Gordon Blair | Computing & Environmental Science

Gordon Blair is a Distinguished Professor at Lancaster University. We talk about his research at the intersection of distributed systems and ecology, Lancaster’s unique interdisciplinary Highwire Centre for doctoral training, and his innovative approach to structuring his work time, both how he structures his week and how he manages multi-year projects. Find out more at http://www.ensembleprojects.org


Gordon BlairI am a Distinguished Professor of Distributed Systems at Lancaster University with research interest in the fundamentals of distributed systems and also the application of associated digital innovations in support of environmental science. I currently hold a 5-year EPSRC Senior Fellowship in Digital Technology and Living with Environmental Change (DT/LWEC). I am also principal investigator of the Models in the Cloud Project and also previously the Environmental Internet of Things project investigating different aspects of this technological support. I am also Theme Lead in the Data Science Institute for Environment and co-Editor-in-Chief of the Springer Journal on Internet Services and Applications (JISA).





Andrew Miller:            So, first of all, thank you for joining me on this podcast. The first question that I ask everyone is, how you got interested in computing or science or research.

Gordon Blair:                I’ve been doing computing for many years. I initially applied to study other areas including mathematics, and then when I first discovered programming, I became hooked, so I shifted to do a full course in computer science. So, I’ve doing that since 1980 right to the very start of some of the more substantial computer science courses, and I’ve loved it ever since.

Andrew Miller:            What is it that really gets you excited about programming or computer science research?

Gordon Blair:                Well, I think first of all, obviously programming is what draws a lot of people in, you know, the joys of programming. I think we all love it when we’ve got time to really dive in and solve a complex problem and write code that solves that problem. Frankly, I don’t have time to do much of that these days because I’m running some very substantial projects, but I look back in fondness to the days when I had time to program. I think on top of that it’s the pace of innovation. The pace is quite remarkable, which means you can’t sit on your laurels, so you have to keep re-inventing yourself, and that’s something I’ve always welcomed throughout my career.

Andrew Miller:            You sort of moved into this area of distributed systems. Was that something that was just sort of a very active area when you were first getting into it, or what caused you to focus on that part of computing?

Gordon Blair:                Well, I started off doing my PhD, and then the initial topic of my PhD was computer networking, and I started programming some device drivers for some early ethernet-like systems, and I suddenly realized that this is fun, but there’s something missing. There’s no way of programming these, what will eventually become very complex distributed systems. So, I became interested in the area that’s now known as middleware, you know, how to offer programming abstractions and programming models that allow you to manage the complexity of the underlying distributed system.

Gordon Blair:                So, it was very early days for the subject, and it was incredibly exciting to be involved as this topic emerged, before the textbooks started to emerge. You know, literally you’re defining the concepts that map out the future of this area.

Andrew Miller:            Right. And you’ve had the experience of actually helping to write a textbook in this area. What was that process like? Did you feel like there was a structure that you could follow?

Gordon Blair:                Well, I joined the textbook Coulouris, Dollimore, Kindberg, and now Blair in the fifth edition, so the job I had to do was to bring it up to date and bring some new material in, so it was very obvious, looking at the previous edition, that the topic had shifted quite significantly, and new areas such as cloud computing were now quite fundamental to the subject. So, that meant a very significant reorganization of all the material, which I kind of led on.

Andrew Miller:            And, you’re now applying this to the area of environmental science. Does there have to be an application area in order to do good middleware research?

Gordon Blair:                I think if you’d asked me that question 10 years ago, I would say that it’s not necessarily important to that application area. I think if I look back at some of the work I’ve done in the past, I’ve developed middleware abstractions that I thought were generic, and now that I’ve worked in an application domain I realize that it’s so helpful to be embedded in the real world, and to deal with the complexities of the real world. You know, it actually makes for better distributed systems.

Andrew Miller:            In this work do you end up collaborating with environmental scientists, or is this something that you can sort of use their data and work on your own?

Gordon Blair:                No, you need to work very fundamentally with environmental scientists. And if you work with environmental scientists, there’s actually a whole raft of sciences involved, you know, the people that bring a perspective on maybe the chemistry, or the physics, or the biology, so environmental science itself is a heavily cross-disciplinary area. And then I come in as a computer scientist and maybe bring along some other people who are data scientists, and we’ve all got to work together, and that’s really exciting.

