What’s it like to be the chief information officer (CIO) and chief data officer (CDO) at an agency whose core mission is supporting the advancement of science and engineering in just about every way possible that serves the national interest, and supporting a workforce that is among the most technologically sophisticated and curious imaginable in the public sector?
We were lucky enough to sit down with National Science Foundation (NSF) CIO/CDO Dorothy Aronson several weeks ago for an in-depth talk about both the challenges and the joys of that task.
We weren’t expecting breaking news, but got some anyway as Aronson talked at length about an internal NSF technology environment created as an initial incubator for data and tech resources available to pursue innovative ideas across the board. And then, how NSF’s IT operation lassos the tech work on those promising innovations to shepherd them into the discipline of the agency’s central IT organization.
Without further ado, enter Hopper…
MeriTalk: NSF by its mission and its reason for being already seems like such a tech-savvy place. What is the climate for tech-driven innovation within the agency?
Aronson: The National Science Foundation is an amazing place. The people that work here, their job is to seek discoveries and to innovate. They themselves are often pioneers. So, our customer base drives a lot of the innovation here because they have business problems to solve, and they solve them themselves using technology and data.
Often, they can do that better than the central IT organization because they’re so familiar with the problem they’re trying to solve. In a typical modernization process, we have IT people and we have customers, and there’s a lot of time spent by the customer explaining to the IT person exactly what business problem they’re trying to solve, and then testing it – then there’s a whole lot of back and forth.
But when the innovator and the user are the same person, the innovations happen much more rapidly, the customers are much more satisfied with the results, and they’re a perfect fit for the problem.
MeriTalk: So not exactly the strictly top-down tech development model that you might see at other organizations?
Aronson: What I observed at NSF was that there were a lot of these innovative people popping up. Even way before the pandemic, the term shadow IT came up, where people would say, ‘well, that’s not right, people on the edge should not be creating their own solutions because they’re not connected to the center.’ They’re wild, they are in the dark, they are in the shadow.
I always call that local innovation. What we’ve done over the course of time is create a center of excellence for innovation in a sense. We’ve said to the local innovators – who are often highly introverted people who are just wanting to get their jobs done – we’ve said to them, we will help you, we love your innovations, we want them, and we’re collecting them.
MeriTalk: Can you explain to us the structure of that?
Aronson: As the CIO/CDO, my job is to ensure that people are empowered by the products that they create, and also that they’re not creating the same products over and over again.
So, we’ve developed a central repository for these creations that we call Hopper. It’s named after Grace Hopper, who was a U.S. Navy admiral and IT and innovation pioneer who served in the military from 1943 to 1986. It’s a place where local innovators can deposit their innovations, and so people can build on each other’s great ideas.
Hopper is like a tool chest. Innovators can put their tools in Hopper, and chat with each other about their work. People who are not innovators also have access to Hopper. They can reach in to find a tool that may help them simplify their work.
Certain innovations will evolve from Hopper – from a lab environment to being nascent, to being mature, and then to becoming integrated into the central business systems.
It’s this evolution from the customer side into the center that’s been very exciting and difficult – a very difficult process to implement. Hopper is my favorite innovation.
MeriTalk: Is Hopper just available internally at NSF, or can the public take a look?
Aronson: No, Hopper is just on the inside, you can’t get to it. Hopper is available as part of NSF’s internal web presence, it’s an administrative tool. It’s not a tool that’s available to external innovators, although when I’ve talked about it before externally, some people have asked if they could put their own innovations in it.
MeriTalk: So the customer base of NSF in this case would be employees of the agency, or does it go beyond that?
Aronson: Just the people that are inside of NSF performing NSF business functions.
MeriTalk: For people that can access it, what would they see? Is there code, are there research papers?
Aronson: There are links, there are descriptions of the tools, there are ways of rating your likes and dislikes and commenting on the tools, and there are links to the code wherever it might be.
People were creating the solutions on their own, and what we’ve done is we’ve wrangled them together so that people who are innovators can make their products available to other innovators, but also to other people who are not innovating but who might want shortcuts or tools that help them get their work done.
MeriTalk: How long has Hopper been around?
Aronson: We released it during the COVID-19 pandemic, first locally, and now it’s fully available inside of NSF.
MeriTalk: Do you have a gauge for its uptake?
Aronson: With respect to utilization, I don’t have metrics on that. But every other week we do have an innovation forum where innovators come together and talk about their tools and demonstrate them, and make comments on each other’s work. So, there’s a constant conversation going on about what’s going into Hopper.
Hopper has been continuously improving in that when it was initially deployed it was a manual process to add new tools, and now people can automatically submit their own products. It’s also linked very closely to another innovation, which is a knowledge base for data and analytics called the Data and Analytics Hub, so there is Hub and Hopper.
They are siblings in that a lot of the innovation that’s happening within the data analytics community; while people are analyzing NSF data and trying to make decisions, and this is where artificial intelligence is trickling into our world. The Hub is the knowledge base about NSF data.
MeriTalk: Is the advent of Hopper posing any particular challenges?
