I tweeted on Sunday about how I had stopped “thinking in the open”, I realise now that doing this is part of what I do. I worked out whilst in Defra that by thinking out loud ideas and concepts can be tested and become more robust.

Reading Clare’s speech over the weekend reminded me that I used to be one of those tempered radicals and the good stuff we did whilst delivering the Defra Data challenge.

Lot’s (relatively) of people ‘liking’ my tweet reassured me that they agreed (it is amazing how we interpret ‘likes’ as we choose).

So, I have said it out loud and therefore I must start doing this again…

(Data) Trust

A few weeks ago I was asked to go to a meeting to talk about “data trusts”, an area of the data field that is (a) relatively new as a ‘thing’ and (b) one I have not been actively involved in.

To this end I did a bit of pre-reading and thinking about what a data trust was for me in the context of this particular meeting (from someone else’s crib notes, thanks Jack). From this how I now understand what a data trust is / could be (note: this is how I understand it) as follows:

A data trust could be a ‘vehicle’ for the collective management of data rights. Rights in this context could be any right accrued. For example by the fact that the data is about a person, so a personal data right.

A data trust may be a mechanism of collectively managing the rights of individuals in an efficient manner so others can access and re-use their data without having to seek permission from every individual every time.

In this case, this is distinct from the right of ownership of the database that may exist under existing database and copyright laws (although a trust could be a mechanism for the collective management of these too).

I stumbled upon an analogy as my brain was turning over during the meeting, pension funds.

In that context money that belongs to an individual is pooled into a pot and then that pot is invested all together according to rules set out on behalf of all of the members of the fund.

There is minimum intervention on behalf of the individual(s). Intervention is not (usually) needed as the fund is being managed on behalf of the collective according to a set of agreed rules.

My bad sketch trying to work all this out.

Tangent -> This led me to ponder flood:re, was this a data trust – data pooled together to ensure a particular outcome… I am still not sure.

Interestingly, in spite of my “detailed” research it quickly became clear in the meeting that this was not the topic that we needed to talk about!

The actual topic was something I know much much more about.

How to generate trust in data sharing not how do we create a ‘data trust’.

Interoperability 2.0?

It is easy to think, if you have worked in and around data for years that many of the conversations of the past are now done and dusted. I remember about 15 years ago talking about ‘interoperability’, in particular in the context of geospatial data, and how it was essential to ensure that data could be used together.

But technology, right?

Computing power means we need to worry about this less now, doesn’t it?

Well, whilst it is true that some of the problems of the past can be tackled by computing power, the basic basic problem that needs to be tackled is still identical and still needs investment in tackling.

It is us.

People.

The “brains” behind the systems.

People. The brains behind the systems.

A lightbulb went off in my head. The conversation in the meeting I mentioned was very similar to conversations from 15 years ago when talking about interoperability in that it was based on the lack of interoperability (communication) between people rather than the systems themselves.

What did that lightbulb mean for the meeting I was having? Well, it meant that I could focus in on the basics of data sharing and the human impact on actually doing it.

There are a set of worries that everyone working in data sharing and open data will have heard, for example:

• We are worried about the data quality,
• What if someone uses the data incorrectly,
• We will lose our competitive advantage
• The data cost us a fortune to collect so it is worth a fortune,
• Surely there is some personal data in there
• What if someone finds something we missed
• It’s just too bloody hard to even think about it

The best way I have seen of representing this is this bingo card:

Credit to Lucy Knight (@jargonautical on twitter)

A lot of these in my mind come down to levels of trust. Those in control of [making decisions about] the data not trusting that in [sensibly] sharing it the benefits accrued could outweigh any of the potential risks.

Ironically, given the amount of effort in releasing and sharing vast amounts of data, particularly as open data there is very little that has been pulled together to create coherent and compelling bespoke evidence to the contrary so it is easy to not take this step.

I have written before about culture change [here and here] and how culture change starts with behaviour change. That is we all need to change to achieve the end goal, life is a negotiation, it is rare that we get exactly what we want without some compromise. This is true too in making the case for sharing more data.

How we engage with these decision makers and the language that we use needs to be adaptable to help make coherent, bespoke, cases for change. We need to become / provide the “interoperability adaptors” (sorry not sorry). One size does not and will not fit all.

