i.
I wake up with the sun today, before my alarm, after a restless night spent tossing and turning in the sticky humidity of early summer. The first thing I do, ironically, is check Facebook. The second thing I do is wonder about my priorities.
I pull on my freshly ironed shirt and khakis and head to my AirBnB host’s kitchen, eerily quiet and empty. Breakfast is a Granny Smith apple and some cottage cheese picked up last night from the IGA down the street. It doesn’t have the same salty tang I’m used to. In fact it doesn’t even have curds, like it’s been whipped into submission. It’s kind of bitter. Who invented this stuff? I need to talk to their product development team.
Where’s my key? I hope I haven’t lost my key on the first day. Is it in the back pocket of the jeans I wore yesterday? Whew. Good.
With that, I head out the door onto the morning-washed streets of Sydney. Within five seconds I realize I have forgotten my wallet. I’ll need my passport and my driver’s license to make the required 100 points of ID. Good thing I thought of it just now. Open, grab, closed, okay back to walking.
I breeze down Enmore Street to King Street, past Nine Thai, past the Turkish ice cream place (how is Turkish ice cream distinctive? what flavors?), past “Café C” (best milkshakes, or “thick shakes” as they call them here). The sun’s getting in my eyes; I need sunglasses, but at this rate I’d probably just lose them. I draw to my left to make way for a jogger huffing past me. Canberra smells mostly like jasmine and tree pollen right now; Sydney, this part anyway, smells like car exhaust and cigarette smoke. Traffic hums on my right, while a train rattles along its tracks on my left. The city’s gearing up for a day of what I imagine will be the frenetic pace of a place where four million people live.
Shenkin Espresso Bar (apparently not Japanese for “coffee”). Cloud 9 (most expensive ice cream sandwiches, but worth every cent). Blossoming Lotus Vegetarian Thai (mostly vegan, in fact). Other Thai restaurants by the dozen. Bookstores, fashion boutiques, shops with newspaper in the windows obscuring a space where once something was sold.
My phone tells me I’ve arrived. This intersection looks like a university, and sure enough, after I cross the street and turn left out of Butlin Avenue, I see the familiar oval of grass, the lounge chairs. No students lounge there; they’re all on their way to their eight-o’clock classes. Somehow I managed to avoid that all through my undergrad years; I don’t know how. Some poor kid stops me to ask about how to operate the parking meter; I might have been able to help him, but I’m on a mission and a tight schedule, so I protest that it’s my first day and I don’t know anything, before zooming onwards.
Fortunately the building I’m looking for is right where the map said it would be. It’s time for my own eight-o’clock class. I’ve visited the University of Sydney for various astronomy-related activities before, but this is different. This town is where I live now. It’s my first day of work as an engineer.
ii.
A quick email summons my new boss, Hugh Durrant-Whyte, to rescue me from my perch on the steps outside the barred, shuttered reception of the School of Information Technology. While I fumble with my messenger bag, he explains with a grin that the Translational Data Science group I have just joined is not technically part of the School, although it’s in the same building. The advantage is that we can naturally incorporate people from a variety of disciplines within the same space. The disadvantage is that we have no reception, or admin staff, not really. Not yet, anyway.
The open-plan offices are bare. I will have a closed-off office which, for the first few months anyway, I’ll have to share with someone else — a machine-learning expert, apparently. Could be fun, and profitable. On the far side of the clusters of empty desks is a shelf of books. Hugh selects a couple and hands them to me; I’ve got homework. There are also stacks of empty cardboard boxes along the near wall. It’s early days yet; the few people who are here have started only recently.
We sit in his office and he hands me a steno pad — first piece of equipment, he quips. Today will be mostly about signing forms, but I’ll also get to meet with some of my new colleagues. Sally, the statistics expert from the School of Business, will show me around and talk math with me. Jane, the Dean’s assistant, will help me corral all these forms. Aldo, one of my colleagues with a CV not unlike mine (a recent transfer from experimental particle physics), will help me get my computer set up and take me out for coffee in a bit.
Hugh describes the challenges ahead. The problem he’d really like me to work on first is a predictive model of human metabolism. There is a substantial locus of expertise on this from the biology side just a few blocks away. Previous thinking on the subject tends to be fractured, focusing on one or more single variables or sites rather than entire pathways, which is where model-builders can really drive progress. Understanding co-morbidities of diseases in which metabolic pathways are implicated is just one of many potential payoffs from the project. The scope is huge, a grand challenge problem that may take quite a while to solve. The approach will be Bayesian, and will likely involve Monte Carlo Markov chains (my favorite) sampling many variables linked in a directed graph. I smile quietly to myself: this sort of thinking is just how I approached modeling supernova light curves.
(I still remember him spending at least half my interview excitedly describing this project, though not in this much detail. Wow, I honestly have no idea how to begin to solve that problem, I said. Neither do I, he said, but we’re gonna find out! He asks me not to have any meetings with the biologists while he’s gone, because he doesn’t want to miss a thing.)
There are other shovel-ready projects as well, with more data than we know what to do with: infectious disease (Aldo is working on this), mental health (they’re still defining the specific problems to be addressed), cancer. The organizational structure of the research enterprises involved in these biomedical problems could charitably be called Byzantine; getting them to share data with each other is nigh impossible with current structures in place. Maybe the future will help us bring them together. Mineral discovery in geology is another grand challenge problem that’s easy to sell to the funding agencies. There are even chances to work on problems in astronomy (surveys for radio transients) and physics (LHC data). Hugh and his colleagues have put in a proposal for a Centre of Excellence — about 30 million dollars over the next seven years — to address these and many similar problems.
