d(PhD)/dt

I’ve spent the past month at the Santa Fe Institute’s Complex Systems Summer School. (Pictured is the view from the library at St. John’s College.) Exhilarating stuff. The lectures spanned computational theory, “econophysics,” food webs, herding behavior, and finance models.

Many of the best parts were, unsurprisingly, outside the lectures. We students organized tutorials for each other and formed groups for research projects. Some groups studied the evolution of modularity, language structure, or public goods games. My group analyzed the co-occurrence network of spices and major flavors present in 280,000 recipes at RecipeZaar.com. The idea stemmed from my wanting (a long time ago) to choose the set of n spices that would maximize the total number of recipes I could make. We’re curious if we can use network information to find better solutions than, say, just picking the most common n spices in the recipe database. It would be nice if this work had broader applications — I don’t quite see them yet — though the primary purpose of the project was to have fun and learn some Python, network theory, and Scrabble words/exotic flavors.

I’m beginning to get a sense of how easily skills can be transferred across disciplines. Non-mechanistic time series models in finance are surprisingly similar to the models in ecology. Replicator dynamics and frequency dependent selection abound, at least conceptually, in social science models. I feel pretty lucky to have had so much exposure to interdisciplinary research and methods through the Center for the Study of Complex Systems at Michigan. These connections to other fields help re-orient me: It’s refreshing to be able to think of oneself as a “scientist” more than a “biologist” per se, and not to feel that certain questions are out of bounds.

That said, dissertation research has never been far. I spent every weekend and most evenings doing the same work I do at home. Partly bad coincidence, partly bad planning — I’ll have a lot of hiking to catch up on after I defend!

This week’s issue of Nature contains an obituary for Edward Lorenz, perhaps best known for his eponymous chaotic attractor (left). If you don’t know the Lorenz attractor, you certainly know the basic idea underlying it: some apparently simple systems can appear to behave randomly, though they are actually completely predictable (deterministic) and highly sensitive to initial conditions.

Chaos has been a recurring theme in my life lately — though it never recurs in quite the same way (ha, sorry, couldn’t resist). It has dominated my personal disease dynamics this term and, perhaps worse, my models’. If almost any pattern can result from a few rules and uncertain starting conditions, and there’s only one set of observations, how can we exclude hypotheses about these rules? Methods to distinguish chaos from noise exist. In ecology, they were pioneered by Steve Ellner and Peter Turchin and George Sugihara and Bob May. I’m still trying to decide how to incorporate their insights, and I’m overdue for a visit to other fields. The good news is that as the biological accuracy of my model increases, opportunities for chaos should shrink, which means there’s hope of rejecting hypotheses with more confidence, getting the work published, completing my dissertation, and achieving deep personal satisfaction and enlightenment on a good salary while improving standards of living around the world.

One of my favorite truisms is “All models are wrong, but some are useful” (George Box?). Many of my friends are beginning their clinical rotations in medical school. Over the course of a morning, they might give the initial order for diagnostic blood tests, research and suggest appropriate antibiotics, and help deliver a baby. When I remind myself that I am, strictly speaking, analyzing oblique projections of the truth all day (i.e., lies, and sometimes incorrectly at first), my work can feel maddeningly irrelevant… for a moment. Slightly more often, I wish medical and cultural practices were more deeply motivated. It takes theory and basic research to know which antibiotic treatments encourage the development of resistance, whether it’s worth vaccinating, whether we can ever hope to predict something like the weather, whether we can expect to drive certain pathogens to evolutionary dead-ends. Other theoreticians defend their work on aesthetic grounds, or they argue that what is pretty is often what is true (after Keats, Murray Gell-Mann). I’ve no idea what makes science an almost spiritual practice for some people but not others.

My tentative conclusion, based on years of unscientific observation, is that most people don’t automatically see the value of most research. Services like medicine are a no-brainer, but knowledge (especially theory) easily appears self-indulgent. Maybe this is a fair inference sometimes, but it’s a lousy expectation. I wish we could better convey the usefulness of understanding underlying structures and mechanisms of natural systems — especially in areas outside medicine — and also the quirky or elegant charm of what we find.

