LabPlus: Lessons about service innovation at French State Incubator beta.gouv.fr

I thought it would just be a day of chatting and now I'm in front of an audience. So I just dug out a presentation I gave last year at the Ninth European Quality Conference. So it's almost exactly one year old. I did not hide that. I really want to stress, though, that I'm not speaking today here on behalf of any French agency. I left my job as a civil servant last September. So I'm just here to discover and share lessons learned. But, yeah. I'm not here representing anyone apart from myself.

So, this talk was mostly targeted at executives. So if you feel like it's too high level and not going enough in the details, just let know, at some point. There's supposed to be one overview presentation. And then there is an in-depth workshop. I'll first go to the overview presentation. And if, at the end of that, you have any questions then just let me know and I can adjust the other part.

So first I want to give you a bit of the bureaucratic context in which we have been operating. It's not the most interesting part. But I think it's really important to understand. And so you can just decide how much this is close to your own context or not. So, we were part of the Prime Minister's services. That's about 800 people. Inside of which is the general secretariat for modernization of public action, which is then split in two parts, which one is the [French] just helping other agencies transform themselves in a very general way by helping them being more user focused as in [French] which is all about creating groups of users that you can then go to to try out new public policies or services.

[French] which is a network of regional innovation centres where local city councils and regional councils can stay at. And then in [French], so the interactional direction of digital and IT from the state. That's very long acronyms there. There's one entity that's performance and resource management thing. They have an awful name but actually what they're doing is quite cool. Basically whenever one project, reaches a threshold of 3 million euros of investment in IT, they can just go and say, "Mmm that's a lot of money. Are you sure everything is all right?" And they can make recommendations. And if the budget goes over 6 million, they can actually stop the project until some major changes are made. That has been very important. We have had some very sad stories of projects going way over budget and nobody having the right to stop them after 400 million has been spent. So that's it.

Then there is Etalab, which is the task force for open data. And there is beta.gouv.fr which is what I will be mostly focused on today. It is the French state incubator. So what you see here in grey is the number of civil servants that are attached to each of these missions. And it seems like it's about 10 in all cases. But actually the count of civil servants is not the proper way to assess our head count. Because we bring in mixed talents by mixing engagement groups.

So what that means is we have in yellow here at the bottom, the actual civil servants that are employed directly by the Prime Minister's office. And then in green are civil servants that are brought in through secondment. Then contractors, in purple is independent contractors. And in orange is the people who brought in through outsourcing companies. I'm not sure if a distinction is extremely clear. But basically, it's whenever you take a single talent or if you just ask for a workforce.

So having this mix is quite interesting and very unusual in Europe, and in France especially. Maybe for you, it's common practise. For us, it wasn't so much. And that's quite interesting to notice because that's also what makes this initiative most fit for innovation because there is such a strong mix of cultures. It's not just civil servants who have been doing the same thing again and again. And it's also not only contractors who are used to being outsourced without caring at all. So interesting mix here.

But the most important thing is that we work in the heart of the state and that all together, our aim is to spread the culture of digital innovation throughout the administration. And the way we do that is through State Startups. So State Startup mostly sounds like an oxymoron. But it's more than that.

The main point is that there is no capital investment. There is no separate traditional legal entity. It's just a match between a team and a mission. So what do we mean by team in this context is two to four people that work together as an autonomous and interdisciplinary team that are matched with a mission which is solving one specific friction between the administration and the citizens. They have to prove that they can solve that friction through a digital tool in under six months and with a limited budget of 200,000 euros, which is basically enough to pay partly two to four people or pay two people and have just a bit of additional work that can be labour such as communications staff.

The very first state startup was the new French open data portal, data.gouv.fr. It was opened in June, 2013. Now there are over, I think, 36 now. So obviously I won't be presenting the whole portfolio today. But I will give you a few examples. And first I give you the demonstration with data.gouv.fr, the open data portal. And the interesting here to notice that it's not just a catalogue of data sets. What made it really stand out, especially back in 2013, I don't know how aquainted you were with the open data movement at that time, but almost, I think, all government initiatives around that time were just big catalogues of Excel spreadsheets.

