It’s Not Science Without Graphs

Posted by Chad Orzel on May 4, 2008
(7)
More »

As a sign of what an enormous geek I am, here’s what I did to pass the time while Kate was getting ready for the wedding we went to yesterday:

i-2379ec813e2f644126c85c245b3bd4fa-blog_traffic.jpg

Yes, I amuse myself by making graphs. If I knew Python, I’d be an xkcd character.

Anyway, that’s the monthly traffic for this blog from January 2006 (when I moved to ScienceBlogs) to the present. As you can see, April 2008 was the third best month since the move, thanks to Reddit picking up my post on What Everyone Should Know About Science, and in influx of crazy people. Thank you, Reddit, thank you crazy people.

The really interesting thing about this, to me, is how little difference anything makes to the general progression. The two big spikes are from when Dave and I got Slashdotted for the Blogger SAT Challenge, and the enourmous BoingBoing/ Digg/ most of LiveJournal spike from Many Worlds, Many Treats. The striking thing is just how spiky those spikes are– traffic goes way up, and then returns to right where it was before the spike. It’s even more striking on a weekly or daily graph, but that’s also kind of silly.

What we see in the graph is a slow and steady increase in traffic, basically doubling over the last two years. The reddish line is a linear fit to the data (or at least what passes for one in the new Excel, which, by the way, sucks even more than the new PowerPoint), excluding the spikes, and it shows an increase of about 1400 page views per month over this span.

It’s sort of impressive how robust this trend is. I’ve occasionally thought that traffic was really taking off in a sustained way, but it always comes back down. And there have been stretches where I’ve felt like I was just treading water, but they still show an incremental increase in traffic.

It’ll be interesting to see if this trend gets broken in the coming months. There are two major changes coming down the pipe that could fundamentally change the blog: one is FutureBaby, whose arrival may very well lead to a major drop in posting frequency; the other is my book, whose arrival will undoubtedly include some non-blog marketing, which might well spill over into more blog traffic. Or maybe those two things will cancel each other out, and we’ll stay right on that 1400/month trend.

Anyway, it’s reassuring to know that if I want to get to 100,000 pageviews/month, I don’t really need to change what I’m doing. All I need to do is stick with this another two years…

(7)
More »

Comments

  1. #1 Colin M
    May 4, 2008

    FWIW, Python is quite easy to learn ;)

    Also, it would be interesting to see a 30-day moving average of your traffic, rather than a best-fit line.

  2. #2 Chad Orzel
    May 4, 2008

    FWIW, Python is quite easy to learn ;)

    I’ve actually done a tiny bit with Python (well, VPython, the package used for some stuff in the Matter and Interactions curriculum). I wouldn’t claim to know it, though.

    Also, it would be interesting to see a 30-day moving average of your traffic, rather than a best-fit line.

    It probably would.
    That will need to wait until I’m both bored and at a computer with a real graphing and analysis program on it, because I’ve done all the wrestling with Excel 2007 that I care to.

  3. #3 John Scalzi
    May 4, 2008

    I do think highly linked posts ultimately contribute to readership increases, but I think it also matters what’s in the post that gets linked to. I wrote about this a couple of years ago, looking at the difference between a substantive post of mine that was highly linked, and one that was just goofy (Being Poor vs. Bacon Cat). The goofy one got more hits fast, but didn’t seem to keep people around, the substantive one got fewer hits at one time, but seemed to convince more people to hang around (it also still gets new links and visits).

  4. #4 Jennifer Ouellette
    May 4, 2008

    I knew I’d become an uber-Geek when I was down with the flu, running a spiking fever, and started charting my changing temperature every hour to relieve the boredom… I’d been learning calculus, you see, and had some vague, delirium-induced notion of calculating the slope of the tangent curve when I done. Except I didn’t end up with a smooth curve, as any physicist could have told me. Not enough data points! Lesson learned…

    But man, THAT is geeky! :)

  5. #5 David Pace
    May 4, 2008

    Plots just make the information easier to understand. I wrote a blog post about my commute time to the lab because I wanted to see whether any particular day of the week had a shorter travel time. Sunday was the best, obviously, but Wednesday was also faster. Graphs make it science, and graphs make it fun.

  6. #6 Wilson Fowlie
    May 5, 2008

    I’ve actually done a tiny bit with Python … . I wouldn’t claim to know it, though.

    Can you pseudocode? If so, you’re at least halfway there.

  7. #7 Greg Laden
    May 11, 2008

    Perhaps you think that the upward trend is in increase interest or a buildup of loyal readers because of what you do on your blog. And that could be true. But it could also just be the overall Expanding Internet Effectd (EIE). All numbers are steadily going up at this rate. Baseline inflation.

    I have exactly the same phenomenon on my site, and I’m pretty sure there is almost no relationship between what I do and what happens, other than those spikes. For the spikes there is usually no relationship between what I do and what happens either. But it feels like a lack of relationship of a different kind.