I'm still really disappointed he didn't go for a more long term solution, federation:
https://matrix.org/blog/2020/01/02/on-privacy-versus-freedom
I think Moxie is right, federation will lead back to recentralized services (for the reasons Moxie outlines in his recent Web3 post). People don't want to run their own servers. In that world you'll have federated services that are mostly centralized with a much worse user experience than centralized competition. Signal's real competition isn't Matrix, it's FB messenger, Whatsapp, and iMessage.
The incentive failures are actually upstream from the application layer of the web. Running a linux server is too hard, spam and auth are issues with the current tech stack, dependencies are a mess of complexity and federated systems built on the current stack can't really solve these issues so end up recentralizing to sysadmins that do (at best). This is the reason I started working on Urbit, I think to fix this for real you have to fix problems farther up the stack.
Given the current landscape, Signal is the best available option imo for most people. Hopefully if Urbit succeeds we can have the federated system we want with UX that's actually competitive.
Thing is, Signal has the exact same issue: the top post of this thread is about how people are feeling compelled to donate to Signal so that it does not rely on one kind benefactor. If people want to donate to Signal, why not donate/hedge a bit by donating to Matrix or to the Conversations (the best XMPP client) developers?
> Signal is the best available option imo for most people.
Signal is still centralized. It is "open source" only in name, as the client code was constantly out-of-date and it is basically impossible to fork it or run your own server. It has a very poor record cross-client vulnerabilities and it forces everyone to be dependent on the security of their smartphones. How many times do we have to re-learn not to put all of our eggs in the same basket?
I don't think Matrix will be able to compete. I think Signal is the best available option today for most people, but the risks you outline are real. I think Urbit is our best shot at a better future.
Matrix is already competing. It's just taking a different strategy (going after governmental institutions and enterprises) than the one you'd like. The fact that Element is going after a market with a very long deployment/usage cycle makes me believe that Matrix has a lot more staying power than Signal.
Signal is nice, but it all it takes is one catastrophic bug - or one Government pushing to break it, or some good amount of years with lots of developers on payroll without any real revenue - away from disappearing or selling out.
> I think Urbit is our best shot at a better future.
No. The problem to "fix" is not a technological one. It's socio-economical. When (if?) individuals and SMBs learn TANSTAAFL, we´ll be able to have a diverse/healthy (and cheap) industry of professional service providers.
Sure, s/compete/win - I don't think they'll win (hence why they'll be a niche).
> "The problem to "fix" is not a technological one. It's socio-economical."
It's both - if you fix the bad incentives, you can fix the behavior. If you don't then you won't.
Because my idea of "winning" is one where there is no central entity controlling what should be basic infrastructure. My idea of winning is one where we have more options and that we can hedge one player against the other.
I don't (necessarily) want Matrix to grow to a point where it goes unchallenged. I want it to grow enough to the point where none of the other players has a clear upper hand.
I use Matrix already to talk with my immediate family, a good amount of my friends also managed to install it. Does this mean that I got to use Matrix exclusively? No, far from it. But I do have the option, and in the past when there were issues with WhatsApp and Signal, a lot of my friends were glad that all my nagging gave them an alternative.
"Bad UI", "complex onboarding", "technical issues"... those are all things that are easy to fix compared with the "issue" of having a oligarchy controlling all of the communication platforms of the Western Hemisphere and then China trying to encroach in.
I guess my idea of winning is like the web vs. gopher. I want the good, federated protocol to become the default. I want my friends to be using it not because I have to nag them to, but because it's just objectively a better experience. The federated nature of the thing being invisible and irrelevant to them.
I think today that's impossible - not because of issues related to Matrix, but because of deeper problems with our computing stack and the web. Federated systems can't match the UX of centralized ones and when they come close its often because they've recentralized anyway.
I think we agree more than we disagree probably. I just see Signal as good given the current tradeoffs.
Any attempt at establishing these comes through some central planning which leads at best to some cookie-cutter "one-size-fits-all" Apple-esque design that while functional, it's just a local maxima that destroys any kind of creativity and makes everything uniform and bland.
> Federated systems can't match the UX of centralized ones and when they come close its often because they've recentralized anyway.
I think it is easier to see it as an issue with the amount of resources that the big companies have vs the "underdogs".