Andrew Miller:            How do you go about finding those people? That’s one question people often ask. “Okay, I think that this application area would be meaningful to my research, but I have no idea where to begin.”

Gordon Blair:                Well, I think we’re fortunate that we’ve been able to attract some substantive funding at Lancaster for this area, and hence we’ve been able to recruit postdocs. Then the trick is to attract postdocs who want to work on environmental problems. You know, certainly if you look at areas like data science, there’s other industries competing out there, but I think we’ve got something to offer. We’re actually offering people the chance to work on the grand challenges of our time around a changing climate, and something that matters, that really matters, and they want to do this.

Andrew Miller:            You’ve just started a new multi-year project on data science of the natural environment. I’m interested to hear how that came about and how you feel your work fits into that project.

Gordon Blair:                Well, first of all, we set up a data science institute at Lancaster, and I was the first head of the environment theme. So, because of that, we started drawing together people from different disciplines from right across campus, and we also established relationships with research centers, particularly the Center for Ecology and Hydrology that has a strong presence on our campus as well. And we started small, looking at what we could do together, and then we had an opportunity to go big and to bid for substantive funding. It all came about from dialogue, from getting to know each other, from understanding each other’s problems and perspectives, and then we were in a position to put together what turned out to be a compelling bid to look at, what does data science look like if applied to the complexities of environmental data?

Andrew Miller:            So, what kinds of things do the centers do in order to promote cross-fertilization? It’s not like you just sort of called up the head of the environmental science and hydrology people and said, “Okay, I need five people to be on this project with me.”

Gordon Blair:                You’re absolutely right. The institutes were set up by the university to encourage cross-department and cross-faculty co-operation. That’s what they’re about. I think in Lancaster we’re a good starting point because it’s always felt like a small village. It’s a relatively large university now, but it’s always felt like a place where you could pick up the phone to anyone from any discipline and where you know people right across campus, so that was a great starting point. And we’ve also had quite a number of cross-disciplinary projects in the past, so I think we understand how to make that work well.

Gordon Blair:                How did we make it happen in terms of data science for the natural environment? We had a relatively modest budget, a budget that was used wisely to talk, you know, to have workshops, to have meetings, to have seminar series, to have a breakfast meeting every Friday. There’s nothing magic. It’s just about dialogue, and drawing people in, and letting them see there’s some really exciting problems at the interface of, say, statistics or AI and the natural environment.

Andrew Miller:            And what specific kind of problems are you tackling in this project?

Gordon Blair:                Well, in the project which we now call DSNE, D-S-N-E, we’re driven by three real-world challenges. So, we’re working on the foundational aspects of data science around machine learning, AI, statistical techniques, but we’re driven by three real-world applications which are representative of the grand challenges of the natural environment, and these are constraining uncertainty around ice sheet melt, in other words, what will happen to the Arctic and Antarctic under global warming. We’re also looking at land use management and how you make decisions informed by data around future land use, whether it be for agriculture or planting trees, what you do with sheep farming in the Lake District near where we live here, and so on, looking at big decisions around land use. And the third one is air quality and human health, so, again, reaching out across disciplines, how is air quality impacted by changing patterns of climate, and what impact does that, then, have on human health?

Andrew Miller:            And for each of those three areas, is it like there’s a cluster interdisciplinary team working on each area, or how have you structured it, I suppose?

Gordon Blair:                That’s a very interesting question. Actually, we thought long and hard about how to structure this to make it effective. We have a core team who are employed as full-time postdocs in the project, and their job is, for three days a week, to work on the foundational elements, so, they may be working on time series analysis, or machine learning techniques, and for two days a week they work together on one of the grand challenges, and then we rotate the grand challenges over time. So, it’s the grand challenges that pull people together.

Andrew Miller:            That’s fascinating. Is that a structure that you have tried in other projects, or is that something that’s new for you in this?

Gordon Blair:                It’s inspired by other projects, but I think it’s quite new for this. I thought long and hard, how do we make this work, building on our experience of previous cross-disciplinary projects, and we thought, this is well worth a try.

