Originally published in the dissertation ‘A Vision of the Digissance’ in September of 2017, and once again on Wonk Bridge the same month
Interview, transcription and editing by Max Gorynski. Further editing by Yuji Develle.
Gavin Starks, a man who founds and raises companies like Dgen and the Open Data Institute as if it were no more of a challenge than a waiting kite on a windy day, is all about scale. Reposing in the man’s unassuming frame are vast notions of how best to design data architecture for society as grand in scope and as multifarious in potential application as the very real feats of physical architecture we see every day in our decreasingly real-seeming world. A man whose life and work betrays a suspicious likeness to the alter ego of Iron Man, Starks sees in the very data that is so incomprehensible to so many of us the building blocks of profound new institutions, of unifying possibilities in innovation and policy. His vision is one he shares generously. If the Internet’s a huge city, Starks made the blueprints.
Max Gorynski: When I look at the now user-centric net I see people with a book open in front of them in a language they don’t really know how to read. Open data, freely available statistics will be essential to ensuring people’s employment, their liberty in years to come; how do you see education changing to accommodate for people needing this new literacy?
Gavin Starks: I think it’s quite a nuanced question; I think we’ve got a lot of work to do on understanding how to talk about data in the first place, let alone how to interpret it in different social contexts. One of the things that we did at the ODI, and this was a central issue for me, [was] in trying to create common language. So we created the Data Spectrum, which talks about data being closed, shared or open, and that changed the conversation radically. It took about 9 months actually to get that right, because prior to that people were talking about big data, personal data and open data, as if they were somehow categories. What that lead to was a kind of mass confusion about data at every conference for quite a long time, and you still get this; if you put those words together, people get very concerned about privacy immediately. That tends to overshadow then the conversation on everything else. If you reframe it [as being] about licensing, I think it would really help to change the conversation; and this is something I’m very familiar with, as I’ve had a lot of experience in the copyright world, in music and video and so on, that copyright and licensing is not something that most people have an exposure to, other than through the notion of piracy.
Even business leaders tend not to consider licensing as a fundamental part of their business model unless they’re an existing, IP-based company, so there’s a lot of work to do on what the language is.
The second piece, I’d say, is how you go about examining data as infrastructure. That was a critical piece of language we created; if you think about roads as being part of our physical economy, and if you think about the internet as being part of our digital economy, then data becomes part of our knowledge economy. That’s part of our working hypothesis, there. What I this needs to lead to though, and this is the reason I got very obsessed about language, is that we need to have a good, public debate about the social contract between citizens and the state, between citizens and companies, and between companies and the state; because, at the moment it’s really not very well coordinated, and that’s another one of the thing we were trying to do at the ODI. There’s a vast amount of work to do on it.
The training and the education piece on it…we’re thinking about how we engage people in understanding of what they’re opting into. There has to be a broader conversation with people so that it’s clear what the conversation is even about.
“I think we’ve got a lot of work to do on understanding how to talk about data in the first place.”
M: Do you think a separation [of state from companies] is the way forward, or does the existing relationship just need to be managed better?
G: Well I think we can shorten the path [between customer and state]. The distance between a customer and a company is as short as it used to be, almost. You used to have direct trade…as our economy has scaled, and globalisation has happened, that conversation has become far more distant, and I think a lot of the reason for events like Brexit is that the ‘control’ the people want is actually about the distance they feel away from the society they’re living in; so good use of the things we’ve built would help shorten that path.
M: You’ve worked extensively with the Cabinet Office during your time with the ODI. How do you think digital law and regulation will develop in the coming years? Will regulation condense or will it be a less centralised proposition?
G: I think, again if you look at part of the language shift: inter-government departments now have to have a data infrastructure strategy, so again, getting the language right gets people to label things, which then helps people to say ‘This is something that you need to engage with.’
One example; I sit on the Ministry of Justice’s data board, which is an evidence board, and that reports to the secretary of the state. They’re very concerned with getting the data available to people in the system, whether that’s about prisons or about particular sentencing that has worked; so for example, if you wanted a judge to be able to pull up every other news case during a sentencing, to say not only what the particular sentence types for this type of crime and this type of profile of prisoner, but also what were the longitudinal impacts of those sentences. If you were going to put someone away for life, which is broadly £1 million’s worth spent, against £50k in rehabilitation: does the latter have a better long-term outcome for that individual? So I think we see it really systemically, and the same with DEFRA: their cabinet minister was really ambitious, to open up all the data in DEFRA that wasn’t under national security. This shift is already happening I think, and it will be interesting to see how it continues in the political climate.
At the same time, Barack Obama led a huge open data initiative, and Trump is shutting it down. So one of the questions that I’m working on right now, is how do you maintain persistent institutional knowledge in government?
“I think a lot of the reason for events like Brexit is that the ‘control’ the people want is actually about the distance they feel away from the society they’re living in…”
M: Carrying along in that political lane, can you see open data empowering small operators in a political sense? We’re starting to see it do with the start-up, could it do so with the activist?
G: I think there, we need to look to the web, and how that has evolved. There are multiple levels there: one is, yes many more people can engage, and you have crowdsourcing and crowdfunding etc. But you also have monopolies; and trying to stop monopoly behaviour is really hard, and one of the actions there that we need to drill into is how do we shorten the path between innovation and policy in the digital age?
M: Looking at it from a monopoly perspective, what do you think of the rise of corporations like Google or Facebook as politically influential entities?
G: That’s a very topical question. Are they wielding a political influence, or are the people paying for advertising wielding the political influence? I think one of the questions there, Tom Steinberg posted recently about the thing he’d like to see would be: Facebook, in the run up to this election, post the amounts of money they’re receiving for the platform.