1. My experiment had lousy timing. Between two illnesses (including flu — the experience was ironically uninspiring), a conference, physical injury, and obligations other than research, I haven’t felt able to exploit my pent-up scientific longing. Two months later, I’m frankly a little discouraged not to have more results. Some old advice springs to mind: Inspiration for Ph.D. research needs nurturing. If you’re happily working, stop only for Really Important Stuff. I keep forgetting that there’s so much positive feedback once I roll my sleeves up, even when half the things I’m doing don’t work out. Had I known I’d be so delayed these past two months, I’d have taken a bit more with me over winter break just to maintain momentum.

2. Much (Most?) of my day-to-day insight and inspiration comes from talking with other scientists. Before winter break, I sounded a tad solipsistic: Me + question + memories of hundreds of books and articles = new ideas. It was supposed to happen spontaneously. Now, I suspect I’m most productive when trying to incorporate my ideas into someone else’s reference frame. Explaining hypotheses and hunches to another person, whether by email or in a casual conversation, is almost always challenging. It forces me to distill my thoughts further than I would for myself. (Current and former GSIs, you have the right to think “duh” here.)

My point isn’t that communication is an essential part of getting science done, because… duh, again. What surprises me is how reenergizing it is to talk to other people in my field. Heck, I blogged about this in September, and it still amazes me. I suspect that a lot of us have no problem getting excited about our work abstractly or over long time scales. It’s the day to day implementation that can feel arduous, especially when we’re overwhelmed with logistical difficulties relating to a small, small part of our dream, and our ideas begin to atrophy. The best response may not be to retreat in meta reflection (though I bet that does help sometimes) but to keep talking with other people knee-deep in research with us. Call up colleagues, join working groups, etc. It also helps to fantasize about future conversations (read: submit abstracts for conferences many months away). I’m really happy about some phone calls and meetings planned for the next two weeks. This time I was lucky, because my colleagues suggested them. In the future, I’ll have to remember to initiate.

Yeah, I know — anecdotal evidence. Let me know if you can add another data point about what keeps you going.

I draw the reader’s attention to a plot in this week’s issue of Nature:

The plot shows the percentage of articles submitted to academic journals on December 25 out of all papers submitted on the 25ths of all months. The positive trend still holds when submissions only from Europe and North America are considered. There has been a six-fold increase in the proportion of Christmas submissions in the past ten years.

The authors attribute the pattern to growth of the “publish or perish” mentality, more intense teaching and administration schedules that push research into vacation time, and increasing availability of electronic submissions.

I wonder if it’s worth it. Setting aside for a moment the question of priorities, I want to know what amount of focus is most conducive to progress in research: Does science move faster this way at the expense of individual happiness? Are there enough hypomanic workaholics to outcompete us mortals for jobs, or are the people working on holidays actually less happy, and are we engaged in a suicidal arms race? Would the Dec. 25 folks accomplish more — perhaps fewer papers, but each with more impact — if they took more time off?

Most of my notable insights have occurred when I’m not really trying to have them (performance anxiety goes deep). It’s like I have to ask my big questions casually and privately, promise self I don’t really need the answer right away, and, it’s okay, I’ll still respect self if self doesn’t come up with anything — and then once I pretend not to be watching, self comes running up with some neat idea. I try not to revise expectations, but it’s hard. What if I really took the pressure off for a week? I know a few successful scientists who schedule regular vacations to disengage from their work completely. For the first time since starting graduate school, I’m going to try the same thing. It’s an uncontrolled experiment. If I think about science in the next ten days, it’s going to be meta, fictional, or somebody else’s. I’ve brought only two books with me, Solé and Bascompte’s Self-Organization in Complex Ecosystems and Nowak’s Evolutionary Dynamics, and I intend to browse them rather than sweat through every derivation. I saw in the New York Times that Giuliani is in the hospital with flu-like symptoms, and I haven’t checked out the local strains. Go Day 1.

This is also an elaborate tease — I bet (hope) I’ll be all over this stuff in January.

On another note, grad students in active self-improvement mode should check out PLoS Computational Biology’s 10 Simple Rules for Graduate Students. All readers should see David’s EEB holiday party montage. My shots are too grainy to upload. Does anyone have a video of Dave (and, frankly, George) dancing to “I Know What Boys Like”? Also, for the record, I have to praise whoever brings the jello shots year after year — they’re consistently the best I’ve encountered.

Happy holidays!

I hope I live to see the day when desktop sequencers are cheap, accurate, and fast; people sample their symbionts regularly; and we can track communities within and among hosts in real time. After the papers are published, it would make a nice screen saver.

