Every two years the instrumentation building community in astronomy get together for conference organised by SPIE, the international optical engineering society. It’s big and tiring and crowded, and the best place to get the updates on how all the major telescope and instrument projects around the world are getting on – not to mention a whole lot of community gossip. It’s also a nice opportunity to catch up with friends and former colleagues from around the world. The conference location alternates between Europe and North America, and this year we’re in lovely Montreal.
I’ve already written about the Hack Day I was asked to organise this year, and I’m excited that that will be happening tomorrow. I’ve had some great conversations already with people about their hack ideas, and I’m looking forward to seeing how things come together.
The first talk I attended all week (Sunday morning! not cool, organisers) was by Phil Crosby of CSIRO in Australia, who spoke about success drivers for large high-technology projects, such as the E-ELT and the SKA. Having had a taste of a management role in my job at MPIA in Heidelberg, this is a subject I’m actually really intrigued by. As a manager, PI, or systems engineer how do you ensure that your project really is under control, and importantly, that you recognise red flags when they arise?
Crosby described some research of his own and by others into past mega-projects, looking at budgeting, scheduling and overall success/failure, to identify factors that can predict a project’s success, and how to protect against things going wrong or ultimate failure. Much of the talk was fairly common sense (e.g. “avoid speculative budgeting”) but it was good to see such a systematic overview.
He identifies 7 aspects that raise the likelihood of a project succeeding; the full content will be in the proceedings paper I imagine (I also found this), but a couple of items stuck in my mind. A first was about “balancing realism with enthusiasm”, which he says is the most common reason why projects get descoped. I remember a section of Daniel Kahnemann’s excellent book Thinking, Fast and Slow describing how many of us are naturally biased towards being overly optimistic about what we can achieve in a given amount of time. The creeping deadline on my own SPIE Proceedings paper would suggest that I am certainly guilty of that (sorry editors).
Complexity is a major factor in any large-scale astronomical project, and one that we’re frequently reminded to limit as much as possible. But in modern instrumentation complexity is unavoidable and Crosby points out that complexity never shrinks as a project progresses and keeping a handle on it is crucial to a project’s success.
With the new mega-facilities like E-ELT, LSST and SKA on the horizon, new ground-based astronomy facilities are becoming as complex and costly as industrial (or space-based) projects, and they should be managed in the same way. That means more industry -style project management, rigorous test and verification criteria, more formal reviews, more administrative overhead. I definitely see this evolution happening in my job, and it’s the cause of many grumbles from those preferring a more laid-back academic approach to instrumentation.
There’s a place for tinkering and trying new things on a more flexible timeline, but I do think that if we’re going to build a €1 billion telescope like the E-ELT, this should be managed like other things in the world that cost a billion euros to build. Particularly if it reduces the chances of lengthy delays and unnecessary stress.
This was just one of many fascinating talks I’ve heard in the past few days. The conference runs until Friday so there’s plenty more to come. As well as hosting the Hack Day tomorrow I’ll be presenting a poster of my work on simulations for the E-ELT first-light spectrograph HARMONI tomorrow evening, so I’m preparing for a long and fun day.
[pictures by SK]