Reading the Twee Leaves: science or just stats?

The big A-lister cage match on Twitter

So this afternoon, search a fascinating debate happened on Twitter. And the topic? Well, story Twitter. Or more specifically, how data and tests about Twitter behaviour are packaged. Can we call the use of those stats “science”? Dan Zarrella makes a living saying yes. Jason Falls begs to differ. The results? Entertaining for the rest of us.

Enjoy the show!

Below, I provide an exhaustive blow-by-blow of their conversation. First, note: I respect both these guys. But I admit a bit of bias. Though Dan Zarrella has lots of interesting insights about how we behave on social media, I’ve always been curious about his characterization of himself as a “Social Media Scientist”. Clever positioning. But is what he does really “science”? You decide.

Background reading if you’re not familiar with the combatants:

Please let me know what you think in the comments: who won? Does it matter?


You should see the story above. No? Click here for the whole match.

(Images from Flickr used under Creative Commons licenses: Dan Zarrella – by Technosailor. Jason Falls by jdlasica)

10 thoughts on “Reading the Twee Leaves: science or just stats?”

    1. @JasonFalls Storify makes that obscenely easy. Took me 1/2 an hour actually. Not that it *wasn’t* a waste of time. Just not much. My question is how did you manage that whole exchange. Were you in the keynote the whole time or did you step out?

    2. I disagree. You guys argued about something that’s actually an important to look at, and yet, who is talking about it? Not the best time to tweet thing – the math vs science issue. Obviously there’s a huge gap in understanding when to use which and at what point data becomes useful. @JasonFalls

  1. 90% of Internet fights are solved when people agree to the terms. The fact is that HubSpot gives you interesting tools for analyzing your content, but it’s not “science.” However, claiming the mantle of Science can be a great differentiator when you’re competing with products offering just Metrics.

    Dan has done a lot of interesting stuff over the years, but I don’t see much in the way of replicating his own experiments. If he did, we wouldn’t see a single post about “What’s the best time to drop a link-tweet,” and instead we’d see “Over the last 18-months, we’ve seen a decrease in the importance of day of the week, and more on time of day…”

    Everything seems like a one-shot stab at sating curiosity — which is interesting, but not everything that’s interesting is Science.

    (Maybe Stephen Colbert will bail us out by coining the word “Sciency.”

    1. @Ike Yes. It’s a strong marketing differentiator for Dan. And as a guy in the Differ business I have to give him that: he’s managed to make himself stand out from the social media pack with this science stuff. BUT if you make a strong claim like that, you’d better back it up. I think there *can* be real science done in this space. But is Dan doing more than giving his Sciencey stats “the Colbert bump”? Not convinced yet.

      You make an interesting distinction between the language of a stats-bolstered tip (best time to tweet) and that of a more scientific approach (really drilling into the evolving state of data over time).

      I’m thinking about this stuff because I’m helping one of my current clients – the Public Health Agency of Canada – work through the difference between the language of science (it’s all about finding relative certainty while respecting the variables) and that of social marketing (unequivocal statements meant to provide clear guidance / drive action). The power generation field surely has a parallel. You’d scare the public if you put all the risk-management caveats that your engineers live with daily on an electric bill. Instead simple stuff like “turn off your porch lights an hour earlier” is more likely to change behaviour.

    2. @Ike (Oh, and thanks for dropping in. I was hoping this would generate more discussion on the @DanZarrella vs @JasonFalls discussion. But I see people – including Jason have been sharing the Storify version more widely than the version on my humble blog. Le sigh.)

  2. Interesting debate, especially with your commentary. Each side has its points but I’m actually falling on Dan’s side with this one. There’s almost NO science that’s completely proven. Ask scientists. We accept a lot of scientific theory as fact and teach them in schools but MOST science isn’t much more than theory that happens to bear out consistently when we observe the results under certain conditions.But just because something behaves the way we want it to when the conditions are right doesn’t make it a law of the universe. Science is constantly in flux because of this. The more the scientific community gets behind that idea, the faster new discoveries take place – this wasn’t the case in the 50s. Now look at the world. Math, on the other hand, is a stiffer, firmer law of what is known, sure. So it’s a lot easier to impose the thinking of mathematical law to science, but that’s a cop out. Math and science processes can’t be argued the same way and I’m surprised Dan didn’t bring that up.In fact, math is even getting tricky though, on the molecular/ atomic level because of the ongoing experimentation in science.. One plus one equals two until science can make the exact same electron appear in two places at once? According to Science Daily, that happened back in 2005. And apparently the experiment was repeated so it wasn’t a fluke.

    1. @Tinu I hear what you’re saying, but I disagree.

      What Dan does is interesting, but it’s not yet science. What he does is the equivalent of measuring the fall of an object, then proclaiming the value of gravity (even if the object was a feather, and not representative enough for good conclusions.)

      Regression analysis actually would provide more useful interpretations of the data, and I firmly believe that’s the real value of these social networks. They will give away the front end for free, and charge a premium to a small list of clients who want to mine the data for truly scientific insights.

    2. @Tinu I get it and I agree that there is a spectrum of grey to science that Dan’s work could *conceivably* fit into. But I guess I’m just saying it’s one of the muddier shades. It’s just too easy to call yourself a scientist – or a guru / rockstar / expert / genius / messiah choose your label. So if you’re going to, you’d better be willing to hold yourself to a much higher standard of proof than the average data-crunching schmo. And more importantly have the humility that’s supposed to come with the scientific method (you know, nothing is ever proven – we just go with the best working model til proven wrong?).

Leave a Reply

Your email address will not be published. Required fields are marked *