And under six months and with a team of two people, we were able to deliver a collaborative platform for the data. So when you have a data set, let's say for example, here you see that what we are featuring-- here you can just search for a data set-- but what we are featuring actually reuses. And that really was a major shift in mindset for government, not just pushing the data and saying, hey, this is what we've done, but actually this is what you have done. And we are giving you tools to empower yourself. So, really showing the reuses--

-- So people submit them, or you went and found them? --

Yeah. People submit them because anyone can create an account and be a producer. So the producers will have a stamp like this when they are certified public services. Otherwise, anyone can really publish a data set. We do moderation only after the fact. We have a few funny lessons, of course, especially in the time when we had someone publishing the pornhub data set.

-- Did that use data sets? --

That was very, very, very interesting. It led to a lot of discussions. But anyway, one interesting thing that you can see here is that we are-- so that's an official public service that's publishing the data set. So what we have here is the reuse. And here is the list of the data sets that I used. And if I go to its data set, I can actually contact directly the data producer, public service that is producing. And I can also file an issue with them.

-- We need that --

That's an interesting example of how you can induce modernization and change in mindset throughout the administration just by adding a few select features in a digital public service. So you only have the hand on that interface. And if you were to ask all administration in this case, all agencies, if they are willing to discuss with anyone, any one from the public about the data they are producing, obviously they will say, no way. We know what is good. And we know what is good for the people. And why the hell would they be able to request any changes?

But if you can provide them with this opportunity of actually showing off their data and proving to them that it yields value because there are reuse cases that are made out of it. So basically it's free advertisement or even free analysis that's made out of the data that they produce for free. They are all keen to join. And then there is just this small feature that we users say oh, by the way, your data is slightly broken in a way. It just feels like it's normal practise.

Sorry.

-- How many people run this product? --

This one now? So if you count all together, that's about six people now. There is a lot-- at the beginning it was a lot of business development, of course. I'm trying to get all agencies on board, then try to convince them to enrol. Now we still need a lot of contacts there, especially when there are new data sets to open. There's usually new people to teach.

-- We have one person --

We have one who is really the expert in all the networking part.

-- Can you show the buttons again. What were the other buttons? --

This one?

-- No, no, no. The report issue. So all those. I see. So you can contact the agency that created the data and you can report that the data is broken. Can you request that the data is more useful? I'm just stealing features. These are great.--

Thanks.

Just to answer very quickly. You can't adapt it. That's too much of an engagement. What's interesting though is there are two things you can do beyond-- two other things that go beyond publishing reuses. You can also publish community resources. So basically, the agency is publishing that data set in a raw formats and it's not very useful. I, for my own reuse case, had to create that set of scripts that really translate the data and make it much more useful by normalising all the field sets, for example. Well, you can then re-publish that as a community resource.

So that's the example of the first State Startup that was created. So that's now four years old.

So what's really interesting to note here is not only a catalogue, but a full-fledged social network that's allowing re-users to connect with producers of open data. So we brought in this feedback loop that the state sometimes needs without even knowing it needs it. Now another example is a web app that helps citizens know which benefits they're eligible to. So here the base friction, you might remember, State Startup: match between team and mission. So which friction are we solving here? For this app, Mes Aides the base fiction is [French]. So it's people who have a right to a benefit but are not claiming them. In 2008, we had 80% of people who were eligible to a complementary health insurance that did not request it. 80% of the people were entitled to it, not 80% of the general population.

So I'm saying kind of very scary stats. 68% of low income workers who were eligible to a complement of revenue did not ask for it. So that just piles up. Because basically, after several months of research that have been made by one agency, they have identified different reasons why. Mostly it was bad information. It was about complexity.

So the answer that the administration came up with was consolidating all forms in a single claim file so that you would only send one form to one agency that would then dispatch it to all the others. But the thing is that there are many agencies that are involved in these benefits, which was one of the main frictions at at the beginning. And obviously, they don't have all their rules aligned. So the best the administration could come up with this huge paper form which just took the sum of all the constraints. So it ended up with a dozen sheets of paper really. And of course, terrible user experience so that would probably not have solved the actual friction. And so that would have most probably ended up like a classic case of, we've put a lot of effort into it, and we don't understand why it doesn't work out.

So that's where we started incubating a State Startup. So this app has been available online since September, 2014. So we suggest for a few basic informations, such as your age, national ID. Same questions for your children, spouse if you have any. And then you get a result that's basically tells you, you most likely entitled to this and this and this benefits.

So we started with seven benefits there, with some specific situations that were not handled. We had about 70 tests that were covering most common cases. Now we have 27 benefits that are covered with a dozen agencies that are there, including local city councils. And that's very interesting, actually. It was a central government initiative. And we built in a way - there's a software and social architecture in a way that you could very easily plug in additional rules.

