Radicle

Hello and welcome to today's episode of SolarCast

and with us today we have two developers

and the core maintainers from Radical.

Unfortunately in this episode today my main recording

microphone broke down so what you'll be hearing

is my backup microphone and luckily that was

still going but the audio quality is maybe

not up to par for the usual episode.

Just a forewarning but I think it's still

good enough.

I did my best to edit it so

it will be a pleasant enough conversation.

And the conversation that we'll be hearing will

be going through a lot of different aspects

and you'll be kind of with me on

the learning journey of what it means to

do peer to peer and local first Git

collaborations and forages.

So if you are as curious as I

am and was then tag along and let's

dive in.

Thank you so much for joining today for

this episode of SolarCast and with us here

today for those of you who cannot see

us is Lawrence and Fintan from the Radical

project.

Radical.xyz. So to get to know y

'all a little bit would it be okay

that you tell us a little bit about

how you entered Radical and what your backgrounds

are?

Yeah so mine was like a long winding

journey of getting to work on like functional

programming during college times, working remote for a

company that actually used Haskell.

So yeah I became enamored with Haskell in

particular during my university days and that eventually

led me to OSCoin.

And that happens to be the kind of

like roots of Radical so to speak.

So we kept in touch and I was

interested in the project.

It was kind of the idea of OSCoin

was to try and bring sustainability to open

source projects through a kind of like weighted

graph system, cryptocurrency and basically if you're dependent

upon a lot then your project would theoretically

get rewarded a lot.

So there's this classic xkcd comic with the

stack of blocks and we have the tiny

little block holding up everything in the corner.

So in theory that block would be getting

a lot of funding to make sure that

the whole stack doesn't collapse.

They realized that not only is funding a

problem but also the like collaboration on code

artifacts and repositories seem to be a problem

as well because it's rooted in centralized companies

and centralized countries let's say.

GitHub being based in America under Microsoft.

So early indications of what we're seeing now

in this like shift towards more sovereign solutions.

So like trying to break away from the

US.

Yeah and it was right about then that

I joined.

That project was kind of like the first

iteration and after that it was a Rust

project and it was the second iteration of

what we call Radical.

And then the name was Radical Link and

instead of a kind of Lisp machine it

was kind of back to basics.

We've got Git, why don't we build on

top of it.

So that's around when I joined about six

and a half years ago.

I've been working on things since then.

I can get into the why a bit

more but maybe I'll hand it over to

Lorenz to kind of give his perspective of

why he got into this.

Thank you Fintan.

Hello to all our listeners.

I am Lorenz and I joined the project

quite a bit later.

So I was not around in this OS

coin phase and the let's say the earlier

explorations and iterations of the project.

I joined quite a bit later just as

a open source contributor.

So I first started contributing just out of

interest because I liked the idea.

I was actually as GitHub got acquired by

Microsoft a few years back I started wondering

okay like where is this going and I

figured why aren't there more projects that work

on the decentralization aspect and making the decentralized

components of Git which are really strong more

usable and more user-friendly.

Because the reason is that the underlying situation

is that Git actually is very decentralized but

the dynamics in the real world have unfolded

in a way that we see lots of

concentration so we see lots of people gravitating

towards GitHub big part of the open source

ecosystem and a few other forges but GitHub

is the biggest by far.

So at that time I didn't even really

know about Radical.

Blockchains were more prominent but I had a

few questions and reservations there so I didn't

really engage much in these projects but then

at some point I learned that Radical actually

switched the implementation to Rust and switched to

one that is not dependent on the blockchain

but is more open to like opt-in

in the future or build on top but

certainly not in the core of it as

it was previously.

In fact the situation that Fintan mentioned before

about the graph to distribute the resources and

funds among open source communities this is a

project that spun out into so this still

exists it's called DRIPS and we say it's

our sister project but that kind of was

realized that it might be better to split

these two aspects and I do agree.

