I don't know. If the experiments were perfectly random, the Buttondown-Substack difference is a 3-sigma effect, so the difference is unlikely to be random chance. But why? What's so special about one vs the other?
I realise it's much more complicated than this, that within a given ESP there are multiple bundles of senders and that better customers get moved to a better bundle etc, but my guess is that for any given level within Substack you're being bundled with an unusually-bad group of peers and therefore getting lower open rates -- I would love to know if this is an accurate understanding, or if there's any accurate way to test this from outside Substack.
This is a great theory. I'm subscribed to a bunch of Substacks, and most send fewer but longer posts which I always open and read. While others treat their Substack like a Twitter feed and send out lots of short posts a day which I hardly ever open.
Alex Berenson has already sent me 3 mails today, and the day isn't over yet...but there are other lesser known people who send far more. And some of those others are crypto/forex/finance writers who by the nature of their content, spam filters may view more suspiciously.