If Element had the same amount of resources as Facebook, it wouldn't be hard for them to say "we will build a client that is just like WhatsApp for those that think that WhatsApp is the benchmark. For those that want something like Discord, we will build another client. And those that want to use Matrix to replace Slack will get another client that does exactly the same".
It does not matter if it is federated or centralized, what matters is that Element does not have those resources, so they can not develop as fast as their competitors and then are seen as "lagging".
Email is an example of this class of failure. Almost everyone uses a centralized provider (mostly Google) and even if you do go through the effort to run your own server since nearly everyone you interact with is using gmail it's mostly pointless anyway.
Signal is not controlled by the USG (see recent doc about what metadata they have access to via Signal). I ran a Matrix server for a while, the UX around setting up encryption is bad (not for lack of trying, it's just a hard problem given the constraints). Most people just use the Matrix.org server and will never run their own (which is the recentralization risk I'm talking about) - at best you'll have a couple providers, and dealing with spam is still a problem. You'll also have a system that adapts slowly because it's harder to make changes to this kind of system, it'll always be worse.
To escape the incentives that lead back to recentralization and to create a federated system that isn't just another niche nerd hobby, you really have to think about the issues that lead back to centralization from first principles. I think Urbit's design and the tradeoffs they make do this.
https://moronlab.blogspot.com/2010/01/urbit-functional-progr...
No. This is a false dichotomy. There is a very healthy market for email service providers. Basically every domain registrar runs one, a good amount of ISPs... there can be a cottage industry for service providers.
> Almost everyone uses a centralized provider
Because the large companies make it free to try to make their money by either exploiting the data or by using the email service as a loss leader. Signal can not do either, so they will have to rely on some other revenue stream, or they will end up like Mozilla.
I think this is just not true, my grandmother would use whatever I tell her to use she doesn't care if Signal has "better UI" according to some people. The reason in my opinion why people don't use Element or other Matrix clients yet is because people aren't aware of them. Signal has a billionaire behind them and that means they advertise the shit out of Signal.
> Email is an example of this class of failure. Almost everyone uses a centralized provider (mostly Google) and even if you do go through the effort to run your own server since nearly everyone you interact with is using gmail it's mostly pointless anyway.
Why can't Moxie and Signal just run a Matrix server and offer this service to people? Your point seems to be - we lost the fight already so there's nothing to do. I think there's a lot to do and Google/Facebook/Apple have not yet won. Even if they have most people at least the people who really care have the opportunity to stay private AND free. Many people still use Hotmail, Yahoo, university accounts, business accounts, privacy concerned providers, government accounts and even famous country specific providers. Gmail is the most used one but it's by far not the only one and the way to fight that is not by giving up on a free world.
> Signal is not controlled by the USG (see recent doc about what metadata they have access to via Signal).
Signal is a company registered in the USA they will have to abide the US law when push comes to shove, so this is not a long term solution.
> I ran a Matrix server for a while, the UX around setting up encryption is bad (not for lack of trying, it's just a hard problem given the constraints).
It's getting better and it's partially like that because it's more private(no phone number) and secure(no US company) than Signal, it also contains an encrypted backup on the server which is a huge feature that Signal just does not have.
> Most people just use the Matrix.org server and will never run their own (which is the recentralization risk I'm talking about) - at best you'll have a couple providers
Most people use WhatsApp maybe Telegram, not Signal and definitely not Matrix the future will tell us if this catches on or not, so let's just stop speculating. There is already now some governments using Matrix and a lot of projects, so there are already many servers and many providers, you can also just pay for your own server at Element and you're already not using Matrix.org
> and dealing with spam is still a problem.
That's a general problem that you'll have with any platform, Telegram and Facebook are still fighting spam with little many failures. If anything this is something that I see a good future with Matrix since a lot of different companies/individuals will try to fight this together instead of a big company trying alone and failing.
> You'll also have a system that adapts slowly because it's harder to make changes to this kind of system, it'll always be worse.
Or better because many people are working on it compared to 30 people (!!)
> To escape the incentives that lead back to recentralization and to create a federated system that isn't just another niche nerd hobby, you really have to think about the issues that lead back to centralization from first principles. I think Urbit's design and the tradeoffs they make do this.
