#OpenDataSavesLives - reflection and moving forward
When we launched #OpenDataSavesLives at the start of 2020, it was part of our open strategy for the coming years, using events and project work to help meet the needs of our sponsors and innovate openly for the benefit of everyone. We never anticipated just how important it would become.
We knew that we would be supported by like-minded allies in the data world such as health informatics specialists Beautiful Information and our partners at Open Data Manchester. However, as the coronavirus pandemic unfolded in the UK, it became apparent that a whole range of people were desperate for data and for answers. So we set up a coronavirus-focused series of online meetings, open to anyone who wanted to share/find data, share work or ask for help, etc. Taking part in #OpenDataSavesLives is easy - join us at one of the weekly sessions (register via Eventbrite), add your details or your resources to the open Google Doc, offer to help someone with their project, or share some data sources on the data page.
#OpenDataSavesLives was derived from #datasaveslives, the hashtag of the Farr Institute, part of University College London and a belief among us that, while we need to be always mindful of information governance, we should be bold and work on the assumption that if you are bringing data together with the aim of making the world a better place then you should have the freedom to do that. Put in place the right ethics approval but work on the assumption that you can help solve a problem by knowing more about it.
We started by sharing what we were doing and this rapidly evolved into a platform for others to share. By bringing people together, connections are made that can advance a project or speed up the release of data. It sounds stunningly simple - because it is. Hosting these online sessions for 6 weeks has revealed many things, not least that people do want to help others when it comes to data projects, they just don't know where to start or who to talk to.
Key moments so far
- British Red Cross spoke about their vulnerability index (available openly via GitHub)
- Ordnance Survey shared more details about the support and resources they are offering to map the coronavirus situation (also mentioned that UPRNs will be published as open data in June, though earlier would be better!)
- Health Foundation shared their advice for starting collaboration during a time of crisis
- NHS Digital data release in the early weeks, guided by user needs shared in the sessions
- NHS Digital also shared their first wireframes and thoughts about the new NHS Open Data Hub in development
- Robust set of use-cases developed that can help inform many other publishers
- Local authorities also open to use-cases so they can publish data that is needed
- Collective demand for ways to move at pace and fix real problems in the Open
The coronavirus situation has changed rapidly over the last 6 weeks, so people's demands and expectations of data have also changed. Despite updates being shared on a regular basis through the #OpenDataSavesLives sessions, there are still some dominant themes & challenges that are preventing progress, and in some cases, preventing a joined-up and efficient local response to the pandemic.
Centralised data, Centralised Thinking & Centralised Institutions
A lot of organisations (such as local authorities and local health bodies) have described their experiences of sharing data to central government but not getting anything back. For example, if something is working in another part of the country and has supporting data, no one gets to know about it because it's not being shared back from central government. This is damaging in two ways - it slows down innovation, and it means there's a higher chance of repetitive work and wasted time/effort.
Data disappears to the centre never to be seen or used again. Unstructured reports are created many times over and the outputs are never shared back. People can't make comparisons, and the opportunities to collaborate, share, and innovate are lost.
We know the solution, and it is straightforward. Analysts can create common datasets with common structures and features, which can be built into data pipelines that the required reports are created from. When shared, everyone gets the comparison data back. These data assets also become the way in which data is shared locally, regionally, and nationally. Done in the open with feedback loops, there is no reason that 'at pace' responses from distributed teams cannot be built.
This does mean that our institutions and organisations need to change and use the web as it was intended to be used - 'many parts loosely joined.' The instinct to control and remove permission to do innovation and share our outputs (positive and negative) has been shown to be exactly what we dont need at the time of national emergency.
Data Hide & Seek
Lots of examples of organisations either not sharing the data at all or making it more difficult to find/use the data they are publishing, and when asked the reason for not sharing there is not a reason at all. The Health Foundation blog summarised below is a great exploration of this.
The comparison with other European countries is particularly hard, as the data that has been published by their institutions has been earlier, more detailed, and more easily accessed than data in the UK. Tom Forth, Head of Data at ODI Leeds, uses examples from France below, whilst Italy and the Republic of Ireland also demonstrate a better approach to data during the pandemic.
Fantastic live data on the economic impact of Coronavirus in France. By region, by sector, up to date, rapidly updated, with graphs, maps, and raw data to download. https://dares.travail-emploi.gouv.fr/...
Specific examples we have are:
- organisations needing to fix geography so that they can speak effectively to other organisations that use different geographies
- using volunteer-led data (seen in our visualisation and many others) because 'official' sources are not reliable, hard to find, etc
- symptom trackers, such as ZOE, which were adopted early in the pandemic (and so have a wealth of data that could help plan and prepare for ways out of the crisis) had no open data functionality built in, even though University College London and National Institute for Health Research were partners
In a recent blog-post, we wrote about the new PHE dashboard, where changes were made to how the data was served to their own dashboard, but broke everything for everyone else who were previously relying on the fixed URL CSV file. The problem was fixed swiftly but demonstrated a failure of design and not considering the needs of all the users - the data is the product, not the dashboard. The dashboard is just one way of visualising it for some people. Other people want the data.
With that in mind, and building on what we already covered in the blog, there are some simple things that people and organisations can learn from this pandemic:
- Use unique identifiers for things
- Don't just be a consumer of data; add back data for others
- Publish on the web using URLs
- Provide in common formats (+ a variety if you can)
- Have persistent URLs for "latest" datasets
- Provide metadata/documentation
- Provide a feedback method so people can help improve
- Geographic IDs (UK/regional/local) and polygons should all be made available as open data, which would allow people to match across datasets
Working together and sharing your work can be hard at the best of times, let alone during a crisis such as the coronavirus pandemic. There are various obstacles to overcome and it can feel daunting if you have never worked in 'the open' before. The Health Foundation published a blog post about collaboration in a time of crisis which outlines the simple ways that anyone can start to share their work. By sharing, you are increasing the abilities and capabilities of others. A good example of this is the proposed data standard that Beautiful Information developed for patient-level data. It is already being used across Kent & Medway CCG, so if other CCG's adopted the same standards, there would be tools and analysis available right away.
What happens next?
The popularity and success of the weekly #OpenDataSavesLives sessions shows that more people are 1) seeing the value of open collaboration, and 2) recognising that working in an open way - publishing work on the web, making it searchable, allowing others to contribute when they can - speeds up innovation for everyone. We want to keep facilitating this work, aiming to keep it positive and always moving. Our vision of a post-pandemic world is one where the collaboration continues. Going 'back to normal' is going backwards, so let's think about going forward instead.
At ODI Leeds, we see two things that we can focus on and help:
- the creation of a 'repository' of tools, data, and resources that is openly available and encourages further innovation
- and working with others to explore cultural change and working more openly.
We've already made a start on the repository through our #RadicallyOpen approach - everything we do is published online, with supporting documentation, blogs, technical help, links to data, etc, as demonstrated by the #OpenDataSavesLives page. Cultural change might take a little longer but we are not alone as advocates, and we welcome the chance to work with others to promote open data and open innovation to organisations of all sizes.
Taking part in #OpenDataSavesLives is easy - join us at one of the weekly sessions (register via Eventbrite), add your details or your resources to the open Google Doc, offer to help someone with their project, or share some data sources to the data page.
We know that being #RadicallyOpen will allow us to work together, build better services, and face the future and that #OpenDataSavesLives.