Aronson: What’s challenging is the cultural distinction between those people who are willing to accept some risk that the product that they’re using doesn’t have a seal of approval of the central IT organization. The central IT organization is forced to play by many, many rules, but the local innovators are not constrained by those rules.
So, there’s a real cultural difference between a person on the edge creating a solution and the people on the inside who are responsible for agency security and data quality. Those people in the middle have a lower tolerance for risk than the people that are innovating, and so moving the innovations from the edge in towards the core to get them integrated into the core systems is a heavy lift.
MeriTalk: Can you tell us more about that lift, and that integration process?
Aronson: For many years before there was cloud computing, central IT had complete control. If they wanted to leverage complete control over the desktop and the network, they could do that, because there was only one way to get to big computing and that was through the network and through the data center, which was under control of the central IT.
As time has passed, and we’ve moved a lot of our computing out of the data center into the cloud, the industry has also started providing IT services from the cloud that were not created centrally. What that has done is if you have some money, you can buy a product or use a product on the internet without the network, without central IT ever having any notion that that’s going on.
In the future, I believe that IT will be even more distributed, that people will be more and more empowered, and that local innovation will be the norm.
And as that happens, it’s very important to recognize the power of the end user and further empower them, rather than try and control them. How do you do that, how do you operate in a paradigm where the end users can do what they need to do? The answer is you create guardrails. And that’s what Hopper is doing.
Hopper says, ‘show me what you are doing, I want to see if the agency can benefit from that.’ Then we on the inside can ask the important security questions, the data management questions, even about the data format.
We want your data, I don’t care how you collect it, but I as the chief data officer – I am both the CIO and chief data officer right now – I want to know about it, and I want the data, because that data can be used for other purposes.
The thing that is most important is network security and data quality. My thinking is how do I get more people to be doing their work exactly the way they want to do their work while the information is also secured and of good quality. Those things are guardrails, they are not hard edges.
What Hopper does is it starts with a lab environment that’s not connected to the network. In addition to the Hopper tool itself, there’s a place in the cloud where people can work freely and experiment with things without impacting the NSF network. I look forward to people joining the lab. If people want their projects to be on the network, they need to register them in the Hopper.
If we discover things that are not registered, we register them. They don’t have to stop using them, but they go through the first step of IT security. As they mature, they go through policy, they go through legal, they go through the technology modernization steps before they get innovated. So, it becomes the full lifecycle including the end user working in any mode.
It doesn’t always happen. A lot of times the tools just get to a certain point where some people choose to use them, others don’t, and they don’t get integrated. But the idea is to get all the way to the end and integrate it.
MeriTalk: That’s quite a process, how is it going?
Aronson: The process is very difficult. We’ve got innovators creating things in the lab and putting them into Hopper. Getting them through that maturity process has been very difficult. It feels like a struggle, and this is where the cultural differences come in because once we get something onto the network, one of the areas of resistance is that it was unexpected.
One of the things I’d like to do to improve things is increase the budget for the unexpected. We have a small innovation pot of money that we can use to move some of these products forward or develop new ones. But we in central IT plan in longer time periods than the innovators, so there’s a synchronization problem where an innovator can come up with something very quickly and spread it to their friends. All of a sudden, there’s a solution. And for the central IT person, that solution isn’t on the roadmap until two years from now.
So, there’s always this conflict of how do we both leverage what we’ve got and continue modernizing for everyone. It just takes longer. If you have 100 voices giving you information, it takes longer to implement the solution than if you have one innovator working on his or her own.
MeriTalk: Hopper sounds like a homegrown effort that would be interesting for other agencies to try. Any advice for other agency technologists who might want to pursue something similar?
Aronson: My advice would be to embrace change and innovation. Changes are coming at a faster and faster rate in the IT and the data worlds, and fearing and opposing that is non-productive. So, embrace the innovation, I would say that’s your only real option.
MeriTalk: Something on the personal side, has technology always been a natural interest for you, or has it been something that’s more been acquired along the way?
Aronson: I did not set out to be a technologist, I wanted to be an artist. I graduated with a dual degree in art and accounting, and I was hired by DuPont for an accounting position. But six months later they put me in a program to learn how to use computers, and that was the start of my technology interest. My journey has been through taking opportunities that came along after that, but I love technology because I love solving problems.
MeriTalk: On the creative front, art and tech are not that far apart…?
Aronson: Within technology is the beauty of it when we see something elegant. I can see the beauty in technology solutions. My primary motive is to make things work, so I guess I am an architect. And when I take the pieces that are there, and put them together, then that’s what a thing like Hopper is. That’s where my career has always been in trying to make something work the best it can work.
MeriTalk: We know that you will moving on at some point soon from the CIO position to become a senior advisor to the agency head on technology, what’s your outlook to the future?
Aronson: What’s happening at NSF is there’s an organizational change underway, and I expect to move over to being focused on data. The agency is growing as a result of the CHIPS and Science Act, and that is a very healthy and exciting change that’s coming to really beef up in both of those areas.
MeriTalk: Last question – what do you enjoy doing in “real life” that has nothing to do with technology?
Aronson: Lots of things – gardening, painting, printmaking, bike-riding, I’m not entirely devoted to tech.