Andrew Miller:            And is your role to supervise the overall project, or to focus on one specific area?

Gordon Blair:                The principle investigator is a colleague of mine called David Leslie who is an expert in machine learning, and I support him, so I’m his, kind of, deputy PI, if you like. I’ve got a specific role in, I lead a particular task around what we call virtual labs. You know, virtual labs are experimental areas that are constructed in cloud environments, effectively, where we place the different techniques that we develop. So, for example, if we develop new techniques to look at extremes in the statistics of climate change, we’ll place that as a module in the cloud, and then we’ll build up from that to software stacks that are able to tackle, for example, the ice melt challenge. The outputs will largely be captured in virtual labs which we then intend to make available to others in the community that want to use them or see the techniques we’re using. I’m a firm believer in open science.

Andrew Miller:            One of the other things that’s interesting about Lancaster is you also have the HighWire Centre for Doctoral Training …

Gordon Blair:                Yes.

Andrew Miller:            … which I know you’re a former director of this Centre. And I was looking the description, and it’s fascinating. The idea is to produce PhD graduates who are innovative and have leadership skills, and then can lead transformation in business and society. That’s quite an interesting collection of skills. Can you tell me a little bit about how this project got started, or how this Centre got started, and how it’s evolved over the years?

Gordon Blair:                That’s a long story, but let me try and give you the shorter version of the answer. I think it came about, as with anything, with an opportunity. There was a research call for doctoral training centers in the UK with an encouragement to think about centers that cross disciplines, and when I first heard of this call I thought we could do something really interesting at Lancaster around digital innovation generally, looking broadly at the kind of exciting innovations that come out of computer science departments and how that impacts on society and business. So, we pulled together departments from right across the university and bid together, and the project was successful and has now graduated somewhere upwards of 40 to 50 PhD students all working in a fundamentally cross-disciplinary manner on real-world problems.

Andrew Miller:            Do those PhD graduates get PhDs in, sort of, like, a home department?

Gordon Blair:                The main home for our students is the HighWire Centre. They all have two supervisors. They must have at least two supervisors from different disciplines, and hence almost certainly from different departments. They also have a training year before they move on to their PhD proper where they get courses in how to work effectively across disciplines, also on leadership that you mentioned earlier, so they can be agents of change in society.

Andrew Miller:            Yeah. The leadership, you know, I highlighted that one ’cause that’s not something we often talk about. I mean, I’ve done several conversations for this podcast now, and that’s the first time the word leadership has really come up. People talk about these, sort of, you know, you have to have the skill of empathy or of understanding, but it makes some sort of sense, right, that you would have to … If these are disciplines that are not typically seen together, you have to be a leader in bringing them together. How did the idea of bringing leadership skills into this come about?

Gordon Blair:                I think because we wanted to have impact. We wanted to produce students that could go out there and make a difference. I should perhaps qualify, though, what we mean by leadership.

Andrew Miller:            Oh, yes.

Gordon Blair:                Yeah. There are some very bad examples of leadership in the world today. You know, leadership does not mean that you’re not empathetic.

Andrew Miller:            Right.

Gordon Blair:                Leadership can be inspirational figures. Leadership could be people who listen. So, there are many different styles of leadership, and I think we were, perhaps, looking for a particular kind of style where people would be trusted and could take organizations or society in a particular direction.

Andrew Miller:            One other thing that I’ve heard in the conversations for this series is some different advice about when one should become interdisciplinary, and some people firmly believe that you really should start in one discipline and know it very well, and then sort of mid-career, then you can bring in the work from the other discipline. It sounds like that is not the way that the HighWire Centre is operating. How did you think through that when you were thinking through the Centre?

Gordon Blair:                Yeah. That’s a very, very important issue. I think in practice the people who joined HighWire were not in the early stage of their career necessarily. Surprisingly, for a PhD program, the typical age profile was not 20-something coming off a masters program. We brought people into the Centre who had significant previous experience, whether in industry or in academia, and wanted to add an extra dimension to who they were. So, in that respect, they were already, for example, well trained in a discipline like computer science, or design, or management, which were the core disciplines that we tended to build on. That wasn’t universally the case.

