Serendipity

Like most universities, U of T had a hiring freeze for new faculty for the last few years, as we struggled with budget cuts. Now, we’re starting to look at hiring again, to replace faculty we lost over that time, and to meet the needs of rapidly growing student enrolments. Our department (Computer Science) is just beginning the process of deciding what new faculty positions we wish to argue for, for next year. This means we get to engage in a fascinating process of exploring what we expect to be the future of our field, and where there are opportunities to build exciting new research and education programs. To get a new faculty position, our department has to make a compelling case to the Dean, and the Dean has to balance our request with those from 28 other departments and 46 interdisciplinary groups. So the pitch has to be good.

So here’s my draft pitch:

(1) Create a joint faculty position between the Department of Computer Science and the new School of Environment.

Last summer U of T’s Centre for Environment was relaunched as a School of Environment, housed wholly within the Faculty of Arts and Science. As a school, it can now make up to 49% faculty appointments. [The idea is that to do interdisciplinary research, you need a base in a home department/discipline, where your tenure and promotion will be evaluated, but would spend half your time engaged in inter-disciplinary research and teaching at the School. Hence, a joint position for us would be 51% CS and 49% in the School of Environment.]

A strong relationship between Computer Science and the School of Environment makes sense for a number of reasons. Most environmental science research makes extensive use of computational modelling as a core research tool, and the environmental sciences are one of the greatest producers of big data. As an example, the Earth System Grid currently stores more than 3 petabytes of data from climate models, and this is expected to grow to the point where by the end of the decade a single experiment with a climate model would generate an exabyte of data. This creates a number of exciting opportunities for application of CS tools and algorithms, in a domain that will challenge our capabilities. At the same time, this research is increasingly important to society, as we seek to find ways to feed 9 billion people, protect vital ecosystems, and develop strategies to combat climate change.

There are a number of directions we could go with such a collaboration. My suggestion is to pick one of:

• Climate informatics. A small but growing community is applying machine learning and data mining techniques to climate datasets. Two international workshops have been held in the last two years, and the field has had a number of successes in knowledge discovery that have established its importance to climate science. For a taste of what the field covers, see the agenda of the last CI Workshop.
• Computational Sustainability. Focuses on the decision-support needed for resource allocation to develop sustainable solutions in large-scale complex adaptive systems. This could be viewed as a field of applied artificial intelligence, but to do it properly requires strong interdisciplinary links with ecologists, economists, statisticians, and policy makers. This growing community has run run an annual conference, CompSust, since 2009, as well as tracks at major AI conferences for the last few years.
• Green Computing. Focuses on the large environmental footprint of computing technology, and how to reduce it. Energy efficient computing is a central concern, although I believe an even more interesting approach is when we take a systems approach to understand how and why we consume energy (whether in IT equipment directly, or in devices that IT can monitor and optimize). Again, a series of workshops in the last few years has brought together an active research community (see for example, Greens’2013),

(2) Hire more software engineering professors!

Our software engineering group is now half the size it was a decade ago, as several of our colleagues retired. Here’s where we used to be, but that list of topics and faculty is now hopelessly out of date. A decade ago we had five faculty and plans to grow this to eight by now. Instead, because of the hiring freeze and the retirements, we’re down to three. There were a number of reasons we expected to grow the group, not least because for many years, software engineering was our most popular undergraduate specialist program and we had difficulty covering all the teaching, and also because the SE group had proved to be very successful in bringing in research funding, research prizes, and supervising large numbers of grad students.

Where do we go from here? Deans generally ignore arguments that we should just hire more faculty to replace losses, largely because when faculty retire or leave, that’s the only point at which a university can re-think its priorities. Furthermore, some of our arguments for a bigger software engineering group at U of T went away. Our department withdrew the specialist degree in software engineering, and reduced the number of SE undergrad courses, largely because we didn’t have the faculty to teach them, and finding qualified sessional instructors was always a struggle. In effect, our department has gradually walked away from having a strong software engineering group, due to resource constraints.

I believe very firmly that our department *does* need a strong software engineering group, for a number of reasons. First, it’s an important part of an undergrad CS education. The majority of our students go on to work in the software industry, and for this, it is vital that they have a thorough understanding of the engineering principles of software construction. Many of our competitors in N America run majors and/or specialist programs in software engineering, to feed the enormous demand from the software industry for more graduates. One could argue that this should be left to the engineering schools, but these schools tend to lack sufficient expertise in discrete math and computing theory. I believe that software engineering is rooted intellectually in computer science and that a strong software engineering program needs the participation (and probably the leadership) of a strong computer science department. This argument suggests we should be re-building the strength in software engineering that we used to have in our undergrad program, rather than quietly letting it whither.

Secondly, the complexity of modern software systems makes software engineering research ever more relevant to society. Our ability to invent new software technology continues to outpace our ability to understand the principles by which that software can be made safe and reliable. Software companies regularly come to us seeking to partner with us in joint research and to engage with our grad students. Currently, we have to walk away from most of these opportunities. That means research funding we’re missing out on.

This only runs on Windows, so I’ll have to wait a while for the Mac version before I can try it myself, so maybe in the meantime someone else can play it and tell me what it’s like:

• Fate of the World: Real Science. Real Consequences | Fate of the World.