So after that period I decided okay now

is the time that I start contributing to

one of these projects and then I started

contributing to Radical.

As usual you know you find those things

that you think could be improved you start

with smaller ones then you somehow understand the

code base you find a few bigger ones

and that's how I grew into a more

senior contributor to the project and then at

some point I was asked whether I would

like to maintain the project together with Fintan

and Adrian and yeah that's what I'm doing

now.

You mentioned something there as usual and then

you describe this process of like finding these

different projects that you want to contribute to.

When you say as usual who are you

talking about here like as usual?

Yeah it's probably colored by my own view.

I would say as usual for really an

open source enthusiast you know the kind of

people that use open source software and if

they see an issue or something that could

be improved they really go clone the code

figure it out do the fix and so

on and yeah I have done that in

like a handful projects those that I use

personally and I like to do that when

possible but of course it's not always possible

to start with when the software is not

free.

So I've been having a lot of conversations

with some friends recently who are new in

the open source space and they want to

start exploring and I think one of the

challenges is kind of what you touched upon

here is like finding what project to contribute

to and so we touched on it or

so why did you find that Radical was

the project worthwhile to contribute?

I think that Radical is pretty far out

there in terms of ambition and also in

terms of challenge in bringing Git back closer

a bit to its roots or at least

how we see it bringing this decentralization aspect

back.

So there are quite a few open source

projects out there in terms of different forges

for example many many will know Codeburg and

that they use ForgeJoe others might know SourceHUD

but as far as I know Codeburg is

an association so they are structured as an

association registered in Germany they invite new members

to join and the core offering is that

they host infrastructure for you to collaborate on

open source code open source software free software.

And they're kind of like GitHub but open

source?

I mean I wouldn't go that far I

wouldn't say they are like GitHub because GitHub

is yeah is a bit of a different

topic I guess but Fintan yeah please take

it away.

Maybe it's good to just clarify what like

a GitForge means in general so most people

my favorite comment is like Git it's already

decentralized when we talk about Radical and we

fully recognize that of course it is it's

a decentralized tool and how developers started using

it is through a decentralized workflow called email

so the Git core team and the Linux

team use email lists for collaborating through Git

and so email can be one Forge and

that the like step after that was SourceHUD

which is an email based Forge where you

get a website where things can be hosted

and you can see the emails coming in

and discussions.

GitHub is another GitForge which is also a

platform and that's what people are most familiar

with as a Forge.

Other Forges now exist such as Codeburg and

also Tangled on the Approto protocol if people

are familiar with Approto and Newsguy.

GitLab another Forge yeah so okay so I'm

coming from the perspective of like peer-to

-peer space and there's tons of different people

coming from a lot of different spaces can

we go back like a little bit to

the basics I think anyone working with code

has encountered like Git has encountered repositories and

have heard about Forges but what actually is

a Forge?

Yeah yeah exactly so it's I guess it's

a home for collaboration so Git gives you

the commits and files so blobs, treats, commits

that allows you to version control your code

over time but as humans we have lots

of other ways that we collaborate and that's

where we see things like issues or task

trackers or discussions on changes identifying as someone

who collaborates on these Git repositories that's I

think that's where Forges bring or like yeah

yeah they paint the picture a little bit

more on top of Git.

And in comparison to like something like CRDTs

like we have this space that's become really

really popular lately with Local First and all

of that and CRDTs are a common means

of facilitating collaboration like what's the differentiation there

between Forges and CRDTs?

I'll let Lorenz take it.

So I think that CRDTs are a technology

and a definition of how a data structure

should work and what features a data structure

should have and you can build something like

a Forge which is a little more abstract

concept like a feature set of things that

people want to do so you approach you

can approach it from two sides you can

say okay what does the user want to

do okay they want to do code review

they want to discuss the changes and then

the CRDT is a bit on is a

bit of technology that can enable this and

then yes what Radical actually does is it

implements a Forge based on top of CRDTs

so many of the things in Radical that

we talk about like patches and issues are

implemented as CRDTs so the underlying structure still

is a CRDT but it is not critical

for the user to know that when they

open a patch there will be an operation

on the CRDT that's abstracted away.

