For me, the primary one is that it's _way_ too easy for everyone to get deeply frustrated with each other. Social interactions and chit-chat lead to a better understanding of people and a more accurate mental model of what people actually mean when they say something.
Written text is the majority of remote communication. That's both good and bad. It's a much lower bandwidth form of communication than face-to-face in terms of being able to pick up intent and understand the context of someone's mood/etc in what they're saying. The same goes of communication over webcam. It's higher bandwidth than written text, but still lower than in-person.
When you're fully remote, you'll much more commonly misinterpret folks intent. You'll also more commonly have your intent misinterpreted.
At least in my experience, this tends to lead to a much higher rate of frustration and more frequent extreme misbehavior when everyone is remote.
I don't have an easy solution. Occasional in person meetings help a lot, but they're hard to do well, and easy to do poorly. Having 200 people in the same room for 2 days doesn't really help the situation, as there's zero time to get to know anyone.