tangled newsletter 01 — hello
53 points by j3s
53 points by j3s
I'm a bit wary of VC driven software projects. How long til Anthropic or OpenAI buys Tangled?
That said, if they follow through with federation support, it seems a good way to give enterprises an option beyond selfhosting Forgejo.
I'm currently working on my exodus from Github. I have about 350 repos I'm moving, to both a homelab Forgejo and Coderberg. Migration to homelab Forgejo has been wonderful and problem free, Codeberg is up next.
I hope the future of open source uses the decentralised nature of Git to actually be decentralised, rather than centralized on a platform we once trusted. But open source is also under a lot of stress right now, and solving problems like supply chains and funding seem like they need real solving first.
I'm a bit wary of VC driven software projects. How long til Anthropic or OpenAI buys Tangled?
Tangled founder here. I have an essay on our decisions around financing still sitting that I need to polish and publish. In short: public funding (in Europe) is notoriously hard to get, and incredibly time consuming to apply to (not to mention the multitude of default closed doors given only one of us founders are an EU resident). VC was the quickest way for us to go full time on Tangled, build a team and scale infrastructure.
Our ultimate goal for Tangled is to be "permanent software"—and this is something we've discussed with our investors at length and they're full onboard with! Regardless of our exit path, we want to ensure that users will be able to spin up a "Tangled 2" (Twangled?) and their data (thanks to the magic of AT Protocol) will seamlessly show up there.
our exit path
No offense, but words like these are exactly why people don't like to rely VC funded stuff. When VC enters the picture, all possibility of a long running sustainable business goes out of the window and customer squeeze becomes a matter of time.
Does anything exist permanently without an exit path? People burn out, people's personal lives become complicated, people even die. The best software is designed to survive those things.
Also, being explicit about it is important to me.
Not really, but in my experience, companies that aren't backed by vulture capital tend to better serve customers.
Also, what is this "best software"? Software, as a field, is maybe 85 years old and I'm unaware of any piece of software that is as old. Linux is only 35 years old, which may be the oldest version of a Unix-like still in use. Microsoft Windows is 41 years old. Oracle (the database) is 48 years old. I'm not aware of much more that is still in wide use that is that old.
For comparison, the Japanese company Kongō Gumi was founded in 578 CE---that's almost 1,450 years old at this point.
[1] Konrad Zuse made a Turing complete computer in 1941.
GNU Emacs was '85. Postgres is from '86. GCC came a little after.
That still proves the parent's point :)
Regardless, eyeing towards exit path when one is starting the business means one is looking at cashing out asap rather than doing something they enjoy while improving the world a bit by having their output help.other in return of modest income.
But then again, thats just, like, my opinion.
public funding (in Europe) is notoriously hard to get
To add to that for those outside of the EU: it's not just hard to get because there's little of it, but also because a lot of funds (or at least the ones coming out of the European Commission) are hype driven, meaning their focus changes every 5 years or so. If your project doesn't fit within that box, you're not getting funded.
Maybe I’m dense, but I’ve never quite understood the benefit of tying together source control with ATProto. Is it a stand-in for an identity provider?
So you have Git: decentralized, possible to mirror across multiple different Git servers. SourceForge, Github, Google Code, Codeberg, Forgejo: they all provide metadata on top of the raw Git servers, in the form of issues, discussions, mailing lists, pull requests, etc. Federalization (of whatever protocol) in theory allows projects to choose whatever code host there is, but have their project seem "native" to the other "forge" providers. Allowing for decentralized metadata, on top of the decentralized nature of Git.
all code collaboration at present uses two protocols:
the "comms" layer refers to some form of identity and message transfer. you need such a layer to build issue trackers and discussion systems. tangled uses it for other things too: repository stars, web-of-trust, an algorithmic feed, discovery etc.
I run three separate 'knots' (the Tangled parlance for git servers): one for work, one for my personal use and a scratch one. I can log into all of these and view them using my ATProto, including metadata (issues/releases) for each of them. It's nice, and if the ATProto network goes down my own git remotes are still available on my own infra.
I'm just playing around with mirroring a local Forejo to ATProto records, so that I can work completely offline on my local instance but still benefit from the online network as well. Tangled is a nice setup.
'knots' (the Tangled parlance for git servers):
I'm glad to know the concept of making vocabulary as opaque as possible didn't learn anything from Homebrew's bottles, casks, taps, and whatever other "cutsey" beer terminologies they could use to hide the idea of binary-tar, external-installable, repository-source, etc.
While reading through the newsletter, it seems they went with "spindles" instead of goddamn "ci runner," which isn't even a vine related word
Naming things is fun, and we explicitly say they’re CI runners…?
Naming things is fun
For the ones doing the naming, maybe. For everyone else, who will have to learn yet another bunch of short-lived, unique snowflake names for the exact same set of long-since-named concepts — not so much.
This is still a "platform".
If they decide to kick you out, like GitHub has been doing extensively to random people for opaque reasons, all your links are still broken, all your contributors still have to make an account somewhere else, all your users are still confused.
And it's on AT, whose authors simply don't grok capability theory. I have gotten used to the dance of getting a link to BlueSky, seeing that the post author tried to make it private, and then going over to SkyView or BlackSky to read it. For all of its faults, I am confident that at least some of ActivityPub's editors understand capability theory.
I have gotten used to the dance of getting a link to BlueSky, seeing that the post author tried to make it private
There’s no notion of private posts on Bluesky. Are you making things up?
Bluesky has an option to „Discourage apps from showing my account to logged-out users“, but it comes with a big warning that this only affects Bluesky itself:
Logged-out visibility
Discourage apps from showing my account to logged-out users
Bluesky will not show your profile and posts to logged-out users. Other apps may not honor this request. This does not make your account private.Note: Bluesky is an open and public network. This setting only limits the visibility of your content on the Bluesky app and website, and other apps may not respect this setting. Your content may still be shown to logged-out users by other apps and websites.