The Leap of Faith

Personally, I believe that sharing data widely is a good thing. For 2 main reasons:

Being more transparent, allowing people to see in under the hood allows people to engage, challenge and in the end understand more which in turn will build and develop trust. Think about gender pay gap data as an example – sharing the data shows the problem and then you can acknowledge it and tackle it. Not sharing the data might make you look like you do not accept the challenge.

Being open to innovation. Innovate or die, right? By sharing data widely you can allow others, the hive mind, to look at it use it, develop with it create things, find out things which ultimately are likely to benefit you. [Note: one of my favourite stories around innovate or dying is the one about Kodak inventing the first digital camera and then deciding not to pursue it as they were a camera film company. Ooops]

So when trying to build trust in taking the step to share more data there is a key precursor.

Start with “Why”

[HT Simon Sinek]

Why would it be beneficial for you/them/everyone to share this data? Is it:

• Transparency?
• Reputation?
• Building a community?
• A loss leader for ancillary services?
• Sharing knowledge?
• Something else?

Key thing is to not assume, but to work it out. I have yet to speak to a company where there isn’t any data that could be shared with minimal impact on them. But often the benefit of doing so is either not clear, not articulated well enough or not relevant to them.

Right, we are starting to work out how to build trust aren’t we.

We are doing this by understanding who it is we are dealing with. Trying to think like them and trying to articulate benefits in their terms. Will this lead to more data sharing? Perhaps, but probably not. Why not? Because we haven’t tackled the actual worries identified above.

One of the biggest things about getting 10,000+ datasets ready to share as open data in Defra was the huge leap we had to make, in many cases, from closed data to open data.

Thinking about it from the ‘data custodian’ point of view, for years this data has been sat on, squirrelled away, only coming out to play when they decide it is needed to be used, it is closed to others.

We were coming along and saying, don’t worry about all the reasons you have kept it closed, release it so anyone can do what they want with it, it will be fine…

Fortunately for the target we were trying to achieve there was a big stick available to make them comply with our plan. But if that hadn’t existed then I think a jump this big would have never been made.

Take Incremental Steps

It is fair to say that it is not necessary to leap to open, especially when trying to build trust. I have been discussing with another group of people that controlled data sharing between a group of known data users is likely to build more trust and more belief that the worries are that, worries that can be overcome rather than insurmountable barriers.

This step might be the biggest step, and will involve sensitive negotiation to ensure it works, and this negotiation is multiplied depending on how many different organisations are involved.

How do you do controlled data sharing?

FIRSTLY

Be crystal clear about the data that is being talked about. There is always a problem that when you say data, depending on your perspective you will think different things. Information Security types may think commercially sensitive data, Data Protection folk may personal information, GIS peeps may think map type data, add in HR, Legal, Finance, Senior Managers etc etc and you will realise that each sees different things when you use the word ‘data’.

I saw a badge the other weekend at UKGovCamp19 which said “metadata is sexy”. I am not sure about that, but metadata is CRITICAL in being clear about what the data is that is being talked about.

Once you know this, exactly, then you can move on…

SECONDLY

Use data licensing, and this in itself shouldn’t be as daunting as it looks on the face of it. Creative Commons licences provide a relatively easy way of issuing permissions without having to create bespoke licences and there is a range of straightforward options from completely giving up your rights to reasonably restrictive use scenarios.

Depending on the amount of protection required, then it maybe that more bespoke licences are needed but the cost of doing this then needs to be part of the decision making process.

It should be remembered thought that the problem with licences is that it is up to you to police compliance with the licence… this can be onerous and over the 20 years I have worked in and around this policing isn’t usually actively carried out.

FINALLY

The other big thing is that sharing (some) data should not be a leap of faith with unknown risks anymore.

There are plenty of examples of where data has been shared widely and organisations have seen benefits. There are examples out there of innovation using data that has been shared outside of its original context.

The message from all of my waffle here is that you need to think about what you want to achieve and why you want that. Then consider whether, in order to get there your organisation need, persuasion, training, coercion, mediation, awareness raising, evidence, an independent view, examples, business cases etc to make the case.

Back to the meeting and what I got from it.

Realisation that we need to remember that data sharing has always been and currently still is all about people as well as our current focus on working out how we share data about people.

So there it is, my return to thinking in the open, it may not make total sense but it is a thinking process.

I love this area, the intersection between data, people, the law, policy and achieving outcomes and trying to plot a course through it. I am amazed how often it comes back to people and needing to work on how people interact.

Share your thoughts, I’m always up for a chat!

Find me on twitter -> @MrMikeRose