I believe the future of a free internet lies with personal web applications. A personal web application is self hosted, often self-developed, has few power users other than the owner himself, yest still provide public access. Elixir is easier to get started (Phoenix generators), easier to reuse other people’s code (hex.pm), and easier to host (self-contained release, fly.io); it is a natural fit to develop you personal web applications.

You can’t really fix something with tech that most people don’t see as broken. There is a reason people flock to the biggest sites and its not because the technology for self hosting smaller sites and forums doesn’t exist.

When I was younger I modded a lot of forums and there was a pattern you’d see on most forums where at some point a group of people would get annoyed at the modding or some other rules and declare that they were leaving to start their own forum with better rules and they’d post a bunch about it then disappear for a few days to chat between themselves in their new forum. No one would follow though because of switching costs and because the conversations and people were already where they were and so after a week or two the people that left would inevitably come back.

We’re seeing the same thing with the push for web3 and distributed web apps now as that. The % of people that care about this stuff is a rounding error so their sites are dead and thus when people do decide to go have a look they see a wasteland and bounce quickly.

That’s absolutely correct save for several exceptions. Calendars work exactly the way self-hosting-people do propose. I host my calendar, G hosts yours, and yet they are both fully functional, because of handy interchange format. Oh, wait, even mail does work that way.

The issue, in my humble opinion, is not people are stuck to their facebooks, but the interchange which is broken (intentionally by bigcorps, but also because self-hosters are not always smart enough to understand what you’ve just said.) Fediverse tries to fix it. To some extent it even succeeded, and one might get to bluesky fellows from self-hosted mastodon instances.

But in general, unfortunately, nobody cares. Can I connect my blog (supporting fediverse shenanigans) to that forum?—Nah. Why?—Because nobody gave a shot.

That said, I think we need a working interchange before self-hosted-world. Then the migration of some of us happens and there might (or might not) be a snowball effect.

This is an exciting topic I meant to talk about in my original post!

inkandswitch.com/keyhive

• Keyhive is a project exploring local-first access control. It aims to provide a firm basis for secure collaboration, similar to the guarantees of private chat but for any local-first application.

I highly recommend for anyone interested to read the whole linked article, I can’t really do it justice. Here’s some pretty pictures from the notebook

image1536×1246 88.1 KB

image1198×1094 78 KB

image1558×1000 38 KB

image1466×1030 27.3 KB

image2693×1601 128 KB

image2334×1438 97.8 KB

image1738×960 60.5 KB

image1942×2013 106 KB

• In this lab notebook, we’ll share snippets of our findings as we explore the problem space and prototype potential solutions.
• A deep-dive explainer on Ink and Switch’s BeeKEM protocol

The elixir comm is most definitely missing out on “web 3”, I understand many people dont like crypto and the usual languages used in this space are golang, java, rust & c#. Elixir can play in the c#, golang, java arena side of things and can help in some aspect for the rust side of things

For example in ethereum there is something called client diversity , the aim is that if you have the same protocol & vm programmed in different programming language’s, if one programming language has a glaring flaw in the future like an rce, since the control of the consensus is spread out between multiple implementations in different PL mitigation would be trivial. I always thought an elixir implementation would show the world how great elixir is at running network applications

Another point, some people in the evm community felt the “standard” golang implementation go-ethereum was slow at certain tasks and then attempted the herculean tasks of rewriting everything in rust, to fit there high performance checklist, I think it took them ~3 years to get to 1.0 and I thought if the standard implementation was written in a beam language and the high performance parts were brought in with rust NIFs how fast would there time to market have been?

I think elixir is perfect for the crypto/decentralized space, now granted ethereum is on the slower end of things with ~12s block times, you have something like solana that builds tx in 200ms, but again how much of the code could’ve been simplified if you handled all the network stuff to the beam and then rust NIFs for the heavy lifting stuff? Alot of decentralized protocols have slow build times