Revolver Kickoff

Revolver is officially a Real Thing™. This is the kickoff announcement, you can also read an explanation of the project motivations. It has been difficult to pry myself away from hacking on it to write about it.

I've been making somewhat cryptic noises about this project for about a year, sketching out the basic architecture, making notes, losing the notes, saying I was going to get to it but it was a slow process, etc. Some recent events upped the urgency, so I said to graf that I was going to start hacking on it but that I expected there wouldn't be much attention on the project. I made a hasty half-announcement on FSE and a LiberaPay team for Revolver with a blurb about the project. This, it turns out, was wrong: people have been excited, some donations have already started rolling in, which means I can afford to just work on Revolver for a little while. I'm shocked and grateful and elated, but it does mean I should write a somewhat more official announcement. (I can't stress enough that the level of interest caught me somewhat off-guard.) Here it is.

What is Revolver?

I'm going to start with the four-color glossies that you have seen if you saw the initial announcement (though I fixed some typos). At present, aside from code on my computer (and soon, hopefully, yours too), it is a set of goals.

• No admins, no masters, no backups: Decentralized systems don't need them when architected properly.
• Resilient: No individual domain name, individual machine, or individual admin is responsible for the network.
• The internet routes around censorship: No reliance on DNS, LetsEncrypt, or any central authority. Revolver federates through multiple channels, including IPFS, and the broader fediverse by serialization to LitePub.
• Lightweight: Small enough to run a node on your phone.
• Low-Tech: Nothing complicated, just well-understood tech we should have assembled in the 90s.
• Spam-resistant: Manual and automatic filtering to keep public nodes clean.
• E2EE DMs: No spying on private messages by admins, the government, anyone.
• Keep the fediverse: Revolver will remain compatible with the rest of the fediverse: Pleroma, Honk, PixelFed, Friendica, Hubzilla, even Mastodon. The project came from the fediverse and is indebted to the fediverse (especially Pleroma), so the software will speak LitePub and be part of the fediverse instead of competing with it.
• Keep your frontend: Revolver will be compatible with, at least, the standard PleromaFE and bloat.
• Plan 9-friendly: Smug overload. Revolver is developed on Plan 9 and Linux and will remain compatible with both.

Project Status, Architectural Notes

The prototype is being actively hacked on, and shortly the prototype's source code will be available and if someone else manages to build and execute it, there will be a handful of nodes that are run by people that aren't me, at which point I'll be hacking on the production version (and accepting patches and probably adding people to the team) until the initial set of features is complete, at which point it will be "beta". At the current rate, within a month or so, it will be talking to the rest of fedi.

At present, blocks slush around and you can post something and it will show up on another server. Right now the interface is 100% text, though. The protocol is rudimentary. Signed posts can be distributed via HTTP or IPFS; there's a primary format and storage/propagation are pluggable. (I didn't hack in attachments yet but that's easy. Nodes that support IPFS should be able to distribute attachments as IPFS hashes, too.) Obviously Revolver will have hooks for integrating Tor: shallow integration is possible to achieve just by being careful about proxy support, and that might be enough. I have some thoughts on BitTorrent but those are still percolating; I think it works as a side-channel but it was an issue with Twister, as noted below.

There is a lot of work to do. I am very excited about that, and am extremely excited by the people that have offered to hack on it.

Interface Notes

I've used other fully decentralized systems. I liked Twister and used to mine blocks for the network. Twister was one of the early ones: it started as a fork of the Bitcoin source code, and used WebTorrent to distribute posts. Pretty cool! But that landed it on the bad side of some trade-offs: namesquatting was rampant, discovery was difficult, running a client was heavyweight, the network was fairly slow, search was broken.

I think Revolver can solve a lot of that. The design has benefitted greatly from fedi, from seeing how the network has worked these years, the performance pain-points on FSE and elsewhere.

Recently Asked Questions

• Licensing: Right now, the license is "nobody but me has the code". Because it keeps the options flexible (I can always dual-license later), it is going to be AGPLv3.
• Language: It's still in the prototype stage, but because I am developing it using ANTS and I want to keep it compatible with Plan 9, it is Go. If it becomes useful or necessary to change the language, it'll probably be C.
• Compatibility: It will be compatible with the rest of fedi. It would be pointless otherwise, I love fedi. It will also be compatible with the Pleroma FE. I liked what lain said in Sean Tilley's interview (unfortunately hosted on Meejum): "That's something I’d recommend to other fediverse projects, too: Write compatible APIs." Let's all love Lain!
• A plan for spam: I don't want to over-plan for this. I have some ideas, but people are endlessly inventive. No plan survives first contact with the enemy, and as Eisenhower said, "Plans are useless but planning is indispensible." This is completely true of abuse-mitigation. So I have some features designed for filtering spam and other unsavory things in a way that should allow you to have control over your network and the things you help propagate.
• Contributing: It will be open-source, obviously. Aside from donations, when it's released, running a node and filing bugs will be helpful. As far as code goes, patches will be gratefully accepted.
• Project funding goals: I don't have any. I planned to hack on it but I didn't expect anyone would be donating yet. To be brief, I have just been writing code and I took myself away from that to write some blog posts, but the project is still a prototype. I didn't think anyone would pay attention so I didn't set anything up, I just planned on hacking. So my response to this is the same as my previous response: if there's enough funding that I can just do Revolver, I'll just do Revolver. If there's no funding, I'll still do Revolver but I do have bills so I will do Revolver plus profit-seeking activities.

About the Name

It might change. If you made the connection, then you're pretty good. There's another parallel, though. Admins on fedi disturb the balance to some extent, and Revolver should act as a great equalizer: people running a server should be free to post like regular people, and people using someone else's server shouldn't be beholden to the admins. If it can do just that part, the project's a success, but I think the other goals make it more than worth the effort.

Funding

You can donate to Revolver via LiberaPay:

You can also send Bitcoin (BTC), the address is 1BZz3ndJUoWhEvm1BfW3FzceAjFqKTwqWV and here's a QR code:

There are plans for Monero/Doge/BCH/ETH/BAT. Like I said, I haven't taken time to set up anything really. The attention being paid to the project is a surprise.