There’s a conference on “Emergence in Physical, Biological, and Social Systems” here on Friday. Evolutionary biologist Richard Lenski will be talking from 2 to 2:45 about phenotypic and genomic evolution in bacteria. If you don’t know his research, check it out — his work is not too different from what’s in the comic.

I’d like to poll some seasoned researchers and ask them what they think they’ll accomplish in the next four weeks. I’d then like to see what they actually do accomplish. Testing effects aside, I bet they’d have better predictive ability than I do with my work. My research is way stochastic. Here’s a sample:

Project 1. A month ago, I encountered some strange results that I thought were due to a numerics error. I explored the behavior more thoroughly and was advised to try programming the whole thing with different software to see if the error disappeared. I could have also explored the behavior with some fancy analytics I don’t know. These were necessary but unpleasant activities. I procrastinated on both. Then, one morning in the dentist’s chair, I thought about the behavior from a biological perspective (as opposed to what happens when I plug the dominant eigenvalue of certain Jacobians into a system of nonlinear polynomials solved by Matlab’s fsolve function), and the behavior made tons of sense. I hadn’t expected it, so the whole ordeal constitutes a minor discovery. I wish I hadn’t avoided it for so long.

Project 2. I planned to run the program over one or two weeks and have results by Thanksgiving, with a draft of the manuscript by Christmas break. We learn that the adapted program doesn’t scale well (in the “it might finish before you die, but not before you run out of funding” sense), and that the approach may be effectively inviable. Sadness ensues, but all is not lost. We’ve got a strategy session planned for the week after Thanksgiving, and we’ll think of new ways to tackle the problem.

Project 3. This project involves (a) obtaining new data and (b) understanding a complicated algorithm.

(a) After months and months of discussion, we finally decided which data we need. It turns out we can get it, we can get more than we thought, and we can probably get it sooner than we thought. This is awesome. The eight-month delay I had been imagining has shrunk to three.

(b) A month ago, I thought learning this algorithm would be straightforward. I spent three weeks deeply misunderstanding it, quite confused and more and more frustrated. Last week I reviewed it in a slightly new representation for the nth time and, voilà, the illuminated truth made me dance. (There are still things I don’t get about it, but I see the structure, and it’s enough to keep me moving.) The lesson here is that I should talk to people more often and more bluntly about what I do and do not know.

Other surprises:

• Challenging discussions about socialism, population growth, and ecological triage during a visit to Howard University (Figure 1).
  
  
  Figure 1. Up-and-coming biologists shading “The Valley” at Howard University.
• Meeting up with a friend and colleague (love how they’re so often intertwined) at Princeton, followed by another meeting over the weekend in NYC. It’s funny how talking about research feels like gossiping and is just as fun, e.g., “Did you hear what flu did in the ’80s?” “Did you read what so-and-so said about mutation rates in PNAS? Do you really believe it?” “He has such nice methods…”
• At Princeton, I ran by the field where I gathered data for my first species-area curve (Figure 2). I realized I’m more into biology than when I started.



Figure 2. Where the author once tossed hula hoops and counted goldenrod and cinquefoil.

The lesson I think I need to learn from all this is to be much more persistent and average over longer time periods — scale up my step size and ignore small failures.

I’d be curious to hear from other people about how predictable their progress is. I’ll report results here.

I’m not going to say any more about the title; I will simply refer the reader to an article about a Dutch museum desperate for a particular insect. It vaguely reminds me of a semi-serious question a college professor asked about the Guinea worm, which is headed for extinction. Should we save the Guinea worm? If half of all species are parasitic, and if we assume they’re not sentient (i.e., undeserving of ethical consideration), do we have an ethical obligation to remove them from sentient hosts?

Practically speaking, our food web models aren’t sufficiently resolved to predict the consequences. We’d surely disrupt the apparent competition mediating coexistence between host species. We’d release some hosts from population regulation. The results could be disastrous, especially if most sentient beings tend to be at higher trophic levels. (Oops, this is theory. Practically speaking, we can’t eradicate most parasites without enormous sums of research time, money, political will, and graduate students.)


Figure 1. Face it: There are benefits.

