When Gmail declares a message a spam and moves it into one's spam folder it undoubtedly keeps a record, even a hash of it, and flags any subsequent copies of it equally [at the backend, prior to delivery].

However, when a recipient discovers obvious non-spam matter in her folder, and presses "Not Spam" button, thus moving message(s) to Inbox....

... does Gmail automagically receive a notice of that (and what)? Is frontend "surreptitiously" communicating this to backend, so Google may update its spam-or-not "files"?

[Background: recently, I had Gmail declare a couple of Salon.com's daily newsletters as spam. All named "Today in Salon: (non-repeating line content)". I looked them over, didn't see anything inside that might warrant antispamrobot's attention.. After "unspamming" them the newsletters continue to be filed normally.]

I would guess they consider both marking as spam and marking as non-spam as data points, to whatever degree... but that's just a guess of course.

They may even consider the "normal" deletion of *unread* messages as data point here to count against a type of message. This is again just a guess, but this suspicion started here when I noticed that the alerts from my tracker script – alerting me of new posts on Google's blogs – were now often landing in the spam folder. And my behavior was this; I often just read the headlines of these tracker messages, without opening the message, and then immediately hit "delete" on them. I now changed this behavior, and unmarked them from being spam, and they're now apparently delivered correctly. But guessing just what brought the change is nearly impossible from the outside (perhaps marking the tracker emails as non-spam once "cleared" this message type for the future, which looked suspicious to Gmail with no relation to me deleting unread messages). And maybe Google doesn't want to share too many details about it either, in order to protect against spammers...

All I know for sure Gmail does not initiate a report after each local un/spamming change – no detectable accesses that I can sniff (at). So, if it does report back, that means it keeps an internal map of folder AND message-ID status in these, and then flushes them back to mothership on each subsequent refresh. Of course they wouldn't tell us, that's why we exist, the Google Kremlinological Evidence Unzipping Trail Force, currently housed @ http://blogoscoped.com/forum

[observe no trailing "." despite obvious need for one!]

"All I know for sure Gmail does not initiate a report after each local un/spamming change – no detectable accesses that I can sniff (at)"

Assuming you mean network traffic sniffing, surely the fact an email has moved folder MUST be sent to Google, so they can detect it from this.

=>no detectable accesses that I can sniff (at).

interesting. what sniff s/w are you using ?

Stephen, given each folder "move" is in reality but internal pointer reassignment, no message data gets moved physically, I'm not entirely convinced it gets reported IMMEDIATELY BACK to server. What's the point of all that advanced multilayered DHTML/ Ajax stuff, if you cannot re-render already wholly memory resident structures without server assistance? Then again, I may be hallucinating, and indeed every single event gets sent back at once (apart from window resizes?).

/pd, sniff in metaphorical sense... I just watch them lamps go blinky-blinky on me 9600 baud modulator-demodulator box, Gov. ;-))