Oddbean new post about login | logout we are working (not high priority) on using Pkarr in Rustls to get tls support for Pubky domains, with no certificates or proxies. Of course won't work for browsers, but perfect for native apps. Definitely would make running a Homeserver less of a hassle, if you don't need to support browsers, since you won't have to buy a domain.
It should be possible to fork or even just update browsers like Brave (i.e. remove shitcoins, make it easy to install Alby or similar, add Pubky support and it's ready to go)
That is quite a lot of work, but yes, it is the way to go.
yes but that is worthless, since no one will download it, and those who would, would have downloaded a native app instead. I would rather lobby existing browser to support Pkarr directly. Even if unlikely to pass, at least the reward is worth the effort
Then start with extension to get a feel if it's worth it?
maybe, but for now, we just expect homeservers to have domains, and that works fine in browser. we need to focus on the maximum leverage first
Well, I learned about Pubky 5 minutes ago, so I guess I'll wait a bit before explaining how you should do it 😅
no you are thinking right, we already considered browser extensions... the juice just wasn't worth the squeeze.
If you build it into an open protocol and open eco system, in the same way Ubuntu did, with transparency and a good devX, developers will build things like this. Easier said than done, but worth it in the long run.
I am allergic to governance and design by committee, tell my something actionable that doesn't boil down to "every one suggest complexity and we maintain it" and I am here for it. Matrix has governance and spec in abundance, but it is complex as fuck. We try to keep our stuff simple, and once they are stable we write them down. If they are not written down, then they are still and experiment and R&D happy to learn to be more inclusive.
It depends if you want to be able to on-board and retain developers (without paying them large sums). This is hard, and doesnt happen by itself. Most FOSS projects dont make it. But put in the work and you will get results. Quite old now, but still probably one of the best works on the subject. Chapter 10 is on Governance. Agree matrix is poor, there are worse. https://www.jonobacon.com/wp-content/uploads/2019/01/jonobacon-art_of_community_second_edition.pdf
The "super simple protocol first to tantalize developers" worked really well for nostr (in comparison with the web3 specs or whatever). It's worth waiting couple years before making the protocol confusing and hard to learn 😉 In general the main metric that you are working with is the ( mental capacity * time invested ) of individual developers that will use the protocol. If you have too many concepts and things, only couple devs will learn it and use it.
In its early days nostr was a simple 2 page spec that worked. Develoeprs were attracted to it because they could go away in a weekend and build a relay or client. It has got way more complex over time. And what could come next (negentropy) is a complexity bomb that may shorten its runway. Pubky talk about a "complexity budget" and they are careful to manage that. It's a good approach, and learning lots of lessons. I really hope there is a project where devs can build on simple concepts in a permissionless environment, and let 1000 flowers bloom. It is much needed!
our repo has an authorization example with a hidden web component in it. once we have a good app for users to hold their keys in, any developer can drop this web component and magically get scoped authorization from users, then basically every app on Solid and RemoteStorage can be forked for pubky
I would rather focus on UX first and DX second. From our hackeeek experiment, onboarding devs is really not hard, if there are users they want to target they will be able to cope. I also think our protocol is way simpler and developer friendly than Nostr. I have seen someone who built bots for nostr building bots for Pubky, and they appreciated the simplicity. all the complexity is really at Pkarr level and this is where we can double down on reliability so developers don't get frustrating errors. but other than that... it is just HTTP PUT, GET and DELETE. How simpler can it get?
Depends how long you want to last. On boarding developers is much easier when something is new. But then it gets much harder. The economics of nostr are that of generosity. Jack & co. very kindly donate about 50k-100k sats per daily user to keep development going. The relay operators donate their time and resources to keep the network going. That goodwill tends to run out after a while (normally 6-18 months). UX is key of course, but nostr has lots of really good UX. Most of the devs that have left are not coming back. And you end up with a finite runway.
My goal is to make something like DNS and HTTP, that works and reliable regardless of daily active users or other VC metrics. I and Synonym team might need to get another job, but if we solved small problem completely, the solution will last