I’m a software engineer and over the last few weeks I’ve been diving deep into messaging apps because of the Chat Control thing. I started wondering how something like that could be implemented, which sent me down the rabbit hole of how messaging apps work (E2EE, Signal protocol, servers, Matrix, etc.).
At the end, to my surprise, I realized it’s not that difficult to build a messaging app. The biggest walls are (1) infrastructure cost if it becomes popular, (2) getting people to join the network (why would I join if nobody else is there?).
I think I have an idea to solve the first problem, so I’ve been building a prototype, yes, with the help of AI, but it’s not vibecoded at all. I have plenty of experience as a software developer.
My question is: do you think there’s room for another private messaging app? The main differences would be:
No centralized server like Signal.
No phone number or any other identifiable information required to sign up.
Not as complicated to use or self-host as Matrix-based apps.
Easy to self-host: think “spin up a docker container and you’re done.”
Not VCs-backed or corporation-backed, just a small team trying to make it profitable to keep the servers running (yes it will have some servers)
Federated.. maybe, I know it’s a big thing, but I think it could work with what I’m planning to make.
SimpleX,Delta Chat, session, and Matrix even if you mentionned as comlicated.
I am not sure you should try to offer another option for decentralized messaging.
Even with with easy centralized messaging like signal. I have only succesfully convince 2 person in my life to use signal, all other refused.
Here people use facebook messenger or MSS,SMS,RCS, I have to communicate with them using email or MSS-RCS.
I guess that you are european, so it’s must be the same thing with whatapps (meta again). I am afraid you would waste your energy trying to gain momentum. If you have skills, maybe start another project that privacy friendly ?
I don’t think we need more chat apps. Signal is really good for almost everyone and SimpleX is good for no phone numbers. The only advantage you are listing to SimpleX are no VCs which isn’t big enough of an advantage on its own, especially with SimpleX’s governance and licensing being good.
If you want something to build that we need, I would recommend an E2EE tasks app, there’s not any killer option currently.
thanks for the comment, yeah the wall to onboard other people is definitely the hardest. I just think maybe the bigger awareness for privacy we are having, because of all these anti-privacy laws that are coming, and the anti-meta sentiment, can push people to other apps.
I kind of stumbled into privacy chat hosting myself - a small group of internet friends jumped through a couple of different social media places looking for something a little more trustworthy, and I ended up throwing together a Rocket.Chat server for them. I wasn’t terribly excited that you have to register with Rocket Chat when you create your instance, but I had already sunk enough time into the project that I didn’t want to start over.
Lately, I’ve been experimenting with Matrix quite a bit. There certainly are complexities, but I think you’ll get more milage out of making an existing protocol easy to deploy than you will by creating a new one. Between Ansible, Docker, and a mature FOSS ecosystem around Matrix, an experienced developer could make an onboarding process that is approachable for a less-than-technical user without sacrificing the advantages of a well-established protocol.
Here are the scripts and config files I have set up for a SaaS I am working on, if it will help. I will say that for my initial use case, I have federation disabled by default, which may be something you want to change.
First seems vibecoded but I’ll look more into it, no iOS app and the second is an overwhelming proprietary one where you only see the first two days of tasks. So doesn’t seem to be saturated.
Famous last words before some cryptographic related CVEs come knocking on your door
Signal is free. If you want money, now you link payments back to people, which is way more identifiable. Maybe you’ll take crypto only, but that’s a whole other can of worms to operate as a bsuinees.
Federated is notoriously complicated. Matrix is still kicking itself and failing adoption because of the jank. And that’s led by people who understand cryptography.
We should welcome fresh ideas into the space, but if you plan on implementing a new cryptographic protocol and have never worked in the space: at best you’re in for a rude awakening, at worst you’ll endanger real people if you aren’t able to make something safe.
I believe are are simpler E2EE projects to tackle, messaging is probably one of the hardest. Try to divert that enthusiasm to another place in privacy which lacks movement.
Thanks for the comments.
I’m definitely not implementing a new protocol, just reusing signal protocol and all the battle tested cryptography.
What happens to me with Signal is that I don’t think donations is a sustainable model, or what if there is one super big donor that wants a seat on the board and comes with some agenda. Or they start to gain popularity so infrastructure gets more expensive, but the donors count don’t increase. Anyway, that’s related to the business side, it would be great if there is a messaging app that is profitable on it’s own.
I don’t recall having met anyone that has paid or will pay for a messenger. People use less than ideal services where they are the product and are happy. And for in between we have Signal and Matrix.
Imagine you want to message a friend, and they say you gotta pay $5 to message them. I’d say no lets just use Signal.
Personal bias here and I’m gonna be very blunt. Apologies if it’s rude:
No. No no no no no no no no no no no. Please dear god, for the love of everything, stop making messengers. We have way too many messengers already, and while there’s always room for improvement, at this point we’re just reinventing the wheel for a very, very tiny RoI, IMO.
Meanwhile I can’t name a single to-do app off the top of my head - like an actual “to-do app” where you can checkbox things and they disappear. (Cue all the people leaving suggestions that look hideous - more on that in the last paragraph.) There is no “task” app on par with the way that Google integrates their “to do” list with their calendar (granted that’s something that calendar providers like Proton & Tuta need to address, not you, but just saying).
