It seems that there is decent evidence that a group meeting that tries to come up with new ideas does worse than if the same people did so individually, and only then had a meeting to present their best ideas. And by “does worse” I mean that they come up with fewer good ideas, even when they are quite confident that they have completely explored a problem.

Personally I doubt that having face–to–face meetings is ever truly necessary. Once everyone in the group has thought of several ideas, they can write a page or two description of it and why they think it is a good idea. Then the decision makers (which obviously might be the same group) can read those papers and decide between them. Mozilla used to work something like this specifically to allow contributions by volunteers outside the company (no idea if they still do though). The Rust programming language is developed in a similar way: anyone can write an RFC to propose making a change to the language (or the associated tooling, the standard library, etc).

I'm not convinced by this. It works but for a subset of problems, not all of them. And it overlooks the fact that people might build on top of each other, I don't think that was taken in consideration. What about exploratory meetings, where people are trying to understand a problem in the first place, those happen to be synchronous because answers evolve in other questions.

Keep in mind that I never mentioned face-to-face, I just asked for synchronous time.

That being said, I'm open to explore the idea, I think I already do this for a certain subset of problems using Slack threads

“Face–to–face” here covers any type of video conferencing, telephone calls, or even text chat that is happening in real time. As you said, these are synchronous interactions instead of asynchronous interactions.

> And it overlooks the fact that people might build on top of each other,

In the research that I read, the researchers came to the conclusion that most of the meeting time went to building consensus instead of actually thinking of new ideas. Some subset of the people in the meeting would be charismatic enough to dominate the discussion, and everyone else would contribute less than they otherwise would have.

> It seems that there is decent evidence that a group meeting that tries to come up with new ideas does worse than if the same people did so individually

Links to any research on this? Anecdotally (about a decade working across different sized orgs in various roles from IC to Staff Engineer), this simply does not pass the smell test for me.

I was looking for my copy of Peopleware where I first read about it, but I cannot find it. Perhaps I loaned it out.