So what you see here on this second row, there are benefits that are provided by the Paris council while on top, it's benefits that are provided by the national government. And it's always more precise. Now we have over 800 tests that have been contributed by agencies.

So here the question is obviously, how do we scale. How did we manage to go from seven benefits to 27? And how were we able to get more agencies in and how were we able to increase precision? Well, the strategy is delivery. I'm not the first one to say that. So what that means exactly is that we made this thing public very early. And we collected feedback. We iterated and we were always very open. Now, if you're all part of the lab, what I'm saying is probably just basic, standard practise. But it's very important to grasp. And it's not often understood throughout the state. Just for the record, I think our panel, I explained a bit more and elaborated it.

If we had waited until all the benefits were perfectly computed, we would never have finished. Because there is no way you can have 100% of the rules perfectly correct. Why so? Because there is what's written in law. That details about 95% of the cases. And then there are always grey areas. And it's really just day to day practise, inside knowledge of some civil servants that are within these agencies. I built in a bit. So if we had waited until we have all the green lights and everything was perfect, this thing would have never been delivered.

But by starting early, by accepting that we would have only 80% coverage, we were able to collect feedback early on and to know what were the most important and the highest priority cases and situations that we needed to handle. So at first that was just-- it had huge BETA signs like here. At first, it was even bigger than that. And we always made it clear that we are not the agencies that will be giving away the benefits so there might be changes.

And at first, we were really willing to take all the flak if anything went wrong. And even after one year of preparation, we still had some partner agencies that were very afraid that literally hordes of zombie users would come to their counters and like literally throw chairs at people, say oh, you told me on the internet that I was entitled to something and I'm not. I hate you all. I'm going to burn this place down. And that never happened. I can tell you how many meetings I've been to, every two months, facing those same kind of reaction. You cannot do that. And yet we did.

And the reception has been great, actually. Most users were very happy with it. And when they weren't, well, they told us. And they were always very understanding, saying, OK, I understand that that was not expected to be correct. I've been very disappointed because-- and then many of them quite often we also had people telling us that's all right. I mean, I didn't even know I was entitled to this and these 300 euros a month have changed my life. Yeah?

-- Just about, how did you get people over the line to manage that risk, to get them to - what were the things that you did to help them be comfortable with that risk, then?--

That will be for the second part of the presentation. That will be for the in-depth workshop I think. There's a lot of depth in it.

No, no, it's a - Exactly. You know, it's a mix of - basically, there are many different tactics but the over-riding strategy is really just do it. Like literally the strategy is delivery. And I will keep on saying that. And, again, is has been stated much, much stronger than I have, much louder than I will ever been able to do, by GOV.UK. But it just takes some time to truly digest and listen to what that all means. But you just need to get started.

So we used this same approach for very different topics. So the very same open computation engine that powers this benefits entitlement tool, it also powers a service that targets companies. So here, we created an embeddable widget that computes the cost of hiring an employee with full taxes and tax refunds included. So this widget is really super simple. It has now evolved. So it is nicer and shinier to look at. But that really is the same underlying concept.

So this widget is really just black and white. And why would anyone ever use such a public service? It looks really crappy. Well, it looks like this because it can be embedded in any layout and made available to employers in their own system. Because what we want is to solve a friction, not to create new, official portals with just yet another partial information, right. So the whole point is that we offer a very basic version of it. But really, want what you want is to diffuse that innovation where the users are already. You don't want to create yet another place where you want them to come.

Even in terms of investment, it makes much more sense. Because otherwise each agency is spending a lot of their own resources trying to communicate over that new thing, that they've done. So obviously there are some things that are quite wrong in the whole bureaucratic incentives system. And this we will probably not be able to fix anytime soon. But just like you have to be willing to take some risk to make an actual change, you also have to be willing to step down some of the glory that you could get and make sure that you find your joy, that your agency finds its budget through other means than by having a lot of communication around what it has done.

So in our case, we partnered with some other agencies and gave them all the spotlight and all the glory. We just have this small logo at the bottom of the embeddable widget. One directs to our agency, saying, oh, by the way, it's being computed by us. And under that, it's OpenFisca, it's the open source computation engine that powers both of these tools. So the point is, if you're curious about how it's made, you can definitely find the trace. If you don't care and just want to use that thing, well, it's all fine if you believe it's been made by the regional council. This is the CC, which is the regional Chamber of Commerce. And perfectly fine if they get the spotlight for that.

