The other day, I participated in a discussion on open-source adoption and was surprised by some opinions. One really stood out: Some technologists think their companies are still deciding on open source.

This is an argument from the early 2000s, perhaps even the late 1990s. Should we use open source in our enterprise?

How amazing to think that in 2025! The odds are that your company has been using open-source software for years, even if it has never explicitly made that decision.

After that discussion, I looked up each company the detractors worked for to find a trace of open source. I didn’t need to go past their websites; WordPress and React are amazingly popular.

How did this happen? A lot of it just came down to persistence and cost.

Why Open-Source Adoption Makes Sense

Open source is diverse, even if the contributors are less so. There are many reasons an open-source project exists, and I won’t attempt to go into each one. The case I find most interesting is “Important but not Valuable.”

This means essential bits of software that hold little value on their own but enable companies to build and sell other things that do have value.

Early computer companies used to share their operating systems because they made money on the hardware. And now… well, there’s Linux. If I had to pay Microsoft every time I felt like spinning up a pile of instances to do some work… I’d spin up a lot less of them.

The data center operating system is important but not valuable. That’s good for Amazon and Wet Dog Weather but not so much for Microsoft. Open-source adoption doesn’t destroy economic value; it shifts it around.

Open Source in Meteorology

On our meteorological side, we use a lot of open-source software with various motivations. Here’s a very unscientific sampling of some packages and their motivations.

Python: ‘Why would you pay for this?’. An enabling technology and science’s favorite computer language of the moment. Trying to make money off of this one would be… complicated.

PySTEPS: ‘Check out my thesis!’ This is common in meteorology and weather modeling, and it’s great. It is nice to have a way for young researchers to integrate their work into a cohesive whole, and sometimes, it can be directly useful.

pyGrib: ‘Use this to read our data’. If you make and charge for data, you want your users to be able to read it. See also: netCDF4.

Other packages, like metPy, are more like shared utilities that various governments view as valuable, so we share the cost. Again, it’s hard to charge for.

Is there a company in weather that doesn’t benefit from open source adoption? That would be incredibly weird.

Open-Source Adoption Is Nearly Universal

I didn’t even mention a variant we work on with MapLibre. Where one company tried to dominate an industry by making a package free, succeeded, and then tried to go closed source to force everyone to pay. Sometimes that backfires, and karma is a b…

Anyway, open-source adoption is like microplastics but good. It’s already everywhere. You can’t do much about that, but why would you want to?