Care over time is the only software success metric now
As far as I am concerned, I now think that care and length-of-presence are the only two qualities I want from a software project now. Elixirforum, for example, now has a continuous input stream of new libraries and projects by enthusiastic authors. I am sure they mean well and are genuinely enthusiastic about their project !
But the jump in ability to produce comes with a jump in ability to let go, and the new project can be abandoned for an even newer project. This pushes me to not release or announce anything that I am not dead certain to depend on anymore, because it is easy to see this pattern in others but we need vigilance to see it in ourselves too. I’ve been guilty of it too and published (now deprecated) libraries 6 years ago due to enthusiasm.
Because effort has gone away from the actual coding and publishing, we have less ability to judge if something new-ish will be able to be trusted. We can initially only trust a gut feeling : who is the author, what are they doing ?
Authors can provide:
- Presence: A new library could be trustworthy if the author has been active in the community for a long time.
- Stakes: It could be trustworthy if the author depends on it for something table-stakes at their business.
- Interests and personnality: Or it could be trustworthy if the author has a track record of being heavily invested in the field of this library.
It’s not a lost cause though ! The time to initial release has been compressed so we don’t get the “proof of work” of the first build. But authors can gain trust by showing care over time. Proof will build up. It’s hard to trust a newly vibe-coded library, but if 12 months later, it has seen continuous activity, others joined, and bugs get fixed, it’s easier to trust the dynamic.
Hardware projects get a bit of advance on trust since a harder cliff must have been climbed… but their software part compressed too.
Care over time stays the primary metric.