Another set of month notes; here’s a link to those from May. Aside from having some interest to the 2-3 people who read them, I’ve found two good reasons to keep writing these:
Writing can improve mental health, and reflective writing can make you better at your job (if that’s what you care about). By the way, this is an article from The Conversation which is now honestly my favourite news source.
They’ve proved really useful simply as a record of activities, as I’ve needed to go back and work through notable events that I’ve spoken at or people I’ve spoken to in order to describe impact and complete KPIs.
As I mentioned last month, we’ve started our activity around making a case for ODI to continue to receive uncontested (ie not put out to tender) programmatic and core funding from DCMS. One of the big questions to answer is whether the UK government should continue to fund ODI in this way over the long term, or whether ODI should only receive funding through competitive bids like most everyone else.
It’s always difficult for me to separate my interest specifically in ODI thriving from my interest more generally to see the UK data ecosystem thriving, so it becomes impossible for me to tell whether any argument I put together is only persuasive to me because I’m already persuaded, and wouldn’t be to anyone with a more objective lens. But I’m obviously making the argument that government should continue to fund ODI.
ODI isn’t a public body (unlike, say, the Centre for Data Ethics and Innovation) but it was created by government nine years ago. There are a number of other organisations like this in the space, including in particular the Alan Turing Institute and the Digital Catapult. ODI, the Turing and the Digital Catapult are all Private Companies Limited By Guarantee which is a fairly common corporate form for non-profits. (Note: unlike those two other institutions, ODI has two people with significant control: Sir Nigel Shadbolt and Sir Tim Berners-Lee.)
Government sets up new organisations on an all-too-frequent basis. In part that’s because new organisations are great things for ministers to announce. In part, though, it’s because sometimes the challenges in a particular policy area speak to an institutional response. Independent institutions can do things the market isn’t doing or won’t do and that governments want to see done, but can’t do or directly shape themselves, like building networks; holding loss-making (but knowledge-disseminating) events; or creating, stewarding and publishing materials. They can be more risk-taking and responsive. They can be more trusted not to have ulterior political motivations behind their activities, nationally and internationally. I believe ODI has and continues to fill a role here specifically around data and data access, and particularly targeting improving business practices, in a way none of the other organisations in the space can do or does.
Even when an institutional approach to implementing government policy is a good idea, governments always want these organisations to be funded by someone other than government. Most such organisations are set targets to achieve a certain amount of match funding from other sources. At ODI (see accounts), of about £5.5m income in 2019, £2m was from InnovateUK (and wasn’t a core/unrestricted grant), £1.5 from other grants, and a bit over £2m from commercial work and membership/events income.
For interest and comparison, the Alan Turing Institute (see accounts), of £36m income in 2019/2020, only £2m came from a core grant from EPSRC: £13m came from universities, another £11m from other research council grants, just over £4m grants from philanthropic foundations, £3.5m from other grants, and a bit over £2m from commercial activities. For the Digital Catapult (see accounts), of about £18.5m income in 2019/2020, about £12.5m came from their core grant and the remainder was roughly equally split between £3m from other grants and £3m from commercial work.
I think (largely thanks to the entrepreneurial spirit that Gavin injected into the organisation during his time as CEO) we have done really well in finding ways to match fund the contribution we get from government, so that we’re not overly burdensome to the tax payer. That’s particularly so considering our scale and the fact that a lot of the government funding we’ve received, particularly in the last few years, has been restricted (which means we can only spend it in particular ways).
But we’ll see what DCMS and HMT think over the next few months…
I’ve spent a fair amount of time this month working with Mark Boyd to create materials in preparation and support for the World Health Organization’s Health Data Governance Summit. Mark is amazing to work with – full of energy and enthusiasm and really responsive to suggestions – so I’ve really enjoyed this project, but it has been just a little insane.
