Tuesday 17 December 2013

The flawless map

Beyoncé recently launched a new album, unannounced. Simon Rogers (formerly of The Guardian and now at Twitter) capitalised on this and created a map using CartoDB and the new Torque. And here it is:


I'm a big fan of CartoDB and also the new Torque engine for creating animated maps with simple, intuitive UI design. But what of the map? The map is a great example of rapidly collating a dataset and publishing it.

I was going to ignore it as blog fodder because it's just one of those transient map objects that we tend to see daily on the interwebs. You look at it, then move on pretty rapidly...but then a TIME magazine article entitled "This Map of Beyoncé-Related Tweets in Real Time After the Album Dropped Is Flawless" appeared. Obvious references aside, the other adjectives used to describe it are..."awesome", "amazing". Now we have blog fodder.

Flawless?  FLAWLESS? I get the reference but this is the hugely respected TIME magazine. It's nearly 100 years old, has the largest circulation of any weekly news publication and a readership of 25 million. The article was written by Laura Stampler, a graduate of Stanford who has held positions at Business insider and the Huffington Post. Now a reporter at TIME she may be a good news hound, she may even be a fan of Beyoncé but based on her article she's got no idea about mapping. She demonstrates herself as just another consumer of the interweb who tacitly accepts its wares. She's not alone by the way...it's a plague.

The map is nothing more than day-glo pinkish-purple splodges of geolocated tweets. Tweets that I've noted before are possibly the worst metric of any modern scrapable dataset. They account for such a biased population they simply cannot be taken as representative of anything. There's so much error and uncertainty which the map implicitly portrays. Sure, it's eye candy but nothing more. It doesn't even really show the true explosion of Beyoncé fandom...it only shows the explosion of geolocated tweets that mention her album as people wake up across the globe...or maybe just mention her name since it's a self-titled album. Can we really gauge the geographical pattern of fandom? Do we know whether people liked or disliked the album? And what about the much larger proportion of people without smartphones, in countries with little network coverage and who don't use Twitter. What of those who use Twitter but turn off geolocation. The map fails to answer any sort of question. Even at it's basic level, the question of  'where'
(in this case...where do people tweet about Beyoncé), the map falls over because it's such a partial dataset.


Ms Stampler offers some in-depth interpretation...the US east coast is consumed by bright purple light whereas Russia is underwhelmed. OK...Web Mercator makes Russia appear much larger, having the effect of dissipating what tweets may exist and comparing the densely populated, spatially proximal cities on the east coast of the US to the vast barren expanses of Russia isn't really relevant. In order to make any sensible interpretation of the extent to which, say, Muscovites and New Yorkers like the new album we need to know the proportion of the population that (a) use Twitter and (b) turn on geolocated tweets. And China? Could people there even download her album?

So it's not a flawless map. It's a map.It has flaws, just like pretty much every other map ever made.  It's not awesome and it's not amazing either. It actually has some pretty major flaws because the data is so weak but then again, these days the ability to recognise, appreciate and acknowledge flaws in online maps seems to be an irrelevance for most people. They're happy in their ignorance consuming this sort of work and not caring.

Like I said...nice eye candy and we move on...but pur-leeaaase let's not elevate such work to being 'flawless' when it isn't. If this is flawless then I'll get my coat. taxi for Field.



8 comments:

  1. Yo girl, you don't know Bey enough to know that FLAWLESS is a reference? http://www.youtube.com/watch?v=Tc8hyBfoBZg

    ReplyDelete
  2. I get the reference, the music may be (though you're right, not my taste)...but the map isn't flawless

    ReplyDelete
  3. I can't agree more. The flowery visualization cannot hide the serious data bias.

    ReplyDelete
  4. I agree with a map not being flawless as like you said no map can be flawless. But do we agree this map show the effect of Beyonce on Twitter? No idea if it is positive or not, no idea if it is representative of the entire world population or not, but...it does show that the announcement from Beyonce made a ton of impact in coverage about Beyonce in Twitter.

    You might argue this is irrelevant, but from a pure campaign value, is clear that the announcement open a lot more of discussion about Beyonce on Twitter, and specially in some particular areas in the world at particular times.

    I understand that lot of these days there is a lot of confusion of what is the actual value of a map of Tweets, but for all twitter maps that I have seen out there I believe this is one of the few, that actually tells you a story.

    ReplyDelete
  5. Thanks for commenting Javier. Yes, the map shows the album caused her name to 'trend but did we need a map for this given the geography is one of the weakest elements of Twitter data? A simple look at what's trending would reveal her surprise announcement got the right publicity. That's surely the only story we can glean from the data?

    ReplyDelete
  6. I wonder how it looks with twitter usage by country/area/region - surely twitter is less popular in e.g. Russia compare to the US. Is it even allowed in China? I wonder would it look like compared to a map of a) overall twitter use in the same period, and b) as a ratio of tweets in a country/area/region. From looking at the map, I am curious about the Turkey hotspot, and if it really didn't get more attention in some European countries (e.g. Germany, France, Spain).

    ReplyDelete
  7. Well, the visualization shows you the tweets over time before announcement too, so that you can compare how much people were talking about Beyonce before too. It is on the change that you can see the impact.

    I agree with you Kenneth, we didnt even need a Twitter trend at all, we all know that Beyonce is great and whatever she does will be popular, we all agree on that right? But for whatever reason there is always someone that needs to see it instead of believing. I dont know why ;)

    ReplyDelete
  8. This is a visualisation that belongs in the "We did this because it's technically possible" folder...it tells us very little, but looks nice (ish)

    ReplyDelete