> You may also want to look at the timing the list software uses for > deciding when to notify people they've had excessive bounces; I got > this on April *24*, fully two weeks after I stopped getting messages; > a similar thing happened on March 21, but that one was only one week > late:
Modern anti-spam measures, especially google's, have gotten fairly non-deterministic, at least to outward inspection. It is especially difficult to reason about delivery timing. For example, the list server may well have had the notice in queue trying to get google to accept it for a number of days. Google is known to hold things for several days, progressively showing them to a few more people, until it decides whether they're spam or not. A single-address notice wouldn't seem to be a candidate for that behavior, but _I_ wouldn't bet any money against google comparing such a single-address message against other singles that were statistically (or machine-learning-y) similar. Yes, two weeks seems long, but if a number of list posts were queued up on the list server not getting delivered to you for some days, and google deferred some of them, and google deferred the notice, and google held the notice after accepting it, that could actually pretty easily add up to more than two weeks. Mailman sends the notice at the point it makes the decision. There are some knobs for how many bounces over what time period result in action. De
