I think this is neat because it’s Forth. I admire Forth from a distance. I’ve never really done anything with it, but I like it in principle. It’s stack based. It has a very minimal syntax. It can run on very small microcontrollers.
I agree- it’s a bit of an obsession for me outside of work. I’ve never built commercial software with it, but I admire the way it can do so much with so little.
There are other applications, such as ssb-chess, git-ssb (Git over SSB) and an NPM alternative. More are listed here, but the biggest apps in the ecosystem are the ones in the tutorial. That’s not to say it’s the only thing SSB does, though.
Just finished reading the spec. Very interesting. Very well written.
edit: it looks like the only medium for discussion mentioned in the docs is the vaporsoft email address. Is this correct? There is no irc channel or mailing list?
All joking aside, this is a continuation of a discussion on Tilde.town IRC. I did a poor job of supporting my argument (SEE: title of this post) and I felt that the author of the article did a better job of defending my argument than I did myself. I hope that the folks I was discussing this with take a look at this article and see why I came to the conclusion I did.
I’d love to have a tape drive for my personal backups. It would be much easier to send an encrypted tape to a friend’s house than an HDD, or to purchase some cloud space to keep my data safe. I should just bite the bullet one of these days and get the drive, but they are expensive…
This sounds super interesting. SSB is kinda cool, but does require a network connection, so something airgapped would be nice.
You may want to look into some of the work done by FreeNet, as they are able to transport FSK objects via sneakernet (Used to smuggle info in an out of North Korea).
That is odd, back when I spent more time on SSB, one of the things touted was that a network wasn’t needed. Updates could be done via a USB device by friends meeting at a pub, for instance. Has this changed?
Despite having the capability to support Sneakernet, there are very few well supported plugins or packages for SSB that reliably support this. There have been discussions about it but if you actually want to replicate your SSB feed today without ever touching a UDP/TCP stack, your options are limited.
Having a fully sneakernet-based protocol would be great for constrained / lightweight systems. Unfortunately, SSB as it is today is best suited for desktops that can run Electron type apps (or install Node/NPM). You will be able to find partial implementations of the SSB protocol in other languages, but if you read through the source, you will quickly find that they either a) only implement part of the protocol b) are just a thin wrapper to a NodeJS-based SSB server on the local machine. I think these folks are working on an early stage Rust alternate implementation. Also, projects like Manyverse are improving this situation for Android/iOS, but are still pretty early. I think the main reason for this stagnation is that SSB has a lot of moving parts and a JS-based reference implementation rather than a fully documented protocol (at least since I last checked). I also think this could be avoided by focusing only on the sneakernet-based use cases (and not worrying about the network related ones) while also designing the protocol from a spec rather than a reference implementation.
TL;DR: The SSB of today supports sneakernet in theory, but not in practice. Furthermore, it does not supply a formal specification for implementations on constrained / non-Javascript systems.
=== DISCLAIMER ===
I hope that readers of this comment do not interpret this as a criticism of the SSB protocol, but rather that I had some very specific requirements that SSB wasn’t really designed for. With that being said, it’s one of the best decentralized protocols in recent years and I use it daily. The SSBC is doing a great job of building a specific technology for a specific need
BBSes did this all the time, via UUCP and QWK packets. This sounds like a great way to implement it, seeing as UUCP is designed just for this type of not-always-on connection.
I think this is neat because it’s Forth. I admire Forth from a distance. I’ve never really done anything with it, but I like it in principle. It’s stack based. It has a very minimal syntax. It can run on very small microcontrollers.
I agree- it’s a bit of an obsession for me outside of work. I’ve never built commercial software with it, but I admire the way it can do so much with so little.
Now I get what you mean with the SSB. So it’s just a personal Facebook or does it have more applications than that?
There are other applications, such as ssb-chess, git-ssb (Git over SSB) and an NPM alternative. More are listed here, but the biggest apps in the ecosystem are the ones in the tutorial. That’s not to say it’s the only thing SSB does, though.
Good to know! I just miss an app purely for private messaging. I guess it’d be useful in cases of censorship or vigilance.
Just finished reading the spec. Very interesting. Very well written.
edit: it looks like the only medium for discussion mentioned in the docs is the vaporsoft email address. Is this correct? There is no irc channel or mailing list?