Because what that means, and that's really important to understand, because as I mentioned earlier you still need to justify your budget. You need to be able to sustain that innovation. It's not just that you got the budget for one year. Let's do a lot of stuff, help others do things, and then we're going to die. That doesn't work either.

But you don't necessarily need to be recognised by the whole public, by the general public for everything that you're doing. What you need to make sure is that the other agencies who really care about their image and their identity and feel like they are hated most of the time, finally they can breathe some fresh air. Because then they will realise that it's still possible. And it has nothing to do with them inherently being in the public sector. They can actually be in control of their way of doing innovation. You've just gave them a gentle nudge. And then you can help them go further. But they can do it on their own way. That's very important to do.

So that's on the original Chamber of Commerce, but this widget is also embedded on the website of a private startup. It helps SMEs make their business plan. So, once again, what we want to solve is the friction between the state and users, here employers. If employers do some of their accounting or their projection through other service providers, that is fine. It's all fine.

What we need to make sure, though, is that there is no free riding here. So that is solved partly through licencing and also by making sure that there is still this trace, that we can trace back the innovation. So I won't elaborate on that. If you're interested, I can speak about it more later.

So the first public version of this tool was done in under six months and with only one FTE. Ooh, one man band. One man startup. And now it's in over-- well, I think now it must be around 15 different department websites. We're seeing investment by our partner agencies that are now taking charge of making the rules of that tool.

So what's really important to understand is that this tool can be used for free by any entity, administration, association, commercial entity, anyone. We just create a digital common that makes [French]. However, this really for free doesn't mean that we distribute without any counterpart. We just replace the monetary counterpart by a contribution counterpart. So you can use it for free. But every improvement that is made by a user must be offered back to the community. And then we merge the improvements that we find are being useful for all. And that's one of the ways we are able to scale, actually. Because there is no way a small incubator or innovation thingy is going to handle rules for all of the state and actually all of society.

But if you make it very easy for others to contribute, that you're consistently transparent, engaging, and you're recognised as being trustable, then people will actually be happy to contribute. Because they know their contribution is useful first and foremost for themselves. Because they can use this 80% of the job that's already done then. And they just add the two percent that they really need. And if you manage to get all actors on board, these two percents add up. And you can end up with 99% coverage. So that's what we're doing with OpenFisca

So that's open source software. And obviously the creation of these shiny services will drive administrations to understand the value of opening up. So we are the first re-users and the first consumers of some internal data. So we get access to it in order to solve a known problem. That's how we actually get agencies on board.

We don't go to them explaining how great it would be if they open up. People have tried that for years. Doesn't work. If you show them shiny services that have been already made by and with other agencies, that's what I call shiny toys, they really want it. I can't tell you how many times after-- so for Mes Aides, for example, the benefits entitlement tool. At first, people were very sceptical about it. Now after six months, it was integrated by the the President of the Republic. Oh, all of the sudden things changed a bit.

Now, six months after that, I had two ministers making a joint statement about how great that tool was. I had a lot of journalists coming around. And oh, in the next two months, I had people knocking at my door and I was saying, "Could I really have that Mes Aides for my own benefits. And could I have Mes Aides for this and Mes Aides for that and Mes Aides for everything. And then, the answer is yes, sure, we're going to help you help you do that. Let's do it. Let's help deliver that product. Oh, it actually-- everything is closed. So we really need to access that data to make sure that we deliver that service that your really want.

We can help you do that. Oh, we've mobilised an API. That's all right. You don't need to open it right now. We understand it's a lot of paperwork and you have a lot of uncertainty around there. But everything is there, technically speaking. And so if you separate the technical concerns from the political concerns, you can get working on one or the other independently. So you're going to have people who start the change with senior executives that get it. And then where you will need some more time is to make sure that all civil servants that are there and intermediary management really understands the value of opening up. And you need to walk them through being more-- acting more as a platform.

And in some of the cases, you will start by delivering the service because senior management doesn't really care at all about opening up, but they really care about being seen in a nice light. In both cases, whatever you start with, it's going to take a couple of months for the other part to follow suit. That's fine. Just focus your efforts where it's the most pertinent at one point. Because what I've seen again and again happening is when they get their service out where you for that specific service you created internal APIs and internal data sets, because your engineers need them anyway. Because they can't work in this environment where it takes two months to connect to one IT system. You'd much rather want to have one or two people focusing on making a mapping and offering a nice REST API of a database. And I'm going a bit into technical stuff here, because I know some of you are technical. If you don't understand, it's fine. Just close your ears, shut your ears for two minutes. It's fine.

