My best experiences working remotely—both during the pandemic and with folks in other locations before that—has been with research-oriented work. The reason? We had the time and space to work and communicate the way that made sense to us. We had plenty of shared pair/group sessions to work out design ideas and experiments, iterate on the system, debug trickier issues... etc. Everything else we could do well asynchronously. Since this was longer-term, more speculative work, we didn't have to worry about tickets or product/management breathing down our necks every day.
On the other hand, the projects that suffered the most were the ones where folks had less scope and autonomy. If you can't unilaterally make decisions for anything that spans more than a day, and you're forced into some ticket-oriented process where everyone has "their" task to do, communication naturally suffers. If you're all in the office, there's a bit more slack to talk quickly and informally outside your immediate silo; remotely, this takes more intentionality, so the friction imposed by your culture and process has an outsize effect.
So my conclusion has been, pretty much, the opposite: remote works better when people have real scope, independence and autonomy. If engineers and small teams can make decisions over longer time-frames, they'll have the space to adapt and work effectively. They might need somebody to set the example or foster the right habits, but it's fundamentally healthy and effective. The frictions with remote work come up when it's thrust against a less flexible culture that makes it harder for folks to adapt.