Good idea. I’ve created a private mailing list. pigeon_developers AT vaporsoft.xyz
I typically don’t do mailing lists, but am willing to learn and I’d be open to advice.
This is like that time Andy Warhol painted the soup can. It was so obvious but so inspiring. Have an upvote.
Up next, facebook.com!
Isn’t the title of this article “The reckless, infinite scope of web browsers”?
That’s my opinion and I’m sticking to it.
All joking aside, this is a continuation of a discussion on Tilde.town IRC. I did a poor job of supporting my argument (SEE: title of this post) and I felt that the author of the article did a better job of defending my argument than I did myself. I hope that the folks I was discussing this with take a look at this article and see why I came to the conclusion I did.
One of my favorites is Krowbar’s IRC Stats
Nice find Kirch.
I’d love to have a tape drive for my personal backups. It would be much easier to send an encrypted tape to a friend’s house than an HDD, or to purchase some cloud space to keep my data safe. I should just bite the bullet one of these days and get the drive, but they are expensive…
Agreed. If anyone in the tildeverse knows hacks for cheap tape drives, please do share.
This sounds super interesting. SSB is kinda cool, but does require a network connection, so something airgapped would be nice.
You may want to look into some of the work done by FreeNet, as they are able to transport FSK objects via sneakernet (Used to smuggle info in an out of North Korea).
Nice find! Do you have any links to that work on the Clearnet?
Well, crap. I did at one point, but they seem dead now :(
That is odd, back when I spent more time on SSB, one of the things touted was that a network wasn’t needed. Updates could be done via a USB device by friends meeting at a pub, for instance. Has this changed?
Maybe I’m incorrect? I thought it at least needed a wifi network in order to discover neghbors.
I haven’t dug in enough to actually find process, but these two articles both reference using sneakernet as a medium for SSB updates.
https://staltz.com/an-off-grid-social-network.html
https://gopher.floodgap.com/gopher/gw?gopher://sdf.org:70/0/users/rusty/Post04-ssb
Despite having the capability to support Sneakernet, there are very few well supported plugins or packages for SSB that reliably support this. There have been discussions about it but if you actually want to replicate your SSB feed today without ever touching a UDP/TCP stack, your options are limited.
Having a fully sneakernet-based protocol would be great for constrained / lightweight systems. Unfortunately, SSB as it is today is best suited for desktops that can run Electron type apps (or install Node/NPM). You will be able to find partial implementations of the SSB protocol in other languages, but if you read through the source, you will quickly find that they either a) only implement part of the protocol b) are just a thin wrapper to a NodeJS-based SSB server on the local machine. I think these folks are working on an early stage Rust alternate implementation. Also, projects like Manyverse are improving this situation for Android/iOS, but are still pretty early. I think the main reason for this stagnation is that SSB has a lot of moving parts and a JS-based reference implementation rather than a fully documented protocol (at least since I last checked). I also think this could be avoided by focusing only on the sneakernet-based use cases (and not worrying about the network related ones) while also designing the protocol from a spec rather than a reference implementation.
TL;DR: The SSB of today supports sneakernet in theory, but not in practice. Furthermore, it does not supply a formal specification for implementations on constrained / non-Javascript systems.
=== DISCLAIMER ===
I hope that readers of this comment do not interpret this as a criticism of the SSB protocol, but rather that I had some very specific requirements that SSB wasn’t really designed for. With that being said, it’s one of the best decentralized protocols in recent years and I use it daily. The SSBC is doing a great job of building a specific technology for a specific need
=== END DISCLAIMER ===
Been using SSB for years now and never seen the USB approach, but I’m curious now.
This writeup had a reply that hit me:
BBSes did this all the time, via UUCP and QWK packets. This sounds like a great way to implement it, seeing as UUCP is designed just for this type of not-always-on connection.
Nothing says you cannot UUCP from a USB drive.
Excellent write up, and great original content :)
Thanks!
woah super cool!
Thanks Ben!
I’ve got an old PIII w/ Ubuntu on it somewhere… I was gonna try to put a cluster of RasPi’s in there, but maybe a vintage OS is more fun?
Might I suggest Risc OS?