We were tasked with conducting research on three topics:
- the worldwide health data landscape in general (which of course isn’t complex at all)
- why health data is a public good (which is an easy thing to assert, but also has a fair degree of nuance around it)
- what good health data governance looks like (again, a nice simple task)
(Sorry, I should have added a /s to the first and last of those to indicate my sarcasm!) And this had to be done in the space of a few weeks, including not only desk research but also interviewing a range of WHO stakeholders so they felt their voices were heard, and the production of reusable visualisations.
I’m not at all complaining about the project, I’d just like to have had six months to do it rather than six weeks! Particularly considering the time constraints, I’m really proud of the fantastic job we (but especially Mark) did creating three sets of materials, which are public here:
I think Mark did particularly well teasing out some of the subtleties and challenges (or “conversations” as he framed it) about seeing health data as a public good. The latter two pre-reads are particularly relevant outside the health sector too.
This month I’ve chatted with:
- the outgoing and interim CEOs of Open North
- people looking at embedding open thinking into the new UK Health Security Agency
- Rt Hon John Whittingdale, Minister for Media and Data about what we’re up to at ODI, and the future ambition of the UK around data, plus various officials at DCMS as usual
- Volker Buscher at Arup, about governments plans (or lack of) to adopt more open data architectures
- Someone from Refinitiv about ESG data
- Cassie Robinson and a colleague at The National Lottery Community Fund, about their Growing Great Ideas Fund and whether it might be suitable for any of our work at ODI – this is a pretty unusual funding opportunity and particularly worth taking a look at if you are doing infrastructural kind of stuff
- someone from HMT, about ODI and the data landscape more generally
- Stephen Bonner, the new Simon McDougall at ICO
- Danil Mikhailov at data.org, and people from the Rockefeller Foundation about some potential work on data institutions for data/code flows underpinning federated machine learning
As I’ve outlined previously, at GPAI, within the Data Governance Working Group, we have defined the two projects we’re going to be progressing over the next nine months or so:
- Enabling data sharing for social benefit through data institutions which will look to do the fundamental work to support the creation of data institutions / data trusts (we’re being deliberately ambiguous at the moment) by GPAI.
- Advancing research and practice on data justice: preliminary guidance for AI developers, policy makers and communities affected by the use of AI which aims to help practitioners and users to include considerations of equity and justice in terms of access to, and visibility and representation in, data used in the development of AI/ML systems.
In preparation for some consultants joining us to do research on the first of these, Neil Lawrence and Jessica Montgomery hosted and ran a workshop on “What is a Data Trust?” to try to get us to a definition that we can at least use consistently within GPAI, if not in the rest of the world.
The results of that should be published soon, but I think it’s notable that one of the features identified was that data trusts “Provide independent stewardship through an expert trustee who has a strong fiduciary responsibility to act in the interests of the trust’s members.” (my emphasis). This is notable to me, because it retains the idea that data trusts (unlike other forms of data institution) have fiduciaries rather than more collective/democratic forms of decision making (eg as in a data cooperative or union), and that they act in the interests of those contributing data into the data trust, rather than, for example, broader society. This – in particular the importance of fiduciary or delegated decision making about data – is pretty much where we got to with our definition of data trusts a couple of years ago.
This month, Jared and Jack in the ODI team created a useful pseudo-venn-diagram of the different types of “bottom-up” or “people-centric” data institutions, ie those where individual people are directly sharing data with some kind of organisation that then stewards and shares it on to third parties. They also wrote about the different variants, which is worth having a look at if you want to understand the nuances here.
But the term “data trusts” still gets used in ways that don’t involve fiduciaries. This great recent piece in Nature about the need for collective decisions to be made about the use (for public benefit) of personal data generated within the private sector uses the term “public data trusts” to imply public stewardship, for public good, of data about the public, with collective or democratic decision making.
“Data trusts” isn’t the only piece of tricky/contested terminology in the data space at the moment. There are conflicting notions of what “data stewardship” means (in particular, is stewardship done by individuals or by organisations?). And about what people really mean when they say “data intermediaries” (my own view is that a data intermediary is any organisation sitting between original data collection and eventual data use; not all data intermediaries are data institutions, eg some are just brokers, and not all data institutions are data intermediaries, eg some are the originators of data).
