> * In-person is much better if you don't know what you need to build, are trying to "figure it out as you go along", and are relying on the implicit fast feedb

> * In-person is much better if you don't know what you need to build, are trying to "figure it out as you go along", and are relying on the implicit fast feedback loop of standing next to each other. My anecdotal evidence is different. I've spent the last 2 years building a large, complex, and ambiguous system working almost exclusively remote, and half of it from another continent. For the first year I never met a colleague in person. However, all this time I have been working closely with another engineer. Both of us are senior and very tenured, got good rapport and were highly aligned and motivated. We were able to execute efficiently, sometimes spending hours huddling in Slack per day. So my feeling is that remote works in some scenarios and not others, particularly involving new or junior engineers.