M: So a simple question, then, with what I’d imagine will be a complex answer: how do we stop companies like Google and Facebook from exploiting feedback loops for profitable purposes?
G: Do you want to?
M: Are such monopolies not bad for the health of political discourse?
G: Those are two different questions: would you want Google to no longer be able to offer hyper-personalised advertising?
M: I suppose it depends on some sense as to whether it’s purely commercial advertising, or whether it’s content with a particular political edge: for example, the idea of Facebook algorithmically determining whether or not a user is to be exposed to more liberal or more conservative political content based upon their history on the site.
G: So are you targeting transparency from Google? Or are you targeting legislation? We have existing legislation to limit the amount of spend on political campaigns. So is it a category problem, or is it a technology problem? Ok, but why is it a technology problem? [Take Cambridge Analytica for instance; is] that Google’s problem? Or is that because the lobbyists are currently able to target people in a way that circumvents the rules? My challenge question to you here is, do you target the people with the money and the people with the political influence? Or do you target the technologies? If you were to go and publish a bunch of leaflets about hyper-targeting, put through your door, would you target the Post Office?
M: Indeed you wouldn’t!
G: Would you be targeting the printers? No, and it’s a good analogy although there are grey areas in there. The question is certainly a valid one, and I’m not saying that Google and Facebook don’t bear a responsibility, but I think we need to be very careful where the responsibility is, and where the liability lies. That’s been a perennial question going back to the early days of the web and of the internet service providers. Was BT responsible if people posted defamatory material on the web? We had exactly this; my direct experience of this was helping set up Virgin Net in the mid 90s, and around the time of Princess Diana’s death- Virgin was one of the first IPs to offer free web space to all its users, and so people could put up their own website. Some bright spark decided to put out a Princess Diana Driving Game.
“Trying to stop monopoly behaviour is really hard, and one of the actions there that we need to drill into is how do we shorten the path between innovation and policy in the digital age?”
It came out just before Christmas, and it was on Virgin. I got a panicked call from marketing saying the press has got hold of this, that we would be in trouble; I asked, “Have we broken any laws”. They said “No, but it’s offensive”. I said “Well, I’m not taking it down because it doesn’t break our rules: it might be in bad taste, but it’s not inciting or defamatory.” I then got a call from the chief executive saying “Take it down”; and I said “Well, you know, it’s not contravening our policies.” So we took it down for two weeks while the media had its day — it was on the Nine O’Clock News, BBC etc. Whose responsibility is that? If you’ve got the whole world publishing content, is it the responsibility of an ISP? Is it the responsibility of Google or Facebook to the police that? They have to monitor it, they have to comply with the law, but there needs to be a very clear separation of where the responsibilities lay.
I think ultimately, it’s as simple as ‘Who’s paying for it? Who’s contracted?” I think one of the positive things is we could get Facebook to have some AI tech, that looks to categorise things that are overtly political, saying ‘During an election, this is going to get flagged up as you will be held accountable by the Election Commission’. There are pretty strict rules; as soon as you spend more than £15,000 on an election campaign you have to report it to them. There is a policing system; but what the response is going to be to [web-based politics]? Man, that’s gonna be hard work.
M: If we move away from the more explicit questions of data for a minute, because I know you are a music man: do you see sustainable possibilities for funding the arts in the near future of digitisation? In the sense that, we once had the self-perpetuating industrial machinery in music, that was then thrown into disarray with the rise of file-sharing. After an interregnum, streaming became a proposition, but if you look at their numbers, an artist needs two million streams for a month of minimum wage [sic]; can this develop such that the whole becomes stable and sustainable again?
G: I think that new services will continue to push the envelope there, but what we’ve seen there is the promise of the long tale, and a reality of zero hours contracts. So that’s one way I would frame it. The other way is that the music industry thought it was going out of business for quite a long time, and they’re now having quite a good time again; and I quite often return back to the fact that you need an ecosystem to play, and one of the things in the transition from physical to digital is everyone said ‘Oh we can get rid of the record labels’. Yes you can, but typically you need someone to do some marketing, some legal work, the business affairs, invest the seed money; and that’s what labels do. If you think you can do everything yourself, great, but you’ll spend more time doing the admin than actually being a performer. The dynamic shifts but the actors actually stay the same; you still have record labels, looking at the independent record labels over the major record labels, it’s kind of two music industries. The independents, a lot of them went out of business; but those that adapted, they’re doing really well.
The interesting thing about that is you still end up with labels, distributors etc aiding the system for successful acts, and one of the things that they do is seed funding and managing a number of artists. Think of it like spread bets: they’ll take ten artists and one will succeed, and the successful one will effectively cash flow the others. So, actually, part of the function of the label is to make a monetised risk; and while a lot of artists then moan about contractual terms, it’s running a business, and if you’re in the music business, you’re a product. And that dynamic gets shifted a bit, and it doesn’t go away; an analogy I often draw is with the open source world, where we’ve seen a transition from IP and proprietary software to open source, and also from product to service. I question people why they don’t just transition from IP-based models to service models, and we have seen that to a large degree with everything from software as a service to cars as a service: AirBnB through to taxis, nowadays, everything’s making that transition. So business models, there are sustainable ones there, but again you have this slightly skewed spectrum that, because there’s so much money at stake, the investment community will pick a winner: ‘Ok, we’re going to put billions of pounds behind Uber.’
So again, technology is not the problem, it never is; technology is only about 10% of what’s going on. The rest of it is how our economy works, all the other things.