Yeah to analogize it would be like asking

what's the difference between a page rank and

a web search or like a search engine

so page rank is just an algorithm and

a search engine can like bring other features

to the table.

Yeah so we were on the topic of

like why and so the why enters into

the space now and then we kind of

like moved into what actually is Radical.

I have an idea of bringing us back

a bit here to where we were before

talking about the different Forges and other solutions

in this space and we talked about a

bunch of those and you can you can

place them on a line and you can

think about okay how centralized are they and

I think we have like peak centralization with

GitHub and then you can place the others

right and then we have this concept of

federation where as there are a few different

ways of how to achieve federation but usually

you have multiple servers and then the user

logs into one of these servers and then

there's data being federated between the servers so

all of these solutions which includes Tangled on

AT Proto and which includes the work towards

federation in Forge Joe I would put them

in the middle.

Email also yes you're very right to point

that out email is also of that kind

you communicate your email server and then the

emails are replicated and then we have another

category where Radical falls into which is basically

where there are no servers there are just

nodes in the network and they sync in

a peer-to-peer manner yeah and there

aren't really any other competitors in a sense

or any other projects that are currently ambitious

and as active as we are to develop

a really peer-to-peer Git layer.

And I have so many questions in this

regard I'm so curious and I've been hearing

ever since I heard about Radical which was

about two years ago now I've been wanting

to figure out a little bit more to

start from the perspective of why is it

important to do it peer-to-peer in

this setting like what makes the big difference

here and then how do you enable those

peer-to-peer connections?

So I think it's important in general to

have a technology that works in this local

first and peer-to-peer manner because it

puts all the power into the user or

at least very very close to the user

they control their own hardware there is no

central authority that tells them when to log

in or whether their account is blocked or

not and it is a technology that enables

this completely decentralized setting where you know I

join Funkfeuer which is an association of people

that put Wi-Fi antennas and on their

roofs and then they might make direct peer

-to-peer connections if you need internet access

to let's say a data center in order

to use your forge then well you're tied

to this access but with Radical it will

just work in such a completely partitioned part

of the of the internet just as well

for example if Fintan and me work in

a room and are logged into the same

Wi-Fi then we can just directly exchange

information over Radical and we don't need to

go out to a data center and so

on so we are immediately much more resilient

and much more sovereign than with these other

technologies.

It enables this routing agnostic perspective where you

can route over different transport layers it sounds

like could you use mesh, Wi-Fi, Bluetooth,

sneaker net whatever kind of here?

Yeah I have to admit that we are

currently using TCP transport so we are a

bit limited in that regard but we could

extend so that's a limitation of the current

implementation but not out of the protocol design

so the data structures that arise from the

sync process they work even if you you

know transfer via sneaker net or whatever but

the current implementation expects to talk IP and

TCP.

I had a previous conversation with IRO and

we talked a lot about map traversal or

hole punching to establish these peer-to-peer

connections.

How do you go about that?

We know about IRO, we are actually in

loose contact with them and we are really

watching the awesome work they are doing and

we are thinking of at least allowing connections

via IRO to in a like a testing

phase.

It lines up really well on like the

the cryptography of it so in radical nodes

are identified by an edwards 25519 key and

the same is true for IRO so we

might get the direct correspondence between IRO endpoints

and and a radical node ids so it

all looks it all looks really nice and

they would give us the hole punching capability

which we don't have currently.

But it's an ongoing process and we're definitely

eager to experiment more.