At ODI, we invented the “data institutions” term in order to avoid distorting existing definitions of these other terms. But perhaps that’s just created another term to get confused about, I don’t know. Back when I was doing my PhD, a paper by Mildred Shaw and Brian Gaines was a cornerstone of my work; here’s the abstract:
One problem of eliciting knowledge from several experts is that experts may share only parts of their terminologies and conceptual systems. Experts may use the same term for different concepts, use different terms for the same concept, use the same term for the same concept, or use different terms and have different concepts. Moreover, clients who use an expert system have even less likelihood of sharing terms and concepts with the experts who produced it. This paper outlines a methodology for eliciting and recognizing such individual differences. It can be used to focus discussion between experts on those differences between them which require resolution, enabling them to classify them in terms of differing terminologies, levels of abstraction, disagreements, and so on. The methodology promotes the full exploration of the conceptual framework of a domain of expertise by encouraging experts to operate in a “brain-storming” mode as a group, using differing viewpoints to develop a rich framework. It reduces social pressures forcing an invalid consensus by providing objective analysis of separately elicited conceptual systems.
Unpacking how and why people are using the same term for different things, and different terms for the same thing, is actually a useful knowledge elicitation technique in itself. I wonder if we should try to do it more formally…
I wrote last month about the op-ed I wrote on the future of data protection regulation in the UK post-Brexit; that eventually got published in the FT and it was quite good that it was delayed because it meant we could add a reference to the TIGRR report which is at least explicit about wanting to get rid of GDPR. I got called out a bit (in private, and fairly) about implying that GDPR is a perfect gold standard, which of course it isn’t. But most people I’ve spoken to seem to agree with the premise that we should be aiming for a high-trust future rather than a race to the bottom.
I attended three advisory groups/boards this month:
- The GOV.UK Advisory Board, where I was very sad to learn of Jen Allum’s departure for new opportunities in the US.
- The Advisory Board for InnovateUK’s Next Generation Services Innovation Strategy Challenge Fund
- GDS’s Digital Identity Programme’s Privacy and Inclusion Forum – I can’t find a good link for more information about this, aside from GDS’s strategy, but think there should be more publicly available about this soon. This was just an initial meeting to outline the goals of the programme. It’s good that they’re recognising the need to engage around privacy and inclusion early on in the process.
- DCMS’s National Data Strategy Forum. This was the first meeting, and mostly focused on asking the forum members what we would like the forum to do.
I spoke at:
- The 7th Annual Meeting of the OECD Expert Group on Open Government Data, in a panel on “Open government data and public services: Facilitating people’s journey throughout the crisis”
- CogX, in a session called “What is Good Data?”. “Good” is such a loaded term, so I made the distinction between data that is good from the perspective of a data scientist (high quality, interesting, granular etc), data that is being used for good purposes (eg towards SDGs, including economic growth / innovation), and data that is good from an ethical / justice perspective, with participation of those that it is about and used by communities.
- A Technical Legal Workshop organised by Natalie Byrom at The Legal Education Foundation, on “Reforming the law around the use of automated and assisted decision making by public bodies”, intended to influence the Law Commission’s plans for future work around algorithmic decision making. Lilian Edwards, Rebecca Williams and Reuben Binns put together an amazing legal briefing for this event which I barely scratched the surface of and while I can wave my hands fairly effectively to lay audiences on the law around data, I was pretty intimidated by the legal minds in the (virtual) room.
- VivaTech 2021, which was a high production hybrid event, where I spoke about privacy and trust, and commented on the ambitions of GaiaX.
