> It’s only confusing because companies like Microsoft intentionally abuse the term “open”.

Not only, but perhaps indeed also.

"Open source" reads as two separate words rather than a multi-word expression. The "open" sounds like as in a book, a shop, a museum: there's no implication of being able to do more than merely look unless you pay.

There's also "open" in the sense of the open/closed principle in e.g. SOLID, but even as a developer that's not the first thing I think of.

This is the exact other half of what seems to annoy people here about LLMs being called "open source" when they can be freely modified into derivative works but the training set is unknown: "open" yes in this sense, "source" no.

The LLM weight dumps we see tend to fail the "open" part too, with fine print clauses saying things like non-commercial use only.