Back to Fundamentals
When the ecosystem stops being the main reason to pick a tool, we can choose software on its fundamentals again.
For years, picking a stack wasn’t really about the stack. It was about the ecosystem around it.
You didn’t pick React, PHP or Postgres because they were the right abstraction for your problem. You picked them because of what came with them. This one has an OAuth provider with one-click setup thanks to a well-maintained package. That one has a Postgres extension that does exactly the thing you need. We were shopping for dependencies, and the framework was just the store they came in.
That’s changing.
Dependencies are cheap now
A lot of dependencies can now be built from scratch, and they make very good replacements. Not all of them, and not the hard ones like a database engine or a crypto library. But the long tail of small packages and glue code that used to tie you to an ecosystem? An AI can write the piece you need, shaped exactly for your use case, in an afternoon.
When the ecosystem stops being the main reason to pick a tool, you’re left with the questions we skipped for years:
- Is this framework the right abstraction for what I’m building?
- Does it give me the speed I need?
- Does it expose the tools and APIs I actually want, or just the ones the community happened to build?
Those are fundamentals questions. And now we can actually answer them on their merits.
Electron is the obvious example
Most people pick Electron because of the ecosystem: npm, the web platform, a huge community, a component for everything. Almost nobody picks it because shipping a browser with every app is the best way to draw a window.
If I can rebuild the little dependencies I need and I don’t lean on the community for every piece, the trade-off flips. I could reach for a GPU-rendered Rust framework instead (the kind Zed built with GPUI) and forget about the browser altogether.
Software has been in a slump
Developers take a lot of heat for software being slow, and some of it is fair. Apps got heavier, startup times got longer, and simple tools started eating gigabytes of RAM. A big part of that comes from choosing ecosystems instead of fundamentals.
If the ecosystem tax goes away, we lose a lot of excuses. We can offer good software again: fast, light, built on the right foundation.
Giving a hand back to hardware
It’s always been the other way around. Hardware got faster every year and quietly subsidized developers. Every inefficiency got paid for by the next chip.
Now hardware is short and expensive. That makes this the right moment for software to return the favor. Pick the right abstraction, cut the weight, and make the hardware people already own feel fast again.