To anyone sending emails (including posterous), I'd recommend going beyond A/B tests. You should know the optimal time to contact each user.
For new users you can intentionally send emails at various times (and days of the week). Then you can track a time-specific clickthru rate for each user. Eventually you'll converge on one or more optimal times/days. You can accomplish this with a machine learning algorithm or you can simply track the timestamp of each user's last N clicks and use that to decide when to email them. Precision isn't of the essence here, but I've found that clickthru rates can be dramatically affected by picking the right time of day and day of week.