If you have the choice to sell out or not sell out, the only logical decision is to sell out, because then you'll have lots of money and one presumes the product wasn't emotionally important to you. You can then move on to making your next product.
Couldn't have said it better. Whatever else you want to build in life will be exponentially easier if you sell out first, and many builders have many things they want to build and not just one.
Focus on open protocols, simple formats over complex vendor-specific cruft. Then you can always "fork" away from an enshittified saas.
I bet a small team of the quality of the kind developers who are attracted to hacking on Ghostty could recreate the subset of GitHub functionality they actually need in ~six months. It's just the problem of how to pay for the ongoing care, maintenance and hosting? Maybe another opportunity for Mitchell's particular brand of philanthropic OSS.
DNS is the cause of all problems, but it's also the solution - just like anyone can run Apache or Nginx, so should anyone be able to run a git setup. Then it scales really well, as everyone is doing their own thing on their own domains.
Of course, you lose out on some things like ease of user access and various protections.
You don't need this. git and a local drive. git and a shared drive. git and an https engine (can be a plugin to apache/nginx, not a full github like solution). git and ssh but people use username/password.
