Oddbean new post about | logout
 Nostr’s super-power is interoperability, yet most nostr services tend to build complete experiences rather than “the best micro-app to achieve X”

Is this due to a belief that each app should be a complete end-to-end experience in itself? a lack of coordination among devs? something else?

excerpted from a phone call today with @Max 
 idk even where to start

the biggest problem of nostr is the inconsistencies of JSON codecs

with the exception of web apps and nodejs backends every case is one implementation vs another

IMO, nostr will start to mature when it uses a modern encoding and API

events, the canonical encoding, the filter and the envelope are all really ragged crappy designs that are really badly underspecified

thus, it is common at this time, to build the two things at the same time, to get the best mesh

this is the thing

when you are a programmer, you don't build the input and output separately, they are part of the same system

but that's exactly where we are 
 This is what I appreciate about Amethyst's approach. Enough integration of other stuff to know it exists, but not enough to fully utilize it. It encourages exploration.  
 Id call it portability.  You can move from one app to another *within nostr* quite easily.  But interop with the rest of the web is low.

People are in the habit of building big apps, because that's what employers ask for, that's what pays, and that's what has been taught to a whole generation.  So there's an overhead for most projects, and you can then do alot.

Micro apps are new because people are not used to *composable* micro apps, they have always had to do it all.

If someone builds a profile manager app, the whole of nostr gets a profile manager app.  This is a new mindset for most people.  Im one of the lucky ones who has seen this before with the Solid project, and webid before that, so I have had 15 years of anticipation of this happening.  Glad to see it finally arrive, after some stops and starts.

As an aside, this is the vision tim berners-lee had for the final version of the web, which he always viewed as a work in progress.  Interoperable + portable micro apps.  We might be half way there! 
 from some replies on the thread it sounds like there’s not enough standardization of interfaces to make micro apps composable. Have you written anything about how you think it should be done? I suspect that might help people discuss the tradeoffs or coalesce around a specific way to think about composibility of micro-apps