But I it's also really important for technical people to hear that because sometimes we get very frustrated. So you just do your job correctly. And it's fine if it's not going to be all open at once. Because after there is public recognition for what has been achieved, then the next carrot for the senior management is, oh, it's probably going to take some more time to identify another service that's going to be as useful for citizens.

But you know, you can still get a lot of public recognition by opening up data sets and APIs because what we've done for you on your budget, you could probably get part of it for free if you only open and you had external re-users doing that. And that's where you can get them to open. There are a couple of other tactics that I can talk about, but that's really the gist of it.

So if applying best engineering practises, which means opening APIs. So API stands for Application Programming Interface. Please raise your hand if you're familiar with application programming interfaces. All right. So, basically, that let's the machines access information automatically which allows for a interconnection of systems that were not designed to interconnect in the first place, which is basically any agency silo. And this is quite a change in mindset. Once again, something that looks like a purely technical change that will yield extreme collaboration changes.

So here, what I want to show you is api.gouv.fr. I showed you data.gouv.fr That's where you can download data sets. Now let's go to api.gouv.fr. That's the catalogue of all open APIs that have been provided by public agencies. So let's say that for example, I want to know about the different regions in France where you have different localities, which might have different names or different contours, geographical representations, and a bunch of other administrative information. So one way to, and the usual way, to maintain that data is to just get it once. You download it once and you just hope that nothing really changes. So I'll take that as it is.

If I was to create a new digital service, whether I'm a public or private entity that needed to show a map of the different regions, rather than downloading that data set or just really recreating it from scratch-- like I found on Wikipedia that map and I'm just going to spend half a day re-creating the contours, which happens sometimes-- you can just go to api.gouv.fr and look up region here. So I see that there's an API that's available for me. I have a bit of description. It does seem to be fit for my purpose. I'll just have a look at the technical implementation here.

So I'm interested in the regions here. I click there. Let's say I'm interested in knowing about Paris. So here I'm just sending a query. So it's really just typing the URL here. And let's say that I'm interested in Paris. I want to know all the communes, all the different city councils that have the name Paris.

So that's the reply I get. This is an API, which means that it's basically aimed at machines reading that, not humans. You can see it is still acceptably readable. So what we have here is a list of all the different cities in France whose name matches Paris. So there is the Paris that you all know, which is the capital city. It has many zip codes because it's quite a big city. And then there are a couple of other towns in France whose name has part of Paris in it.

So what's interesting here is this is a public service. And we're now slowly getting to this idea that we could have public data infrastructures. We're used to thinking of public infrastructure as roads and electricity and plumbing, water systems, all that stuff. Now there is a lot of information that's going around. And maybe the role of the state is not only to maintain the technical infrastructures over which information can flow, but maybe also to ensure that there is consistent, high quality access to the information itself.

When we think of water as a public service, what we really want is to have clean water available, not that the plumbing works well just for the sake of the plumbing, et cetera. So here, what I'm showing is APIs that are provided by the public service and that are maintained by it and for which after a couple of years of usage, we have law stating that there's a set of data sets-- and we're looking at nine now-- that should always be cared for by the state because they give vital information about the country. So that's probably the way forward in terms of data management, its interconnections that are offered directly by the different agencies. That's part of their commission.

So what I showed there was just a way for you to get, as a citizen, to get information from the state. But obviously the next step is to for the states to take actions through the APIs. Just like you currently send a paper form saying, hey, I would really like to have this done. And now most of the time you can also use the web form to request, hey, could I have this done.

Well, the next step is to have an API which means there are ways of authenticating me as a user requesting some action to be taken on my behalf. But it won't necessarily be the responsibility of the agency who is doing that action to provide me with the interface for doing it. Or at least, it probably won't be the only interface for doing it. That's how we are going to break the silos.

It's not by talking about silos. It's not going to each agency and saying, oh, it would be really better if your just cooperated. Because they won't be able to do that. There's just all the incentives, all the budgetary constraints are there to make sure that this will never happen. But we have an opportunity through technical change to make the silos obsolete. Not necessarily to remove them, but to empty them of their substance by just making it so easy to interconnect with other agencies to make actions, to share data that you might still feel like you're operating in your own budgetary constraints, but actually you are all working together. Because what you're doing is impacting the work directly of a couple of other agencies.

