Wednesday, October 26, 2011


I had hoped to flex my newly awakened blogging skills with something light.

Instead we have the sight of police chucking tear gas into protestors at #occupyoakland.

That is, of course, the hashtag used to follow the news with twitter (it seems that more traditional media sources, acting on police advice, went home early)

Oh, that sounds like a conspiracy. Well, I suppose it is, and I'm as much of a sucker for the outlandish theory as the next man.

I also possess a certain forensic streak as well and I became a little puzzled with another 'conspiracy'; that none of the 'occupy' tags were trending. This has been discussed for a while back and forth, with no clear conclusion being reached.

I went looking.

First off, I tried simply counting the number of tweets associated with #ows and the death of Steve Jobs (#ThanksSteve was very prominent at the time) They seemed to be appearing at comparable rates. However, Twitter's trending algorithm isn't just a question of total tweets; it tries to show people what's hot and what isn't. So, there may be some reason why ows wasn't trending? Possibly it had saturated.

It turns out that there a number of sites that plot trends for a given set of tweets over time. Most are geared to individuals trying to find out how they're doing.

There is one, however, allows you to select hashtags. That is Trendistic. Using it to compare #ows with #ThanksSteve shows that the latter was a clear spike, while 'ows' was an ongoing grumble (with spikes). OK, so maybe 'ows' doesn't trend, but it's persistent.

Then came news of tear gas in downtown Oakland...

Again, no 'OccupyOakland' appeared in the trending list. However, what *was* appearing was 'Oakland PD' and 'OPD' (Oakland Police Department?)

What is the Oakland PD doing that doesn't involve OccupyOakland?

Going back to Trendistic now revealed something *very* interesting.

Here's a plot of that trending 'Occupy PD':

Occupy PD Trending... at 0.18%.

There's a respectable spike there. But have a look at the column on the right. Do you see 'OccupyOakland' displayed at the top? I do (I also see ows, shown less prominently)

So, here's the trend for 'OccupyOakland' as the dawn breaks over the Bay Area.

OccupyOakland Not Trending... at 0.81%.

Whoa! That is a spike that is over 5 times the size as 'Oakland PD'

This seems like a case of 'same data, different conclusions'.

Twitter has been a mainstay of a lot of popular movements, and it seems churlish, even ungrateful, to point this out.

But, dear @Twitter, I think you need to respond to this, and either fix your algorithms, or list the tags you are 'demoting' as well as 'promoting'.

Update: Twitter did recently point to a fairly detailed account of how tags trend here. As I said earlier, this may account for #ows, but #OccupyOakland fits the profile, and Trendistic lists it!



Post a Comment

Subscribe to Post Comments [Atom]

<< Home