- The G20 Multi-stakeholder Forum on “Digital Transformation in Production for Sustainable Growth” where I spoke on a panel about “Enhancing data access for mSMEs”
I also attended a private roundtable organised by the Institute for Government on “How to implement a data strategy”, which really surfaced a lot of the challenges (without, perhaps, identifying potential approaches to tackling them). The main points I made were about the need for a National Data Strategy to be implemented across multiple bodies, and the implications that has for what a central(ish) team can do.
I was interviewed by the FT for a piece on data use in healthcare, which then wasn’t used.
I was also interviewed by Creative Commons for a position on their Board. This was a more rigorous selection process that I’ve been used to for other Boards (which is no bad thing!) and I’m yet to hear back about whether I’ve been successful. It’s an exciting time for Creative Commons as it approaches its 20th anniversary, so I hope I can be part of that.
This month is the first month where I’ve had Fridays as non-working days. This hasn’t exactly worked out as I thought it would because, of the four Fridays involved, two of them were ones where the kids were home from school so felt more like holidays than pseudo-working days. The others were taken up by a lot of life admin (on which a bit more below). But I did manage to work through my ideas about a book to write.
As I flagged last time, my first idea is a book vaguely titled “Data Strategy: A guide for organisational leaders in an interconnected world”, in which I want to really highlight how data strategy meets organisational strategy and how you can use data to change others’ behaviours as well as inform your own decisions as an organisation. Rough outline so far looks like:
- Part one: Understanding context
- Do you need a data strategy?
- What is data anyway?
- What do we do with data?
- How do technologies help?
- What do people think?
- What does the law say?
- How does data work in a wider ecosystem?
- What does your organisation think?
- Part two: Prioritising effort
- Strategic decision making
- Operational efficiency
- Customers and partners
- Part three: Implementing change
- Collaboration and communication
- Open data
- Portable data
- Sensitive data
- Equipping change
- Motivating change
But I also had another idea for a book with a working title of “Together data: Why our data is about us all, and what we should do about it”. Vague outline looks like:
- Part one: understanding rights and interests in data
- Personal data rights
- Intellectual property rights
- Public and state rights
- Indigenous data sovereignty
- Group and collective interests
- Data ownership
- Data commons
- Part two: foundations of consent and participation
- Participation levels
- Individual consent and its challenges
- Personal data stores
- Data trusts
- Public and patient participation
- Data collaboratives and commons
- Part three: implementing participatory data
- Understanding conflicting interests
- Embedding accountability
- Listening to people
- Learning data governance
- Making the case
To be honest, this is the one that I’m most passionate about. I’d love, personally, to make the case for rejecting the current narrative around data that paints individual informed consent as the best/only way of making decisions about what data gets collected/used/shared (which is as ridiculous as giving individuals primary responsibility to make decisions about when to wear masks, imho). I’d like to build on my recent blogs on community consent and on individual and collective rights in data, weave in the thinking being done by scholars like Jathan Sadowski, Salomé Viljoen & Meredith Whittaker (see their recent piece in Nature) as well as Linnet Taylor and others, and make it practical for policymakers and businesses.
Thoughts and suggestions still welcome.
I spent several hours on one of my Free Fridays working through what it will take to officially change the name of my youngest child, who is non-binary. Action on this was motivated by the fact that, as they were doing their mocks, they were listed in exam listings, and their places labelled, with their old name, which triggers their gender dysphoria and adds an extra level of stress on what’s already a stressful time (see more on this below, too).
Changing a child’s official name is a much more complex process than you’d think or hope, and hard to understand. The relevant forms are all here, which you have to complete and send in to be registered at the Royal Courts of Justice. The difficulty is in understanding what forms need to be completed by whom. There are three sets:
People with parental responsibility have to complete an affidavit of best interest to confirm that they’re OK with the child’s change of name. These need to then be sworn in front of a Commissioner for Oaths / Solicitor (which solicitors will do for £5 each).
People with parental responsibility have to complete the deed poll itself, and sign it in front of two witnesses. There don’t appear to be any constraints on who those witnesses are.