Gordon Blair:                If there’s anything kept me awake at night, it was probably thinking about other careers for people that are fundamentally cross-disciplinary. You know, in practice, perhaps I should have slept better because our students that do have these skills are in great demand and are now in positions of … got to use the term leadership, significant leadership. They’ve got really good jobs, and I’m very proud of that. But it is much more difficult to carve a career in a cross-disciplinary space.

Andrew Miller:            And going back to your own story, there must have been a point at which you sort of realized, “This is a domain that’s ripe for the kind of work that I do.” When was that? How did you have that realization?

Gordon Blair:                I think there’s two answers to that. First of all, a good friend of mine, Keith Beven, who’s one of the world’s leading hydrologists, came and knocked on my door many years ago, saying, “I think you’ve got skills that I need,” and we started collaborating. So, I cut my teeth, for example, looking at wireless sensor networks to study the behavior of rivers under conditions of flooding, and that got me interested in the domain.

Gordon Blair:                I think the true answer to your question is, actually, when I went to the movies. I went to see a film called Chasing Ice, and that had a profound effect on me. I remember leaving that film saying, “Climate change is real. I want to do something about it.” You know, it wasn’t a head thing. It was a heart thing. I was established in the distributed systems community, and I had the textbook, I had the H-index, and I thought to myself, I need to apply this in something that’s really important. So, from then on, I have been seeking ways of funding a research group that would do fundamentally cross-disciplinary work around the future of our planet.

Andrew Miller:            I think it would have to be from the head and the heart in order to deal with the many challenges.

Gordon Blair:                Yeah.

Andrew Miller:            So, something else that you’re interested in is digital innovation. I’m wondering if you can talk about how you see that in connection with your other work.

Gordon Blair:                I think digital innovation inspires everything we do. I mentioned earlier that computer science is a fast-moving field, and we’re responsible for many of the innovations that are impacting on society.

Andrew Miller:            I mean, so much so that I imagine it can feel hard to, sort of, stay centered when the fundamental technologies that underlie the work that you’re doing are changing so rapidly.

Gordon Blair:                Well, I think you stay centered on the problem you’re trying to solve, which in my case tends to be climate- or environmental-related problems. That’s where you can have a focus. You’ve also got to be aware of how fast technology is changing. You could be made to look foolish by thinking that can’t support X at the moment when in five years X’ll be trivial to support. See, you’ve always … need to have that crystal ball to map out the future and to think, well, in the very near future we can have unprecedented data around, for example, areas of outstanding natural beauty, and to use that data to help manage the associated ecosystems. That’s very close.

Gordon Blair:                We also have the capacity, through cloud computing, to store incredible amounts of data about the natural environment, and to run models, and to run models many times so we can help to resolve the uncertainty in the models.

Andrew Miller:            It also seems like there’s a policy implication to a lot of the work that you do. How do you deal with doing fundamental, hard computer science work, doing fundamental, hard environmental science work, while also trying to get it rigorous enough to influence public debate?

Gordon Blair:                That’s a very good question. I think all the research that we do nowadays has stakeholders involved. The work we’re doing will inevitably be working very closely with, for example, the Environment Agency, the government department responsible for the environment in the UK, and Defra, which is responsible for agriculture. So, we’re working with stakeholders, and they’re an intrinsic part of the research. I’m very keen on our research methodology which is agile software development but used in the context of research.

Gordon Blair:                The way we tend to operate is we build a team which involves computer scientists, environmental scientists, also stakeholders that care about these problems, and then we tend to iterate very rapidly. You know, for example, if we’re looking at the use of internet of things technology to study soils or soil moisture content, we’ll build a system very rapidly and say, “Is this what you mean?” and have a dialogue. And as you’re having that dialogue, you’re drawing people in. Everybody is equally involved. Whether you’re the programmer or whether you’re the stakeholder, you have an equal role in the team, and that’s very profound. And that, in turn, means the stakeholders know the science of the data science, and hence are in a strong position to make decisions, which ultimately is what it’s all about.

