Here’s a link to my June month notes. I’m just going to focus in on a few highlights this month rather than listing everything religiously, in an attempt to not take so long over them!
A highlight of July has been finally getting our Experimentalism project off the ground. This arose out of conversations that Milly and I had about the challenges of policymaking around data when the data ecosystem and economy is so complex, new technologies are continuously being invented, and there’s little scalable or replicable evidence about what works and what doesn’t because outcomes are so context dependent.
This kind of complex environment can make policymakers freeze up, feeling that they need more evidence before they can act. But I’d argue that it instead speaks to a need to experiment, monitor, iterate, accept that mistakes will be made but correct for them rapidly. So we wanted to explore with data policymakers and funders where they have been experimenting, and what lessons they’ve found as they do so. We want to encourage them to be more intentional about their experimentation, which includes being cautious: perhaps not experimenting at all in particular areas, starting small, and being prepared to change direction.
(Over the last couple of days I’ve started reading Emergent Strategy, which isn’t a book for everyone, but highlights themes of adaptation, and nonlinear / iterative change. The language of “experimentalism” speaks to hard science, data, and evidence; “emergent strategy” is a much more organic way of seeing the same process. One thing I like in “emergent strategy” is its focus on intentional adaptation – not just adapting to circumstances but towards an intended goal: I think it’s important to recognise that we bring our values to our experimentation, and choose what good looks like.)
Milly and Adjoa have put together an ambitious programme of roundtables across three groups: UK establishment; under-represented or marginalised communities in the UK, North America, and Europe; and with stakeholders working around international development and the Global South. During July, we held discussions with friends in both the latter two groups to work out how to pitch and frame the roundtables with them most effectively. We also had our first actual roundtable, with the UK establishment. We’re aiming to hold more, for each of the three groups, all the way through to March next year. There are still opportunities to get involved if you’re interested.
I spoke at an excellent event organised by Natalie Byrom at the Legal Education Foundation to launch new research by Judith Townend on justice data in Australia, Canada and Ireland. I particularly enjoyed the general skewering of overinflated claims about the ability of AI to predict the outcomes of cases during the event. But it was also interesting to me to draw parallels between how worried judges are about data being used to performance manage them, and similar concerns that doctors have. I think it’s worth looking at the approaches taken with prescribing data, in particular within OpenPrescribing, to highlight how careful interpretation is needed of the figures by third parties, and to actually use it to proactively support GPs to reflect and alter prescribing behaviour (towards cheaper generic drugs where applicable).
I had an interesting discussion with Jack about the relative merits (when developing data institutions) of just getting on with building something and then iterating vs. spending a lot of time understanding context, stakeholders and so on before doing so. There’s obviously a lot to be said for a JFDI attitude and approach, but one thing to bear in mind is that the distinguishing feature of a data institution is that it is there to find a balance between a community that includes multiple stakeholder interests – of data providers, users, those affected by the collection and use of the data, and those trying to create system-level change through the data institution. We found in our research on trustworthy data institutions that new stakeholders (of any of these types), or changing needs from existing stakeholders, can change the dynamic within that community and prompt re-evaluation from all sides (potentially leading stakeholders – such as data providers – to drop out). So iteration can be quite costly for a data institution, which I think shifts the balance towards a more careful, deliberative and inclusive approach from the start.
I tweeted about reading A Skeptical View of Information Fiduciaries by David Pozen and Lina Khan (brilliantly, current chair of the US Federal Trade Commission). It’s definitely a good read, but the thing it picks apart is the idea that organisations like Facebook or Google could themselves be information fiduciaries; a concept that is pretty easily dismissed because of their conflicts of interest, between delivering shareholder profits and user benefits. Most of the work that I’ve seen about information fiduciaries (aka data trusts) has suggested that this role is taken by third parties, which wouldn’t (or might not) have these same conflicts of interest. But, as the paper points out, there are still some features of the information fiduciary relationship that make it challenging, and different from traditional fiduciaries such as doctors or lawyers. In particular, it seems unlikely to me that information fiduciaries will be people you have conversations with – or “develop thick relationships with”. They’re much more likely to be digital services that you interact with, and that need to make money from those interactions somehow, which actually doesn’t make them all that different from the organisations they’re being set up to protect us from.
I also tweeted about an idea that’s been buzzing in my head for a while now, about how to bring people who aren’t involved in citizen’s juries themselves accept the decisions they make about things like acceptable secondary uses of health data, or the ethics of uses of location data. I really like citizen juries as a way of getting at what informed public opinion, but there’s a whole process involved in getting participants to being informed. So people who haven’t participated might understandably feel that their views haven’t been taken into account. I tweeted a thought that a reality-TV-show-style mass broadcast of the process might be an effective way of bringing the wider public on that same journey. There was some good discussion about it, and some legitimate concerns about the idea, in particular about how it might discourage people who don’t want to be televised to being involved, or change what they share, and thus skew the result.
I’ve been playing a lot of Cozy Grove which is an Animal-Crossing-like game with a lot of chill grinding. At some point I will wonder why I’m spending so much time on it, but I like the art style and there’s always a sense of incremental achievement when I play it.
I’ve also just bought and have been playing Ring Fit Adventure, which is a very basic RPG that you play by exercising. I suppose a mark of its success is that for the last 5-6 days I’ve got up early to play it (exercise!) before getting down to work. If you’re a game-playing nerd who wants to exercise but doesn’t like gyms or enthusiastic work-out videos, it might be for you.
The one TV show I’ll recommend this month is GLOW. I admit I wasn’t particularly drawn in by the premise (it’s centred on a women’s wrestling show in the 80s), but I do highly recommend it for the humour, drama and characterisation.
We hit a few teenage milestones this month. The ones I’ll share are: not knowing whether either kid would be home for dinner; both of them staying up after me & my partner went to bed (so they could watch a Dream SMP stream together); and my 17yo making the trip to see her grandparents in Cambridge on her own.
Finally, and talking of the Dream SMP, if you share my sense of the ridiculous you might like this story about how the flag from L’Manburg, a fictional country in the Dream SMP, ended up being waved at the recent anti-vax rally in London.