Of course statements like this are just rationalizations - not arguments employed to arrive at objective conclusions.
For other experiments, YC has discussed publically (fellowship, accepting people with no idea, early decision) there were what seemed externally at least to be well-run experiments with multiple accepted and some longer-term tracking to see how that worked. No remote seems to finally be getting that proper experimentation and treatment, I guess we'll see how that goes.
There are a lot of benefits to in person, I personally prefer working on things in-person to remote (in biotech, in-person really matters), but it'll be an interesting experiment this batch to see how much room there is for fully remote startup development across all the sectors they invest, both in terms of the batch itself and the companies individually.