Google won’t release the AlphaFold3 code.
(They say they won’t release it “yet,” but don’t hold your breath).
What a bummer. But to be honest, not a huge surprise. There was a lot of wringing of hands when OpenAI didn’t release the GPT-2 code. But people got over it.
Also worth noting, Nature’s submission policy states that:
Any previously unreported custom computer code used to generate results reported in the manuscript and that are central to the main claims must be made available to editors and referees upon request.
Unless its Google.
Not every commercial product requires a research publication. But a Nature paper, especially when it comes with a response from the journal, will certainly bring eyeballs to this new product. So is the AlphaFold3 paper research, white paper, or marketing?
Or all of the above?