Someone who is (a) not a relative, (b) a British citizen, (c) a home owner, and (d) familiar with both parents and child needs to complete a statutory declaration for a deed poll of a minor and have it witnessed by a Commissioner for Oaths / Solicitor, alongside getting two exhibits – the deed poll and the child’s birth certificate – validated in some way.
Then all these forms have to be sent off to the Royal Courts of Justice, alongside a form about publishing information about the change in the Gazette, and about £40 to pay for their administration.
It is interesting to see both how complex the process is and how much it relies on people swearing oaths (in front of people who don’t know them) as a means of determining the truth.
Video games we’ve been playing:
I played Pikuniku which is a very simple little puzzle/exploration game; a few annoying boss fights that you will likely have to play repeatedly to get through, but a fun and whacky story.
I watched my eldest play The Wild At Heart which is an exploration/puzzle game I’d certainly recommend. It has an interesting mechanic where you have a swarm of “Spritelings” with different strengths/weaknesses that you grow and then basically throw at enemies and puzzles to interact with the world.
We’ve been playing a fair amount of Griftlands which is a deck-building / role-playing game. I really like how it uses cards to simulate verbal sparring as well as physical battles.
I played Rain on your Parade which is a silly puzzle game a bit like Donut County, where you play a cloud who can rain and later snow on people, rain different substances, do lightning zaps and move things around with hurricanes. There are a number of fun “themed” levels of which my favourite was leading a bunch of zombies around a city.
I played most of Minute of Islands which is a fairly linear exploration/puzzle game with a very warped/creepy feel which I enjoyed. I’m overall less enamoured of the gameplay though (overall just a bit too linear) and I gave up when I hit a particular movement/timing-based puzzle that I just couldn’t get through on multiple tries.
We started actually playing Good Society, and had a really good time just letting the characters interact and making up parts of the story. We’ve had a few weeks of downtime, while real life has intervened, during which we’ve been writing in-character letters, which I’ve really enjoyed.
TV we’ve been watching:
- Star Trek Discovery – now up to date.
- Lupin – watched Series 2.
- Sweet Tooth – the devastating impact of a pandemic on society was a little close for comfort, but recommended.
- Loki – a number of great things coming together here, in particular Owen Wilson, whom I would watch in anything, and a fantastic retro-futuristic set design; waiting to see how the story plays out
- The Irregulars – We enjoyed this series and the complexity of the characters and relationships.
- Law & Order – now just started series 12. So sad that Adam Schiff is gone.
Family film watching:
- 10 Things I Hate About You – teen romcom that got over the bar with the grandparents because it’s based on Taming of the Shrew.
- Raya and the last dragon – we found a bit formulaic (“an animated video game”)
- Harold and Maude – I enjoyed the dark comedy and 70s outfits!
June saw mocks for both my kids. My youngest found this particularly difficult because their neurodivergence means they find sitting in a room filled with small sounds from lots of other people really difficult to cope with. They explained this to me and I tweeted about it in a thread that seemed to resonate with other neurodivergent people (not just those with autism).
The mocks were also challenging because of their gender dysphoria. When going into exam rooms (and assemblies in general, I later learned), the kids in their school are asked to line up in two lines on each side of the corridor – girls on one side and boys on the other – before alternating in as they go into the room. This is supposed to prevent people who know each other from sitting next to each other and being inclined to chat, apparently. But which line should a non-binary student go in? What about trans students who aren’t yet out?
This hits hard because unlike when someone misgenders you (a micro-inequity you can choose to object to or shrug off), in this situation you are forced to self-identify as either male or female. It is like a 1984-inspired brainwashing technique where you have to say something that you don’t believe is true, repeatedly, day after day, creating cognitive dissonance and psychological stress.
Frankly, no wonder non-binary and trans students have particular mental health issues.
I wrote to the school about this, (and about a careers worksheet that asks whether men or women typically do particular careers) and they are sympathetic with the problem and the experience, but like other embedded practices that assume binary gender (cf PE classes), it will require some imagination on their part to change.