I’m particularly intrigued by the fact that Myles Allen (famous for climateprediction.net and the Trillionth Tonne study) was a consultant in the game design. Does this mean it brings on board some of the dynamics in the latest GCMs? The Guardian previewed the beta version of the game back in the fall, but they don’t appear to have actually played it. PC Gamer magazine did play it, and concludes that it really does succeed in it’s goal of making people thinks seriously about the issues.

Hmmm, almost makes me want to borrow a PC to try it…

Here’s an interesting opportunity for young coders. The Climate Code Foundation is a mentoring organisation this summer for the Google Summer of Code.

I’ve mentioned the Clear Climate Code project before, but it’s time to give them an even bigger shout out, as the project is a great example of of the kind of thing I’m calling for in my grand challenge paper. The project is building an open source community around the data processing software used in climate science. Their showcase project is an open source Python re-implementation of gistemp, and very impressive it is too.

Now they’ve gone one better, and launched the Climate Code Foundation, a non-profit organisation aimed at “improving the public understanding of climate science through the improvement and publication of climate science software”. The idea is for it to become an umbrella body that will nurture many more open source projects, and promote greater openness of the software tools and data used for the science.

I had a long chat with Nick Barnes, one of the founders of CCF, on the train to Exeter last night, and was very impressed with his enthusiasm and energy. He’s actively seeking more participants, more open source projects for the foundation to support, and of course, for funding to keep the work going. I think this could be the start of something beautiful.

The British Columbia provincial government has set up a Climate Change Data Catalogue, with open access to data such as GHG emissions inventories, records of extreme weather events, and data on energy use by different industrial sectors. They recently held a competition for software developers to create applications that make use of the data, and got some interesting submissions, which were announced this week. Voting is open to vote for the people’s choice winner until Aug 31st.

(h/t to Neil for this)

On March 30, David Mackay, author of Sustainable Energy without the Hot Air, will be giving the J Tuzo Wilson lecture in the dept of Physics (Details of the time/location here). Here’s the abstract for his talk:

How easy is it get off our fossil fuel habit? What do the fundamental limits of physics say about sustainable energy? Could a typical “developed” country live on its own renewables? The technical potential of renewables is often said to be “huge” -but we need to know how this “huge” resource compares with another  “huge”: our huge power consumption. The public discussion of energy policy needs numbers, not adjectives. In this talk I will express power consumption and sustainable production in a single set of personal, human-friendly units. Getting off fossil fuels is not going to be easy, but it is possible.

The book itself is brilliant (and freely available online). But David’s visit is even more relevant, because it will give us a chance to show him a tool our group has been developing to facilitate and share the kinds of calculations that David does so well in the book.

We started from the question of how to take “back of the envelope” calculations and make them explicitly shareable over the web. And not just shareable, but to turn them into structured objects that can be discussed, updated, linked to evidence and so on (in much the same way that wikipedia entries are). Actually, the idea started with Jono’s calculations for the carbon footprint of “paper vs. screen”. When he first showed me his results, we got into a discussion of how other people might validate his calculations, and customize them for different contexts (e.g. for different hardware setups, different parts of the world with different energy mixes, etc). He came up with a graphical layout for the calculations, and we speculated how we would apply version control to this, make it a live calculator (so that changes in the input assumptions propagate like they would in a spreadsheet), and give each node it’s own URL, so that it can be attached to discussions, sources of evidence, etc. We brainstormed a long list of other features we’d want in such a tool, and we’re now busy creating a first prototype.

What kind of tool is it? My short description is that it is a crowd-sourced carbon calculator. Because I find existing carbon calculators to be very frustrating, because I can’t play with the assumptions in the calculations. Effectively, they are closed-source.

At the time we came up with these ideas, we were also working on modeling the analysis in David Mackay’s book (JP shows some preliminary results, here and here), to see if we could come up with a way of comparing his results with other books that also attempt to layout solutions to climate change. We created a domain model (as a UML class diagram), which was big and ugly, and a strategic actor goal model (using i*), which helped to identify key stakeholders, but didn’t capture the main content of Mackay’s analysis. So we tried modeling a chapter of the book as a calculation in Jonathan’s style, and it worked remarkably well. So we realized we needed to actually build the tool. And the rest, as they say, is history. Or at least will be, once we have a demo-able prototype…

I’ve been hearing about Wolfram Alpha a lot lately, and I finally got a chance to watch a demo screencast today, and I have to say, it looks really cool. It’s a combination of a search engine, a set of computational widgets, and a large, curated knowledge base. Exactly the kind of thing we need for playing with climate datasets, and giving a larger audience a glimpse into how climate science is done. The only thing I can see missing (and maybe it’s there, and I just didn’t look hard enough) is the idea of a narrative thread – I want to be able to create a narrated trail through a set of computational widgets to tell a story about how we build up a particular scientific conclusion from a multitude of sources of evidence…

[Update 19 May 2010: They’ve added some climate datasets to Wolfram Alpha]

Having posted last night about how frustrating it is to see the same old lies get recycled in every news report, this morning I’m greeted with the news that there’s now an app for that. I’ve posted before about the skeptical science site. Well, now it’s available on a free iPhone app. I’ve downloaded it and played with it, and it looks fabulous. Here’s the screenshots:

Just perfect for bringing the science to the masses down the pub.