I assume you've been looking at the $NORM graph. It goes up, it goes down, but it doesn't look any different now as it did 5 years ago. It looks like... the norm :).

Maybe a little on the low side, but nothing to be alarmed about. I would not attach any significance to it until it reached either 7, or 13.

PTAV, another of jcwren's Grand Unfinished Projects gives you a good indication of long-term trends, as it gives you the count of number of top-level nodes posted per month. Basically, the monastery hit cruising speed in 2001, and the traffic has been pretty stable ever since.

I think nodes garner less responses these days, so yes, in some ways, the site is a bit quieter than it used to be. But it still seems in good health to me.

The $NORM graph only covers 30 days .. I don't see a way to extend to a longer period, which is what would show any long-term trends. What was $NORM, say, a year ago? It's those long-term trends that I'm thinking might possibly tell us something interesting, not the minute day-to-day fluctuations.

People are upvoting less ... More nodes are getting downvoted

It could also mean there are more posts relative to active voters. Depending on one's perspective, this could be a good thing. Just from my anecdotal, off-the-cuff experience, it does seem like nodes these days get less attention than they used to. But I've been wrong about things like that before, so take it with a grain of salt. :-)

Does $NORM tell us anything useful about the "health" of the Monastery?

Sort of, but it's risky to interpret, since it collapses two different measures--volume of posts and volume of votes--into one number.

From casual observation, posting volume has been going up. Inferring from $NORM, voting volume hasn't been going up as fast. It's difficult to say whether that relates to post quality or not.

Something I do wonder about is, if there is a statstic about root nodes versus child nodes. IMO this differentiates activity versus re-activity: Is the community appreciating asking more than answering? This statistics might need to be related to the base frequency but might be interesting.

Do you know something about this?

Bad karma: Java and .Net start taking over the world. Former Monks have to take jobs using these technologies (or start selling hot dogs) so they can't hang around the Monastary as much. So there are less votes and less response to nodes. The new people coming to the Monastary are the kids out of school who now maintain a few Perl scripts sitting around and come in just long enough to get an answer to their question. They post poor nodes and never get hooked or start voting themselves.

Good karma: Perl experiences an uptick in interest, so more new people come to the Monastary looking for help. Since they are new, they post more new nodes without looking around first. These nodes get less attention and possible down votes, but if they really want to learn they eventually find their way around and get their answers. So short-term $NORM is down, but if some of these users stick around and contribute, it trends back up and Perl stays popular.

Reality? Who knows. You'd need all the voting data, plus data on server traffic over time. And it would still be speculation.

What I've noticed, and you didn't include it in your list of possible causes, is that the number of posts by "anonymous monks" seems to be up. I, for one, am less likely to vote on an anonymous post, even if it is useful. So maybe the anonymous postings are "diluting" the results.

And even if people voted in the same way for anonymous monk posts as for others, the $NORM would still drop. Anonymous monks don't vote.

Of course, it could also mean that more and more people realise that XP is a pointless numbers game, not unlike the lottery (except that there's no price money), and no longer bother to vote. Or that the novelty of voting wears off over time - it no longer produces the thrill.

Or it could mean there's an increase of the number of posts per user (or rather, per vote cast). It's still possible for $NORM to drop even if no downvotes are cast at all. And even if every one logging in on a day casts all of their votes.