So, this evening I went to see the marvellous and challenging film Ex Machina, which is about artificial intelligence, power and abuse – I strongly recommend it.
While I was watching the film, I experienced my first bit of controversy with the Party Funding database. Someone listed as a donor on the site claimed that our records were incorrect and demanded a correction:
A little background: Rupert Read is the Green Party Prospective Parliamentary Candidate for Cambridge City. He’s recently been embroiled in a bit of controversy over some comments made about transgender people, which he has subsequently now apologised for. One blogger, Zoe Stavri, used this website to point out that he was a major donor to the Green Party, which is how, I presume, Rupert Read came across this site (full disclosure: Zoe is a personal friend of mine and I’m grateful to her for helping to promote the site; I also know Rupert, largely though my previous work at the New Politics Network and Unlock Democracy).
It was then suggested that there are other discrepancies on the site, involving other Green Party donors. Specifically:
This is potentially a big problem: if the website is showing erroneous data, that is a very big deal. Fortunately, I have checked with the Electoral Commission database, and it would appear that this is not the case. I can’t link directly to the Commission’s database (nor can it be used on mobile devices!), but the following are screenshots showing what the Electoral Commission themselves have down for Rupert Read and Noel Gillian Kirkby:
(Note: there are two screenshots for Noel Gillian Kirkby and Noel Gillian Kirby – I’ve merged these two on the Party Funding database because they are presumably the same person and no-one has yet challenged this assertion).
As I hope you can see, the data on both the Commission database and my database are identical. This does not mean of course that either Rupert Read or Noel Gillian Kirkby did indeed make those donations, merely that that is what the Electoral Commission’s register says. I make no claims as to the accuracy of this data; indeed, the whole point of this project is to subject it to greater scrutiny than I believe it currently enjoys.
It could be that the Electoral Commission has made a mistake. My understanding however is that since the mid-2000s, the political parties enter the data directly themselves via the Electoral Commission’s own e-filing system. On the face of it therefore, the data came from the Green Party directly. I suggest to Rupert Read that he begins his investigation into how he came to be listed at the Green’s ninth biggest donor of all time (and third biggest in 2014) with his own regional party.
Lest you think I’m complacent about this little episode, I can assure you that I’m not. This experience has thrown up a whole number of issues which I need to contemplate.
The first is that while the data presented here is to the best of my knowledge accurate, it isn’t as transparent as it could be. I recognise now that I should have created pages for each individual donation containing all the data provided so that they can be properly scrutinised directly. I’m going to include that in the next build (which is coming along very nicely, thank you). The second is that aside from emailing me, I haven’t developed any kind of automated mechanism to allow people to flag up what they believe to be questionable data. I have no intention of changing any data here that hasn’t been corrected by the Electoral Commission itself, but I think it is reasonable to allow people to flag concerns on the website if they feel they have been misrepresented. I’m already planning a commenting system, but I’m now considering whether a “red flag” system is needed alongside that.
The third issue is more troubling. Because this data is taken from the Commission website at a certain point in time, I don’t have a way of correcting old records if they are subsequently changed. I’m all set up to add new quarterly data as it is published, but there is no way I can correct older data, other than by uploading the entire dataset periodically. Currently, this is the only solution I have, but it is a time consuming one. Fortunately, I’ve designed the system so that if I do reload the original Commission data periodically, it won’t stop the rest of the website working (in principle, at least). But I’d rather be able to simply update any changes on an ad hoc basis.
To summarise: I’m pleased to say that the database has passed its first real test, but even then I’ve learned a lot from this experience and it has caused me to rethink how it should work. It’s positive then, but it involves more work and might delay the release of 0.2 by a little.