Replies to post #13017 on The Question and Answer Board (IHUB)

[Bob Zumbrunnen]

Your figures are different from mine but the trend is the same. I'm showing the current 7 day moving average right at 1k posts per day with a high mark of 1281 in late March. I use a 7DMA so it blends the relatively slow weekend into it with the fewest meaningless dips and spikes.

On the flip side, we got yet another record number of page views yesterday (and this excludes the considerable traffic I and others are making in the csstest directory -- some folks, myself included, are in the habit of using the site solely from there) and the average is definitely trending higher. Pretty strongly. That's more meaningful where Alexa is concerned. Once I put the latest CSS stuff into production, clear that directory, and redirect folks back to the production directory, there will be another spike.

This points out something that intrigues me, though.

I used to think the hit/post ratio was pretty stable. It's not. According to my figures, average hits per day have climbed about 20% in the past 30 days while posts per day have dropped about 20%. I'm somewhat curious why that is.

Any ideas? It seems counter-intuitive to me. Posting activity dropping this time of year is normal and predictable (end of January effect, nicer weather) and bear markets always have a depressing effect on posting. The drop in posting volume makes lots of sense and doesn't bother me. Actually, it should be a larger drop, but we're still in a stage of stronger growth than a more mature (age-wise) site would be.

The increase in hits is the part that puzzles me. Why are more people reading if fewer people are posting and there are also fewer new posts to read? And at a time when hits should be dropping (if you assume they should follow the same trend as posts)?

There are probably a bunch of possible reasons, all of which would be difficult to identify.

Getting back to the Alexa thing, I haven't charted it, but my WAG is that we'll be ranked in the top 20k by the end of June. Might be a good idea for a WAG contest. If I had to submit a guess without benefit of charting it, I'd take 19,438.

We're getting into territory where gains in ranking are tougher to get because we're grouped with a lot of pretty busy sites. Where an increase of 10k hits per day might've moved us up 20% or so in the rankings when we were ranked in the 80k range, now we're at a level where 10k is far less meaningful and might just move us up a few points.

The rate of ascent (in Alexa ranking) has already slowed pretty dramatically though we're continuing to set daily hit records.

[Mattu]

Does Gary ever plan to return to the boards?

Haven't heard from him or seen him around any of the old digs in a long time.

[Ben Dover]

Do you trade stocks in the stock market?

[Bob Zumbrunnen]

Was noticing something looked a little wrong with my 7DMA tracking of posts per day. Weekly peaks and valleys I didn't think should be happening since any 7-day period will only have 2 weekend days in it.

Looked at my source code and indeed there was a mistake. Your figures were more accurate than mine, although yours were a tad on the low end. Not much though.

I'd made a rookie error. You know how a lot of beginners get tripped up by zero-based counting, so, for example, the 8th element of an array is identified by a 7, etc? Well, this was similar to that.

I'm storing posting totals in an array so I can do my averages. Figured "Okay, subtract 7 from the current subscript, add them all up from there to the current one, divide by 7, and I've got a 7DMA."

Unh-uh.

If today's the 29th and I subtract 7, I get the 22nd. Problem is, from the 22nd through the 29th isn't 7 days. It's 8! So what I was doing was totalling 8 days of posting and dividing that by 7 and calling it my 7DMA.

Can't believe I did that! Especially since I *never* have trouble with zero-based arrays and such.

I fixed it, and the numbers leveled out like I thought they should've to begin with.

I think there's a saying that goes something like "Figures don't lie unless liars figure." Well, it should be "Figures don't lie unless liars figure or the figures were produced by a programmer who threw the figures together too quickly."

PS. I'm still excited about the Alexa score. Especially since for the past 4 weeks they've been making sure only unique pages viewed by unique users are getting counted. I don't know how many people let Favorites autorefresh, but it's good to know that's not impacting the score.