And this is actually we're getting into an

area that has been a big part of

my research in the past which is this

what's starting to emerge is kind of an

ecology of protocols from previous years like 10

years ago a lot of projects were trying

to build the entire solution on their own

which meant rebuilding the same wheels over and

over but now these kind of protocols are

entering more of a modular form of design

so it sounds like you're solving a quite

a specific part of the puzzle piece of

peer-to-peer networking and specifically for git.

What is like where do you see yourself

and what is your solution more concretely?

Yeah that's a great question and I do

hearken back to when I started a phrase

that was thrown around was like protocols not

platforms.

So yeah I guess where we're trying to

fit in is building the protocol on top

of git for collaboration and as we spoke

before we're talking about we're talking about it

in a self-sovereign way we also fit

into this local first even offline first scenario

where you can hop on a plane still

work on your collaboration not just your code

so I can like reply to issues reply

to patch comments and then come back online

later and sync up with the network so

yes putting that protocol first and hopefully making

it repeatable so that people can come and

boost the ecosystem with other implementations in the

future I think so

I'm also thinking like you mentioned in the

beginning like a question that a lot of

people have asked you is like okay so

why are you doing this git is already

decentralized like what do you say when people

ask that like where do you because okay

I think it boils down to decentralization be

a very fuzzy term because there's so many

forms of decentralization and federation and peer-to

-peer are often dumped in as it clumps

some under decentralization is it there in the

distinction that by enabling what or by doing

what you're doing with git you're kind of

bringing it to a new level of decentralization

my thoughts here are that we are we

have realized and I think many people will

agree that software engineering and managing a software

project being open source or maybe even not

is not just about git commits and and

tags and so on there's a lot that

works on top like code review and discussions

about the code and tasks and so on

and these are just fundamentally not part of

git so here we compete in a sense

with these other forges so git is decentralized

in the aspect that you can you can

put the repository on any computer and if

you set those computers up well enough then

you can have this network of git repositories

and that is decentralized but you're decentralized only

at the level of git features which aren't

really enough to do all the things you

want to do as alluded to before so

this is one aspect you you need more

than just git and you also want it

to be decentralized that at least we do

and then one other thing that comes to

mind is that for using git in a

decentralized way you have to make careful configuration

like you have to share these urls of

all the murals of your repository with everyone

so if five people want to want to

collaborate on a git repository in a truly

decentralized way then each of these five persons

has to note down the git url of

all the four others and we make an

improvement here because we have a globally consistent

way to address git repositories and that's something

that git itself doesn't have there is no

standard way to say this repository which is

agnostic over various locations of the repository that's

something that we add and note that this

is not centralized so there's not a global

registry where you register the repository but this

is done in a decentralized way are you

using dhts like distributed hash tables no we're

not using dhts currently we do node discovery

via a gossip network so at first you

join the network usually via what we call

a process of bootstrapping so you first have

to know some node in the network and

then from then on the nodes will give

out addresses of other nodes that they have

found on the network and then very quickly

you build an address book of other nodes

awesome so i'm wondering now like you talked

a bit about like the importance of enabling

the collaboration aspects in connection to git and

we also mentioned platforms and stuff like that

where what is like dream scenario where do

y'all want to go like what is

the roadmap what is a future that you

envision when radical is like a project that's

complete fully because i guess it's also still

in development yeah um what does that look

like for you that's a it's an interesting

question because there's a question of when is

it a software project ever done and i

actually it reminds me of a post by

moxie um one of the founders of signal

and in it they write about kind of

like envy or jealousy of musicians where it's

like you release a piece of music and

then it's kind of just done and you're

you're done i mean you can always like

remaster edit remix whatever push software just like

it feels infinite it can go on forever

but yeah but then yeah is it mostly

like scope creep or is it like is

it because there's maintenance what is the main

thing that makes it infinite do you think

maintenance does make it feel infinite the most

but of course i think what makes it

feel infinite is just like the power of

ideas and um yeah maybe scope creep is

kind of like a way of describing that

but i guess it's all about like constant

improvements like git was