So we are able to create this network of civil service and of public services through mostly technical means. And the political agreements can be made after the fact and enshrine these interconnections, which were just experimental at the beginning. Once it has been proven that it yields so much value, that's where we will be able to get a lot more cooperation going. So that's how we are making the state a platform, by making public services reply to queries and take actions through APIs.

So to recap, a State Startup starts with a friction then improves through iteration. It's all about collaboration and agility. There's really nothing magical here. It's just a different approach to risk management than what most agencies are used to. Demonstration's usually focused on reducing uncertainty. That's the question that we had earlier. Now we do accept uncertainty. And we do use it, somehow paradoxically, to reduce risk.

There is an increased risk of execution. That's right. But the thing is that you will know about that risk much earlier. So we will decrease the risk, which I believe is much higher, of delivering something that no one really cares about or that is already obsolete by the time it has been shipped. Something that made sense two years ago, but doesn't anymore. So we do accept the risk of failure, not only on small scales, where iterations may fail. We may have spent two weeks working on something and it actually doesn't deliver. That's fine.

But startups as a goal, so new public service, new digital public service, might fail. Because we really thought this specific friction could be solved by a digital tool. And maybe after three months, we've just proven that no, there's no way. We tried four different ways and it just hasn't-- the problem really doesn't find an interface. There's no way we can solve that solely through digital tools. So then it has to be tackled through different means. Because remember, all of that is just a way to modernise the state and improve public service. It's not only about delivering new, shiny interfaces. It's just a means to an end.

Sorry. That was targeted at the EU. But I still think it's quite interesting to know that this is getting really international. So obviously what we've been doing is very similar, conceptually speaking, to the United Kingdom's GDS Initiative, which you probably all know about, GOV.UK information portal. But actually, between the UK, the US, Estonia, France, there are very different approaches. And the differences are obviously in terms of the context but also the political momentum that is behind these initiatives.

So depending just on how much and how strong the political support is, you will be able to have different ambitions. Because, in any case, sometimes you will have to go under the radar. And sometimes you will have to really raise your voice and be annoying at some people. And sometimes you can do business as it should be. Just how much you can do one or the other really just depends on your political support.

For GOV.UK they were lucky enough to have a Prime Minister that said, "That's the future. We're going all in. And it's going to be digital first." Well, if you don't have digital first as a political motto, it's hard. In the States, they made a point more of having an agency and like an internal outsourcing agency. Let's say they would probably hate me if they heard me say it this way. But that's the shortest I can say it. In GOV.UK it was really about becoming bigger and bigger and assimilating all the services.

And for us in France, it was mostly about proving that we can do something and then opening doors as much as possible for agencies to come in. In Estonia, they never really had to think about that at all because they just created it from the beginning like this. So there was not even a need for a strong political support for change. It was just political support for doing it right in the first place. So there are as many ways of implementing change on an organisational level as there are governments and context.

So you have to find yours. I hope this example gave you another point on the map of all the different initiatives. And hopefully you can get a few of your ideas out of it. And, yeah, just one last interesting point I wanted to stress. I talked about OpenFisca, which is an open source computation engine. It's re-usable. It's being already used by other countries. I would be happy for New Zealand to try it out. We have quite a nice community. Just last night here, we had Italy who committed its first phase of their model of their legislation. The committee already made some reviews there. I was just doing that an hour ago. So we have quite a lot of momentum going on at the moment. It's probably a good time to jump on that ship.

And that's about OpenFisca, but for example, the engine update at data.gouv.fr, the social network based around datasets. It's been re-used by Luxembourg. And we have another country that's quite keen on re-using it also. So that's the kind of power that comes with making things open. So it's, of course, nice in terms of feeling like you've done something right. But it also means something about just how much you can share resources. You can just share that cost over time. And by creating commons that go past the boundaries, the nation state boundaries, but also the different sectors boundaries, making sure that you have the private sector and the public sector and other sectors that are all using, relying on the same tool.

Well, public service might be the first investor. It might be the one who's responsibility it is to maintain it. Because it has become such an important tool for the whole of society, doesn't necessarily mean that the state will be responsible for all the improvements and all the costs there. So that should be something quite interesting to keep an eye on. Thanks a lot for your attention.