I’ll probably run out of air before folks get this but, hey, one more time for those in the back:

If you’re creating “alternatives to #BigTech” that scale vertically, you’re doing it wrong. You’ll never beat Big Tech at vertical scale. They invented the game. They’ll destroy you at it every time. And say you somehow do manage it… congratulations, you’ve just become *checks notes* Big Tech. *golf claps*

So what do you do?

You limit vertical scale and #design for horizontal scale.

#SmallTech

What does it mean to design for horizontal scale?

It means limiting the growth of individual instances of a service (e.g., number of accounts, etc.) and investing in ensuring that as many different people and organisations can run their own and that they can all talk to one another and interoperate.

It means keeping the power relationships between instances as level as possible.

Take it to its logical conclusion and you get the #SmallWeb: instances of one.

https://ar.al/2020/08/07/what-is-the-small-web/

#SmallTech

@aral The number of people who have the money, time, and expertise to run servers prohibits this from ever reaching critical mass. Check out what’s happening recently in P2P, we just shipped https://sockets.sh, that’s step one. This month we’re open sourcing the P2P libraries that make it possible for users to communicate without servers, even if people are offline. No servers == no landlords.

@heapwolf All of those can be solved in various ways. Currently, with pre-alpha code and zero investment, you can set up a small web site that is maintenance free in under a minute with zero technical knowledge for less than €10/month. All of those numbers can be improved in the future with investment from the commons but anything beyond that is premature optimisation at the moment.

Regarding sockets (congrats on shipping, btw): how do you handle offline messages if not without always on nodes?

@aral servers are and always will be landlord-tenant oriented, this is antithetical to a free and open internet. I don’t object to commercial ventures but the fundamental stack must be absolutely free and open. No black boxes. No gate keeping. No rent extraction.

We designed a packet relay algorithm that was originally intended for partition tolerance, however its design also makes it suitable for *significant* partitions (hundreds of hours at the moment)

@heapwolf Ah, so nodes also act as relays for the system at large (assuming by storing and forwarding end to end encrypted content-addresses packets? So more nodes = more reliable offline support?

@aral correct, it’s highly nuanced but we have a huge FAQ on the way that includes a lot of maths and models. Even as a (relatively) small network it works nicely already. My whole team can be offline and we still send chats and files reliably. E2E FTW 🙌

@heapwolf Best of luck with it :)

(What I’m working on takes a much simpler approach. But they’re definitely compatible. And I’d love to see efforts like yours succeed.)

🙌