Open source development has been running on remote work forever and it’s doing great.
The recent article about OpenAI and Google having their lunch eaten by a bunch of remote hobbyist show how this thinking is flawed.
One bias is that most open source projects are people solving their own problems. They have a deep understanding of the issues at hand and the developers working on a product are also the one using it. This is also why the open source community produces a lot of developers oriented tools. It is also why a lot of remote companies are working themselves in for the tech sector.
Compare that with a company that produces software for other industries. The developers working on a product are not the one ultimately using it. One of their main challenges is to ensure that the problems/goals faced by the users are ultimately properly understood by the product teams and developers. This is far more complex and requires far more communication the further away the dev is from being an expert in the target field.
I want open source language models to be good and succeed but be realistic.