Tonk

Tonk

@tonk.xyz

Data spaces you read and write to with a link

181 followers58 following48 posts

Project Cybersyn, designed by British cybernetician Stafford Beer, aimed to manage the national economy as a living system that could sense and respond to change in real time. Tonk is a system that syncs and computes data across multiple computers, big and small.

A few hardworking agents just saved their work on to the Tonk network

Partial map of the Internet based on the January 15, 2005 data found on opte.org. Each line is drawn between two nodes, representing two IP addresses. Tonk runs a network on top of the internet for sharing data as link-addressable space.

Punch cards (1935). CD-ROMs. Digital pets (2004). Personal computers used to carry personal software. Tonk is a system to carry small software.

Is this what coding looks like in 2026? Who knows! Show and tell starts in ~1hr in Tonk Discord :)

Community Memory, the first public computerized bulletin board, was accessible by a terminal installed in Leopold's Records and connected to a timesharing computer in San Francisco. Tonk is a system for networked software, inhabited by friends, collaborators, and group chats.

Tonk reposted

Loving ability to see members of the space in @tonk.xyz which also reflect topology of responsibilities. It also happens to be a representation of @ucan.xyz delegations chains. Space delegated ownership to founder who then delegated access to other members

AI is becoming an operating system for your life. Will that system be yours? Or will it be rented to you? We believe user-sovereignty provides the better experience. If that resonates, join us in the Discord and together we'll make it a reality. --> discord.com/invite/hBPQ9...

If you’re like us, it’s not unusual to create a new piece of software in just a couple hours. With AI, we’re going from idea to working tool over and over again every day. Dashboards, workflows, little utilities, that type of thing. But what happens when you have to share it? 🧵

View full thread

9/ Tonk will make that possible. We’re getting close to something we can share with you, and we’re really excited for that moment.

If you’re like us, it’s not unusual to create a new piece of software in just a couple hours. With AI, we’re going from idea to working tool over and over again every day. Dashboards, workflows, little utilities, that type of thing. But what happens when you have to share it? 🧵

View full thread

7/ When anyone can create software, automations, whatever they like, with their agents, the challenge becomes access to data and connecting that software together. We think end users are increasingly going to demand that access and flexibility.

8/ Software should be as easy to share as it is to create. Your software wants to be free.

If you’re like us, it’s not unusual to create a new piece of software in just a couple hours. With AI, we’re going from idea to working tool over and over again every day. Dashboards, workflows, little utilities, that type of thing. But what happens when you have to share it? 🧵

View full thread

5/ AI is pushing us into a new paradigm where software is like water. Bowie said the same about music in 2002. Streaming proved him right. Software is next. We’re not going to put an operational stack around every little thing we make every day.

6/ And that’s where Tonk comes in. From the beginning, our bet was that AI would make the platform-centric web increasingly difficult to sustain. You may know this idea as the #SaaSpocalypse.

If you’re like us, it’s not unusual to create a new piece of software in just a couple hours. With AI, we’re going from idea to working tool over and over again every day. Dashboards, workflows, little utilities, that type of thing. But what happens when you have to share it? 🧵

View full thread

3/ If we want it to be collaborative, live updating, support multiple users, include granular permissions… we’re back in the world of deployments, databases, authentication, hosting, monitoring, maintenance, all that stuff. That sucks. We don’t want to do it, so we don’t do it.

4/ The old model assumed that applications were scarce and valuable, made by teams of admins and software engineers. In that world, you could surround each application with an entire operational stack.

If you’re like us, it’s not unusual to create a new piece of software in just a couple hours. With AI, we’re going from idea to working tool over and over again every day. Dashboards, workflows, little utilities, that type of thing. But what happens when you have to share it? 🧵

1/ Some of them we use for a day. Others we keep returning to, and they become a part of how we work. We no longer feel any friction building software. But we feel a lot of friction trying to share it.

2/ If we want to send somebody a report or a dashboard through Claude or as an HTML file, we can do that. But then it’s dead on arrival.

We’re producing software faster than ever. AI has changed the economics of software creation so dramatically that writing code is no longer the hard part.

View full thread

We think software should move in the same direction as documents. Handing someone a tool should include handing them the ability to understand it, change it, and keep it useful. The creator shouldn’t automatically become the admin of everything they build.

We’re producing software faster than ever. AI has changed the economics of software creation so dramatically that writing code is no longer the hard part.

View full thread

You don’t want to own all of that software. So you don’t share it. The cost of creating software keeps falling, while the cost of stewardship doesn’t.

Think about a Google Doc. When someone shares a document with you, they aren’t signing up to make every future edit. The knowledge is shared. The people using it can continue improving it together. Wouldn’t it be great if software worked the same?

We’re producing software faster than ever. AI has changed the economics of software creation so dramatically that writing code is no longer the hard part.

Today, when you build a piece of software, you’re also signing up to maintain it. The software may get handed off, but the responsibility usually doesn’t. You’re still the person people come back to for changes, fixes, updates, and questions.

That made sense when software was expensive to create. The builder and maintainer were naturally the same person. But if people can create hundreds of useful tools every year, that model becomes painful.

Tonk reposted

Your next phone will be something closer to a companion than a tool. It will know your rhythms and hold your memories. But before it can talk to your neighbor's companion, it needs a shared platform, matching accounts, and a server that is still running. That is bad and here's why ->

Ten years chasing the same coordination problem. Gavin on why local-first is the only credible answer he's found.

Gavin Owens@gavinowens.net

For ten years I've been chasing the same problem: how can groups of people coordinate without routing everything through a central authority. It's a common failure mode and a wicked problem. At @tonk.xyz, we're working on the most credible answer I've seen; one that is local-first all the way down.

Intelligence is a social process.

Kevin Kelly@kevinkelly.bsky.social

This is a section of my New Rules for the New Economy book published in 1998. Full book can be read at: https://kk.org/newrules/category/this_new_economy/ #kknewrules

The source of truth problem is older than AI. Agents just make it urgent. Gavin wrote up why knowledge fractures at the edges, why files and traditional databases both fail, and what Dialog-DB does differently.

👀

cdata@cdata.earth

Getting close to being able to demonstrate the complementarity of #DialogDB and #ATProto. Bi-directional sync between a general purpose, local-first (private by default) database and any PDS.