This is translational data science — meaning that the goal is not only to drive discovery in all of these different domain areas, but to build sophisticated models that push the boundaries of machine learning and statistical inference. So while I haven’t gotten under the hood of the toolkits I’ve been using to apply techniques like random forest and Gaussian process regression, I’ll be expected to do so here. This actually suits me fine — it’s part of the charm. What used to be “practice” in astronomy is “theory” here.
Eventually Hugh hands me off to Aldo and zips out the door. Aldo helps me attend to my computer first, but it looks like I’ve still got eduroam access via my ANU ID, for the time being. Nobody else will be available until 10:00 at the earliest. So it appears to be coffee time.
iii.
As Aldo introduces me around, I’m asked what my role in the new group is — student? postdoc? I bristle at the term “postdoc” a bit, partly because of the history that led me here. I’m not thinking of my current position as a postdoc because I’m no longer insisting, or even expecting, that it lead to a professorship.
I try not to think about it too much because that shift of focus is fairly recent, and I still get depressed if I brood too long on it. Or try to write about it, as many others already have (I’m hoping to be a “Destigmatizer” for myself and others). The reality is that in the last round, despite my best efforts, I got neither a job offer for a continuing position doing astronomy full-time, nor even for a fellowship which would’ve allowed me to continue my independent research program. I hadn’t let myself fully face this possibility because I felt that if I did, the prophecy would fulfill itself. Up against the prospect of yet more temporary contracts with less responsibility and autonomy than I had working for Brian, I got the sickening feeling that my astronomy career had likely hit a dead end, as though it might be time to do something different.
It’s very hard for me not to think of my situation in terms of value judgments. Was I just not good enough? Hard to say; I didn’t write the most papers and I definitely wasn’t always the last to leave the office, but I did enjoy some measurable success. Am I a quitter, writing self-indulgent “quit lit”? Maybe a bit; but we all struggle to find meaning in the narratives of our lives, to carve out a niche and an identity that hangs together. Was I the victim of an unjust system? In a sense: the economics of academic research drive the job market towards more short, casual appointments, and towards more conservative funding decisions, both of which make it difficult to establish a track record. This is terrible and carries a substantial human cost, but it’s widely known now (by those on the inside, and I feel it should be more widely advertised to those entering the system) and applies across academia, not just within a particular field or even science. I knew it myself when I came to Australia. Blaming the system for being unfair also seems to imply that the people (many of them my friends) who came out on top somehow don’t deserve it, which I simply reject as untrue; there are many more deserving people than there are jobs. More lauded prize fellows than jobs, now.
Unpacking the roots of these problems and how to address them in policy terms could fill another whole blog. But that doesn’t help me make much sense of my own story as an individual; it’s like trying to blame anthropogenic climate change for Hurricane Katrina. Both my own actions and the structure of the system contributed, and on average the way the system currently works hurts lots of people. But except in really egregious cases,* it’s hard to blame the system for the specific outcomes of individuals, especially since we kids are all supposed to be above average. In any case, the victim/self-blame duality keeps me tied to that past failure and doesn’t help me move forward either way. Indeed, I hope to go back and give talks to astronomy Ph.D. students precisely to head them off at the pass from thinking in these terms.
I’m looking for door number three: the one where we acknowledge that life is nonlinear and has no user’s manual, that good judgment comes from experience which comes from bad judgment, that we often have to rewrite our own scripts at great expense and at the last minute. In fact I’ve already done this myself, moving from theory to experiment, from high-energy astrophysics to optical astronomy, and from “particle physics with telescopes” to “astronomy”. It was tough each time, but interestingly, my interests in Bayesian inference carried me across all three transitions. In my last position, I can at least say that Brian gave me what I felt was a fair go: five years to work on the most important problems I could solve. I accomplished some things I can be proud of when I look back on them in ten, fifteen, twenty years. In fact, that’s become my new criterion for working on a problem.
So after digging around in my troubled little soul for a few months, I decided that it was important to take a new tack that would enable clear progress towards greater leadership, autonomy, and impact along paths other than the traditional tenure track. I decided that data science, broadly construed, and software engineering made the most of my existing skills and talents. I looked for jobs which would teach me absolutely everything I could learn about these new disciplines while giving me the most interesting possible work along the way.
From this standpoint, I hit the jackpot in landing my new job with Hugh. Technically it’s still a postdoc, in an interdisciplinary area that is growing explosively. But connections to what I used to derisively call the “real world” are made manifest, rather than remaining implicit (“you’ll have no trouble finding a job with a physics degree”). I’ll have the chance to work with people in academia, but also industry and government, on high-impact problems that will help make people’s lives longer, healthier, and (one might hope) happier. I’ll learn a lot about new techniques and technologies, and about working with other people with backgrounds different from my own. And sure, I can look for faculty positions in engineering, which may be less bitterly contested than in astronomy; but if, after trying this out, I decide I want a job in industry, I can in all honesty claim to already be a data scientist. And I won’t have to feel guilty or bitter about it.
Since Aldo has a very similar story, I don’t really need to go into any of this with him, or with the others. As we head off to acquire coffee, they ask me where I’m from, what my background is. Bayesian inference? Yeah, I’ve heard of that. Tell me more.
* Here I’m talking about the oversupply of junior academics which makes life miserable for everybody. But I also have to acknowledge that as a middle-class able-bodied white cis het male, I have basically won privilege bingo and have no a priori reason to think the system might be out to make life miserable for me personally. Other junior scientists in marginalized groups, though — who are deluged constantly with reminders that they’re not wanted and/or not seen as professionally capable — have a lot more reason and motivation to be angry; not only is the playing field sunk into the ground, but it isn’t level either. I would count those among the “egregious” cases, and there are a lot more of them than anybody wants to admit.