2 ms·
Yeah, it's hard for me to believe his views are completely separate from the project. Note how the Urbit universe is organized is highly hierarchical, likely be
by mempko 5y ago
Yeah, it's hard for me to believe his views are completely separate from the project. Note how the Urbit universe is organized is highly hierarchical, likely because he believes this is how things must be.
When I wrote Firestr (http://firestr.com http://firestr.com) back in 2013, which is similar to Urbit in many ways like being P2P and being able to download and share apps, It's clear to anyone who knows me my beliefs drove the technological decisions. For example, I am anti-hierarchy. So your Identity on Firestr is completely decentralized. There is no sub-ownership. There isn't even a central repository of identities. Identities are shared in a peer-to-peer way. Also, it doesn't attach itself to some bigger system like Ethereum and running it is free.
I can easily see Borges's ideas in Urbit and it's a shame as we need more decentralized software.
- _6hmp 5y agoThanks for sharing your project as a counter example. To be clear I'm not damning Borges here, just suggesting that naming the company after that particular story gives some weight to the idea that we should consider the author's values and goals when evaluating it.
- olah_1 5y agoFirestr looks great! It looks like a cross between Nostr and Gun. Regarding the architecture, where does the data live? On the clients? How are you achieving a multi-writer functionality? Meaning, how can two people edit the same chat data structure?
- mempko 5y agoThanks! Unfortunately I haven't done much with it in a while, just enough to keep it running on modern linux systems, but it still works nicely. (code is here: https://github.com/mempko/firestr https://github.com/mempko/firestr) Data lives on the clients. How multi-writer is handled is up to the app you write but I have built in support for vector clocks. So for example, the built-in app editor uses vector clocks and a merge algorithm. I thought about adding CRDTs for making syncing data structures easier. Maybe there is someone out there that is motivated...