Some of the things we’re missing have a high barrier to entry (I’m thinking VoIP & payment masking) but many don’t. Where are the budgeting apps? Where’s the language learning apps? Health and fitness? There’s ONE food tracker that looks like it’s from 1995 and the food database is incomplete (that’s a crowd-sourcing problem, myself included). Where are the security cams/smart doorbells that can be self-hosted? Shared grocery lists with your partner? Where are the private marketplaces that accept Monero or Bitcoin? Where’s the consent-based location tracker so that when the single 95-lb girl goes on a blind date she can let her best friend track her location for the next 24 hours for safety? I have to go prep for the podcast today but if you really want, give me half a day later this week and I’ll come up with a literal list of ideas that it’s like “I have at some point wanted this app myself or known someone who did and there’s no real good options here.”
I’m sure some of these have problems I’m overlooking (the doorbell one probably requires hardware to go with it) but I’m sure many are more solvable than we realize.
I read this question and went and asked my wife to name stuff off the top of her head as a “normie.” Her IMMEDIATE answer is the UI. We need to fix the UI for most privacy tools. Signal looks pretty. Proton looks pretty. Pop OS looks pretty (my wife LOVES Pop OS and how cute and cozy it is). A lot of privacy tools look like they’re form the dot com bubble, if they have UI at all. If we want a privacy tool to have mainstream appeal, it needs to be cute. It needs to have color and look attractive. It needs to capture the neurodivergent crowd. We’re not doing enough to appeal to them. (I’ve been wanting to write a blog post about this for a long time but I’m having a hard time putting it into words since I’m not neurodivergent enough to really relate.) We need to stop making everything look like The Matrix and start making it look cute and friendly and appealing. That doesn’t necessarily mean make it look like a crackhead on a manic ADHD episode designed it (thinking of a very specific social media startup I saw the other day when I say that), but think like ente. Ente’s duck is cute af. Honestly a little shocked I haven’t been able to get my wife to switch with how cute that sucker is. Anyways, my point is that if you’re not interested or qualified to work on any of the other projects, try to find the ugliest one you are interested in and offer to help rebuild the UI.
Thank you for coming to my TED Talk. Tip your servers. And again, apologies if my “direct” was too “rude.”
That doesn’t necessarily mean make it look like a crackhead on a manic ADHD episode designed it (thinking of a very specific social media startup I saw the other day when I say that), but think like ente. Ente’s duck is cute af. Honestly a little shocked I haven’t been able to get my wife to switch with how cute that sucker is.
Hot take, but I’m not a fan of Ente’s duck…
Like they couldn’t have chosen any other mascot? There aren’t that many privacy focused companies out there. But they had to use the same animal as the most popular privacy focused search engine company?
Duckduckgo at least sort of tells you what it is as it’s based off the pre-school game, duck duck goose. I like to think of it as which link in my search results am I going to pick before venturing off.
A duck mascot says nothing about an encrypted cloud photo gallery.
The name, Ente, also reminds me of the pokemon, Entei, a legendary fire beast, which easily overshadows a little duck.I
I think the only space that has yet to be filled in terms of messengers is a centralized alternative to discord that offers the same convenience and more privacy. This might be impossible to do, at least at scale. But its the only thing that no one has done yet. Honestly though with the way things are going it seems like nothing will be safe besides self hosted instances
And honestly that’s fine. “Cute and colorful” isn’t going to resonate with everyone. But the point is that we don’t need everything to try to do the “Hackerman” look. Will that resonate with some people? Sure. My wife thinks it looks super cool to run updates from the terminal because it looks “hackerman.” But it’s also really cool that you can run updates from the app store (in Pop OS) like a normal, mainstream distro. Zorin had a moment because it emulated Windows & Mac instead of asking people to relearn a totally new GUI. My wife doesn’t want to do everything from the terminal and she doesn’t want everything to look dated. She loves that her desktop is cute. She loves that Brave “just works.” She loves that Signal has GIFs and stickers and uses them generously.
These are the tiny little things that will keep “normies” around. My wife is a pro at getting people to use Signal and I don’t think she’s ever once led with “it’s secure/private.” She leads with things like “we can do group chats with GIFs and stickers and emoji reactions” (before RCS became cross-platform) “and we can send voice notes and bigger attachments” and then at the very end she might mention “also btw it’s like, super secure and private. Nate knows more about that stuff, you can ask him. But here’s why I like it.”
At the end of the day, to me, it’s about options. We’re lousy with options for messengers. Name a type of messenger you want to see and I bet it already exists in a form that’s at very least “close enough.” And that’s fine. I don’t care that those exist. I do care that those exist to the detriment of everything else. Even when you subtract for the privacy tools that have a high barrier to entry like infrastructure or regulation (VoIP, payment masking, email, etc) there’s still a whole ton of options out there that either don’t have an app at all, or only have a couple options, and out of those options they all either look super outdated, are missing critical features modern users expect, and/or haven’t been updated since I was in high school. (There’s a separate discussion to be had about how a lot of these are “for fun/on the side” projects and where’s the reasonable expectation of responsibility there, but like I said, that’s a different rabbit hole.)