I heard of this project before, it sounded cool at the time but I didn't get enough info the details to remember much, I don't really understand how this is relevant though, care to elaborate?
I don't think Element or Matrix is aiming to be a niche nerd hobby, they have a lot of money invested in them (last I read it was 30 million) and the UI is super cool (much cooler than Signal)
At the end of the day Signal and Matrix are both niche and it's just such a pity that we don't go for the obviously better choice but instead put all our eggs in one American basket yet again.
Protocols are slower to change, this is why XMPP lost to competing messaging services. Maybe grandma is a bad example, but a highschool kid is going to use the messaging service that has the emojis and crap they like - not one with a crappy app and confusing encryption setup. I don't think it's lack of knowledge I think it's just harder to create an experience as good and you're at a perpetual disadvantage to roll out new features.
> "Why can't Moxie and Signal just run a Matrix server and offer this service to people?"
Moxie mostly outlines the reasons here: https://signal.org/blog/the-ecosystem-is-moving/ - the answer is mostly the same as above, it won't be competitive and it'll lose.
> "Most people use WhatsApp maybe Telegram, not Signal and definitely not Matrix the future will tell us if this catches on or not, so let's just stop speculating. There is already now some governments using Matrix and a lot of projects, so there are already many servers and many providers, you can also just pay for your own server at Element and you're already not using Matrix.org"
This is kind of my point? Except that I'd go further and say most use FB Messenger, iMessage, or Discord. I did pay for an Element server for a while to support the Matrix devs. While I want them to succeed I think the existing incentives are such that they will remain niche. Most will use centralized options because they provide a better experience and most users don't care about federation. The UX is everything.
> "[Spam] is a general problem that you'll have with any platform"
Unless you think about it from first principles and actually come up with a way to solve it: https://urbit.org/understanding-urbit - if you have NFTs as a permenant pseudonymous ID with a low but non-zero cost, moderation becomes trivial and spam is no longer economically possible. Today centralized services have an advantage fighting spam (email is the easiest example) and decentralized services struggle with this.
> "Or better because many people are working on it compared to 30 people"
I think this has been shown to be empirically false with this kind of service. Not because of the number of people, but because of the reasons federation on the existing stack is hard.
> "I heard of [urbit] before, it sounded cool at the time but I didn't get enough info the details to remember much, I don't really understand how this is relevant though, care to elaborate?"
The pitch is basically that the problems that cause federated systems to fail (at the application level) run pretty deep. To fix it you have to rethink some assumptions with modern operating systems and how you handle identity across the web. If you fix these issues, then it becomes possible (but not easy) to solve the UX problems that causes federated applications on the web to fail and provide a better experience than the existing web for applications. If you can fix the incentives then you can escape the trap we're in.
> "I don't think Element or Matrix is aiming to be a niche nerd hobby, they have a lot of money invested in them"
You're totally right - they want to take over the world, and I want them to win too, I just don't think they can because of the current local maximum we're trapped in. It's moloch (https://slatestarcodex.com/2014/07/30/meditations-on-moloch/).
it does not, see Email
You could argue that's better than trusting one service (debatable imo), but you're still a thin client vulnerable to the whims and data protection capabilities of that provider. Even worse, if everyone is using just 3 or so main providers then since you're communicating with people centralized primarily on these other services you're also vulnerable to their security practices too.
All of this is because of a few core problems with the current stack. Linux servers are too hard to run. Modern dependency requirements are a mess, operating systems were designed without an eye toward the web or identity so spam is a problem. You can't fix this without thinking more deeply about these issues, I think Urbit might be a way out. Without fixing these issues the options available in the current system are niche communities or centralized services. The idea of people running their own decentralized nodes at scale given these constraints is dead on arrival. PGP is a failure - even when people's lives literally depend on it they cannot or will not do it.
Trust me, I wish this wasn't the case - but I care about winning more than I care about idealistic purity. You have to think pragmatically about what the best option is today and what crazy changes are required for this to succeed in the future. Today with the existing tech stack I think Signal is the best we can do at scale in competition with whatsapp, fb messenger, and iMessage. Tomorrow I hope urbit gives us a way out.
There are thousands if not millions of companies participating in the federated network, not including some nerds who self host