a publishing tool, then we don't need to lean on Web2 network effects to grow. People won't need to take it on faith that Urbit's community is good - you'll be able to show it by sending them a %blog link.
With %blog, Urbit became a good publishing tool, but it can become the best publishing tool. Urbit can rectify exactly what went wrong with Web1 and HTTP, and build a true decentralized hypermedia web.
The first problem with Web1 is that old links go dead; the second is that there are no backlinks - i.e. you don't know the pages that link to your page. Urbit fixes this: if everyone is running the same server code, we can establish a new Urbit-side protocol to implement the back-link and permalink:
%blog, your Urbit can send a poke to let them know that they've been referenced%blogs will all have a list of all associated backlinks, it can let all those references know that your ship has changed it's URLThese two features make Urbit+Web1 much more discoverable and much more robust than any previous version of the web. We can use these primitives to build up a decentralized Web2.
The main innovation in all of Web2 is the feed. People enjoy consuming content in a linear, top to bottom list. We can build a system on Urbit where everyone can publish multiple feeds. For example, a "personal" feed (ala a Twitter/Instagram/FB profile), which contains original content and "reblogged" content from other people's feeds, and a "timeline" feed of everyone you "follow".
This will be a similar experience to "Web2" as it exists, but without any of the drawbacks. All of the code runs on personal servers; every detail is owned and customizable. This means you will not be locked into communist design, algorithms, or speech; there will be no ads, no tracking, and no dystopian "attention economy".
This is the future we can build with Urbit if we use it as a server instead of a login. Two hooners could make this happen in just 6 months from their garage. If this is you, talk to me on Urbit
~dachus-tiprelI'm not against building centralized web2 apps on Urbit. 4chan is a good example of something that will fundamentally never work if we use Urbit IDs as a requirement to post; 4chan isn't 4chan if it's not anonymous. Instead, we should be building 4chan on Urbit as a centralized Web2 site. The only difference between a normal 4chan clone and a 4chan clone hosted on Urbit is that Urbit will dramatically decrease the barrier to running a 4chan clone. There are probably more examples of a centralized web2 apps that should stay centralized, but powered by Urbit.