I’m about to hit the reach wall head-on, but I’m giving it a shot anyway.
For several months, maybe even a year or two, I’ve been saying here and there, “we need an alternative to LinkedIn.” Why? Two reasons: LinkedIn has become (or perhaps always was) a dumping ground for heavy and artificial marketing. And more recently, it has become the sole central hub for professional exchanges—I don’t really see other alternatives (https://www.xing.com/ seems to be the closest), and it’s not European. As a long-time Linux/open source user (Red Hat 4 in 1997?), I’m looking for something ad-free, without bogus algorithms, European, and open source.
I was looking for developers, friends who code, and two of them volunteered last year. But I couldn’t fund it. Especially since the success of a minimalist European LinkedIn clone is more than uncertain, as REACH is the name of the game. REACH: the ability to attract enough users for the service to actually have an impact, and the paradox: how to get them to come when that impact isn’t there yet. And REACH, having enough users, is a wall we often hit, regardless of the quality of the idea or the product.
So two visible walls: developing the solution, and taking a hit to the face from the lack of reach. Other invisible ones: production risks will teach me those soon enough.
The first one was solved in the last 6 months, thanks to artificial intelligence. I equipped myself with Claude Anthropic (Opus version). I have a very good knowledge of software development, but I’m not a coder (I was one a bit, but I was mainly a sysadmin). What I mean by that is I know—more or less—what kind of architecture I want to go for and why, I think I know most of the security and performance pitfalls, etc. Naturally, the various automated tests, unit tests, functional tests, etc. So I know how to set up the whole framework. But I haven’t read a single line of code from Ponos.
And so I wanted to embark on building a “poor man’s LinkedIn” (to use my eldest son’s expression, and I take it as a compliment: an “essentialized” LinkedIn), if only to see whether developing all of this, with quality, without knowing or seeing the code, just with the framework, was possible.
A week and a half later, I already had a well-advanced first version of the application that I put into production—it was Wednesday. I’m blown away. Structural changes made during development. Explanations and constant dialogue. It’s a powerful learning experience: I’m discovering new ways of doing things, new languages, new tools, I test them, I choose them. I’m blown away.
All the code for this open source project: https://gitlab.com/pablopernot/ponos You’ll see my code, my coding rules, my automated tests, my agents, etc.
No software difficulty: it’s a classic and well-known management application. No complicated algorithms. Here I’m on familiar ground; it’s more about choosing between uses than discovering new ones.
This shifts the entire challenge to ops, to production: the ability to handle the load, to resist attacks/hacks, to perform updates seamlessly and transparently.
Now I’m waiting for your help. You connect, you sign up, you upload a photo and write two or three descriptions, and from time to time you come take a look (no notifications in Ponos). Think of it as your backup account, just in case.
Then I hope not to stay alone (but not to be too many either). Having input from bizdevs, designers, architects, infra/ops people would be a plus. I have no idea what form this conversation group would take.
I’d like you to alert me to security holes and to let me know privately. (I have a dedicated agent for this, but that’s not enough.)
And then I’m kindly waiting for Elon Musk’s help –> Europe blocks Twitter –> American retaliation –> European blocking of LinkedIn.