I think it's a good thing to study it scientifically, but also to be extremely conservative in the application of results. Some very thorny problems just off the top of my head:

1. Monitoring/counting "spontaneous interactions" is extremely difficult.

2. Simple "counting" of spontaneous interactions seems like exactly the wrong approach, because one would probably expect that 99% of interactions would be mundane but then 1% would be important.

3. How do you define "innovation"?

4. One might suspect that the benefits of in-person interaction would show up on longer timescales rather than shorter timescales. For example, this article is long on pointing out that how in-person interactions can be detrimental to minorities and women, but my opinion (not fact, just opinion) is that basically all of my closest business connections were formed in person. Now that I'm late in my career remote vs. in-person is not as critical to me because I have a strong network, but right out of college it was critical, and my career would have suffered greatly had I not forged those relationships early. I got every single one of my jobs after my first through my network that started in my first job (not necessarily directly, but often times through the "hey, I heard from someone they are looking for XYZ, and I heard you might be looking" channel).

I was more talking about the article itself. Does it require a scientific study to poll experts and distill into an article?