Andrew Miller:            One challenge that I’ve experienced in doing this kind of work is the different speeds that the different pieces of the operation work at, and I’m thinking especially for policy makers being able to … or government agencies being able to react to what you’re finding in the project. It may take them longer. Or similarly, it may take you longer in order to come up with a new technique to give them the answer that they’re looking for. How do you manage those different speeds?

Gordon Blair:                Well, as well as the DSNE project I’ve already talked about, I’m fortunate enough to have an EPSRC Senior Fellowship in Digital Technology and Environmental Change. In that program we deliberately modified the pace at which we work. So, the way we work is we’ll work in a focused problem domain for one year — three months to understand the problems we’re working with, three months at the end to disseminate the outputs and reflect on what we’ve achieved, and a very intensive six months in the middle where we work with stakeholders in the agile fashion I’ve just described. That mean’s we’re working at a pace that’s not common in research, a much faster pace, but we’re working with our stakeholders, so we’re matching the pace that they need to work at in terms of making fundamental decisions.

Andrew Miller:            The thing that I’m going to have to think through is how intentionally you’ve set up the social organizations across these different groups, the three days, two days; the three months, six months, three months. I suppose if you’re going to get anything done you have to be intentional about that, but that’s something that I haven’t really thought through.

Gordon Blair:                It was quite challenging because we’d finished working on what we called our flood sprint, so, we had a year of working with the flood community, and then we wake up the next morning and we’re working on biodiversity problems … have to build new relationships and understand that domain and the problems, and start afresh. But of course, we all have kind of intellectual drive as well. We’ve all got things that we’re interested in from the previous work that we carry forward.

Andrew Miller:            Mm-hmm (affirmative).

Gordon Blair:                [crosstalk 00:23:54] got cross-cutting themes we work on. It’s an experimental way of working, but I like to try new things. And sometimes the old ways don’t work very well for cross-disciplinary research, so try things out, and just constantly reflect, and be prepared to change if they’re not working.

Andrew Miller:            It’s really fascinating to see that, you know, at the PhD level, at the postdoc level, and at the distinguished professor level, you’ve got these different models for how to engage in interdisciplinary work. A lot of the people listening to this are perhaps thinking about doing interdisciplinary work but are coming from fairly traditional computer science backgrounds, so they may be masters students, or early-stage PhD students. When you talk to those people, what do you say? How do you advise them about how to pursue this kind of work, or strategies they should use in order to do it wisely?

Gordon Blair:                I think what … A PhD is hard work, and I think we all know that. But I think the first advice I would get would be to find a problem that really interests you. I would probably, now wearing that hat I wear now, just say, find a problem that’s really important. You only live once, so if you’re going to be working in something like a PhD problem for three or more years, pick something that’s really meaningful to you and to others around you. I’ve been drawn in to working on environmental problems. That’s given my life real meaning, and I think other people can follow that as well.

Gordon Blair:                I’m not saying that might be right for everyone. I would love to have many more people working in this area ’cause the area needs it. But what I’m saying could apply to many other areas as well. There’s plenty of challenges in health informatics. There’s plenty of societal challenges where computer scientists have a major role to play. Find something that’s meaningful to you. It will put a spring in your step halfway through your PhD when things are getting very difficult and sticky.

Andrew Miller:            Is there anything else you would like to share, either advice or thoughts that have been spurred in this conversation?

Gordon Blair:                Perhaps I could finish with an advert.

Andrew Miller:            Oh, yes.

Gordon Blair:                We recently held a workshop in the Lake District, which is a very beautiful part of the world close to Lancaster, and the workshop was looking at digital technology and environmental change. We finished up that workshop by producing what we called the Windermere Accord — Windermere is the largest lake in the Lake District — and that Accord lays out a challenge to build an international community in this area. So, if any of your listeners are interested in how computer science can help the challenges of the natural environment, please look at the Windermere Accord and get back to us if you would like to join a more global cross-disciplinary effort in this area. You can find the Windermere Accord, I’ve just written a blog entry on our webpage, which is ensembleprojects.org.

Andrew Miller:            Well, thank you. Yes. I encourage everyone to check that out, and thank you so much for sharing your insight and experience with us.

Gordon Blair:                It’s been a real pleasure.