Advice to Mollom

Since they asked in the latest survey, i said...

Here's an important thing we need to know: when do we get spam because Mollom's limits for the free service were reached, and when is it because Mollom isn't recognizing it?

and would most want:

To be able to set the range of the "edge" between messages mollom is sure are spam and messages mollom is sure isn't spam. That is, to set the range of the not completely certain area and send those to individual moderation (which Mollom could of course learn from). Ideally the components of the spam score for each message would be shared, and the administrators own certainty of what Mollom is doing a good job at and what it isn't could be used to set what precisely what goes to human moderation (trust Mollom's IP based blocking but not its link-suspicion for instance) or even tweak the weights of different components in the final score.