I don't know how many look at the *MathOverflow Weekly Newsletter*, but these were
the "Greatest hits from previous weeks":

"Video lectures of mathematics courses available online for free" from Feb 2011.

"Undergraduate math research" from Nov 2010.

I assume this is an artifact of people editing old, popular posts. It doesn't put the best face on MO. (On the other hand, the "Top new questions this week" do represent MO more accurately.)

manyanswers are considered as 'good' (in a direct plus in an indirect way via total score of all answers). Anyway, as I said in one of the last discussions on 'tea' I think to make a good newsletter for MO is difficult (for AIandHI), to the extent one might best give up on the idea, in my opinion. For some other selection that is in part a bit, well, not optimal mathoverflow.net/questions?sort=frequent (but really only in part, it is actually not so bad). [For that selection number of links is key.] $\endgroup$Theoretical Computer ScienceNewsletter. In the Aug 1st newsletter, which I haven't yet deleted, the "Greatest hits from previous weeks" are: "What is the complexity of this edge coloring problem?" and "How to check if a number is a perfect power in polynomial time." These seem much more representative of TCS than do the "greatest hits" represent MO. $\endgroup$