I have found that what is in general most effective is that EVERYONE on the team is remote, or everyone is co-located. You can do hybrids, remote work with an on site team if the remote folks are strong communicators and/or people that are more task oriented (Assuming you have someone to manage them) but there is a tax on doing so, and often they won’t feel as connected.

Even in the above hybrid example, there are specific things you need to “plan in” to be effective (things like an office buddy, regular sync up times, clear responsibilities/action lists, etc).

I'm way more likely to ask a co-worker over to my desk to show them something or have a discussion if they're a few feet away within sight and earshot. In an open office I also overhear a lot of pertinent information -- I'm aware of issues that might be relevant to me as they develop, I don't have to wait hours or days for someone to communicate them in an email, issue tracker, or meeting.

There's a cost in terms of readily available heads-down concentration time to be sure, but if I want to mess around with math formulas or something I pop on the headphones, go to a quiet room, or do it at home. It's much harder to replicate the benefits of an open office environment in a closed one than the reverse.

Why does it make sense to optimize for the former instead of the latter? The benefits must be studied in context of how overall productivity and bottomline is affected, not just easy it is to start talking to someone.

Perhaps I'm weird, but I'm not interested in working remote. It's isolating socially and politically, and I work slower over the medium term.

This factoid detracts from the argument; When the op is about a (or several) company individually worth about a trillion dollars.