“Digital Agenda” has been used in such creative ways by our political parties that I was not surprised when my friend Marco asked this old data juggler (I did data-driven web usability assessment in the ’90s) a hand. His idea was to monitor the political platforms for the general elections that will be held next weekend (that would be 24-25 february 2013, if you’re reading this in the future), and you can download the resulting “Observatory on digital policies for the 2013 Italian general elections” (in Italian). Alessandro Garbagnati joined in too (and was invaluable).
Can three lone geeks (please, no “Three Am1g0s” jokes) take on the propaganda behemoth and come out alive? Yes they can.
Well, I’m no political analyst. I am also painfully aware of the supreme rhetorical levels politicians can achieve at campaign time: political language is crafted to mean absolutely everything. But I love a challenge, I love to see through hype and I burned to do some network analysis on the worst possible data. So here’s a part of the story of how the Observatory came to be: the funnier one, and with graphs.
The first problem with data analysis is data preparation. The first problem of data preparation is telling data from hype, wheat from chaff. Or signal from noise, as Nate Silver (gotta love Nate Silver). When it comes to political platforms, there’s an additional problem: you can’t. There is no way you can tell hype from data. People way above your pay grade have sweated every comma so that everything and its exact opposite can be inferred from the text.
Comparing apples and oranges
Choose what is better:
- a nation-wide program to teach digital technologies to school teachers
- economical incentives to businesses who donate pre-obsolescent PCs to schools in need
- bringing Internet into every classroom.
Avoiding statistical pitfalls
Enter Network Analysis
Beware: this is little more than an entry-level exercise, very much like shooting flies with a shotgun. Overkill, yes, but it gives us the advantage of catching one or two flies. All graphs were produced with Gephi, the open-source interactive network visualization and exploration platform.
1: What Areas Are Really important?
In the chart, areas are ordered clockwise by increasing number of platforms they appear in (their degree). To reduce clutter, only those arcs representing co-occurrence of the areas in 3 or more platforms are represented. Note that three areas appear together in almost all platforms: Digital Literacy, Tax Incentives, Broadband Investments (from top, counterclockwise). Let’s remove some more clutter.
Now, an arc appears only if it stands for two areas appearing together in 4 or more platforms. That is, there are at least four parties whose platform prescribes actions in both areas. With less clutter, a few things become apparent:
- there is a large consensus that Digital Literacy, Tax Incentives, Broadband Investments should be pursued together
- when it comes to reducing the digital divide, consensus is already diminishing
- bringing digital technologies to the Public Administration (PA) and Incentives to Startups already belong in open-triangle-land: Incentives to Startups is strongly linked to Reducing the Digital Divide; this in turn is strongly linked to other areas, none of which is strongly connected to Incentives to Startup; here is where consensus becomes brittle
- all other areas (from Digital Citizenship at 6:00 counterclockwise) appear way less often; these will likely be the Cinderellas of actual government action.
It is instructive to see what area are “Cinderellas”; from 6:00 counterclockwise we meet
- Digital Citizenship
- Open Source for the Public Administration
- Smart City
- Open Government
- Open Data.
You may probably say that these areas are crucial to any self-respecting digital agenda, that all of them should appear in a party’s platform, and you would be right. Here we have some indication that when it comes to digital, there is no strategic vision.
We can also note that as we walk the nodes counterclockwise, the keyword becomes less and less fashionable and more and more “technical”. We can safely assume that most professionals and entrepreneurs in the digital realm would deem the right side of the chart at least as important to the economy as the left side. Parties seem to think otherwise. Fashionable keywords get all the attention. This is another indication that platforms have been designed more to impress the public than to pursue strategic goals. Also, plain ignorance of “technical” issues cannot be ignored.
What does all this means in terms of potential policy? Simply that, whoever wins the elections, will find it easier to pursue thickly-connected areas than thinly-connected ones. That is, for instance, to pass laws on tax incentives to promote digital literacy than on Open data to digitalise the Public Administration.
2: Who Agrees With Whom?
You’ll like this: in the following graphs nodes are now parties, and an arc represents that two parties propose actions in a same area. The more areas both parties want to act in, the thicker and whiter the arc. Look what happens…
Here, nodes are political parties. An arc connects two parties if their digital agendas share at least two common areas. As you can see, everybody agrees with everybody else, to a greater or lesser extent. Consider that the entire political spectrum (though not all parties) is represented. Yes, there are stronger bonds among the parties on the left side of the chart. The interesting thing is, half of them are from the political left, the other half from the political right. Let’s remove some clutter and see the really strong bonds.
Now it’s better. Platforms from Lega Nord, SEL and Rivoluzione Civile have less than 4 areas in common with anyone. This is no surprise as their platforms cover 2, 2 and 3 areas, respectively. When we get to the stronger commonalities (the thicker arcs in the upper left quadrant), things become interesting:
- Monti (political right) shares more common areas with PD (political left )than with PDL (political right)
- PD (political left) and PDL (political right) have more common areas between themselves than with any other party (excluding Monti)
- the M5S (Grillo, political left) platform shares more areas with Monti (political right) than with any other party
- FARE (political right) has as many common areas with Monti (political right) as with PD (political left, and more with these than with PDL).
- which platforms have an ample digital agenda, and which only pay lip service
- which joint areas are more likely to attract a majority of votes, whoever wins the elections
- which potential alliances may be formed on digital issues, regardless of election results.
All this, of course, requires that political parties stick to their pre-election platforms, or are so required.