---

I can think of a few other reasons why parasites are useful:

1. We can use them to reconstruct evolutionary histories of hosts. However, we can usually work with hosts directly to discern their evolutionary histories. More pruriently, we can use parasites to infer our ancestors’ sexual history. In the near future, they may help us resolve transmission networks.
2. With a sprinkling of coalescent theory, phylogenies can be used to reconstruct the population dynamics of hosts.
3. They’re probably extraordinarily useful for learning how to outmaneuver immune systems and manipulate nervous systems.
4. Artistic inspiration.

Though Darwin was referring more to the “random walk” aspect of evolution, parasites make me recall,

What a book a Devil’s chaplain might write about the clumsy, wasteful, blundering low and horridly cruel works of nature.

Parasitism can make predation look downright tender.

On that note, remember you can get your flu shot starting this week at UHS. It looks like there’s a new variant of H3N2 emerging, and the vaccine may provide some cross-protection (see the CDC’s summary of recent worldwide activity). In a year, fingers crossed, I’ll be able to say something about how much.

Happy Friday: Here’s some research-related inspiration for all of us.

First and foremost: Second year students, you’re almost there! It’s just a matter of stringing nouns and verbs together in a coherent, articulate way, to form original and insightful arguments. And I think it’s natural at this point to have no idea if you’re doing that.

Second, I thought I would dispense some social capital, fresh from the social mint that is Rackham:

• As most of you have probably heard, there’s a free breakfast this Monday from 8:30 to 11.
• December 12 is stress-down day. Last year, I remember the invitation describing dinner, board games, video games, movies, and free massages… and I still managed to miss it. Mark your calendars.
• There are free classes and seminars on yoga, self-defense, not smoking, dealing with performance anxiety, stress, and the impostor syndrome throughout the term.

Since I’m the EEB rep for the Rackham Graduate Student Forum, please tell me if you have any questions or suggestions for Rackham’s policies. (Apparently, the free bus rides on AATA came from this forum. Rackham is also now accepting electronic submissions of theses due to members’ initiative.) One student has already suggested I inquire about increasing funding for domestic and international conferences. The forum meets periodically throughout the year. More on this at the next GREEBs meeting.

It’s a beautiful autumn in Ann Arbor, and prudent science bloggers across the Northern Hemisphere are expounding theories of why leaves change color (Figure 1).

Figure 1. What’s out the window.

---

I won’t do that. Like most scientists, I’ve developed an easy facility — that’s my euphemism for compulsion — for generating hypotheses. It’s so sweet and precious to look at the leaves and think, “Woooah, pretty!” for a moment before my thoughts turn to anthocyanin and carotenoid evolution. Others also seem torn by competing perspectives:

The Problem of Describing Trees

The aspen glitters in the wind.
And that delights us.

The leaf flutters, turning,
Because that motion in the heat of summer
Protects its cells from drying out. Likewise the leaf
Of the cottonwood.

The gene pool threw up a wobbly stem
And the tree danced. No.
The tree capitalized.
No. There are limits to saying,
In language, what the tree did.

It is good sometimes for poetry to disenchant us.

Dance with me, dancer. Oh, I will.

Aspens doing something in the wind.

-Robert Hass

Science gives us a richer, though slightly schizophrenic, perspective on the natural world.

Science can challenge us in other ways. (That’s my euphemism for saying that research has been hard this week.) I’m embarking on a really interesting and tough question, and I do not currently have the tools to answer it. It’s both fantastic and, when I haven’t slept enough, slightly overwhelming to realize how much I’ll have to learn to make a decent pass at the problem. Further, I can’t say that these are the tools I’ll be applying in ten years — it’s hard to know what the questions will look like — though what I learn now will undoubtedly shape how I think in the future. What combinatoric possibilities! I fantasize that my mind is growing increasingly protean and adept, though some days it feels more like a random walk in a dark room with sharp, pointy furniture.

A few years ago, I heard Brian Arthur argue that innovators were essentially people with diverse tool sets. U of M professor Scott Page makes a conceptually similar, though more formal and elaborate, argument about problem solvers in his book, The Difference. I imagine that’s what getting a Ph.D. is about. It’s exhilarating. I’m constantly shedding stereotypes of myself, talking with advisers and peers about what to learn, studying with some of the keenest minds around, and crossing my fingers that I’m assembling a good tool set (and that it’ll be slightly recognizable as such by the time I leave). It takes courage and audacity, and yet, damn, this is how knowledge happens.

One of the highlights of my week was collecting paper wasps (Polistes spp.) with Mike. We spent a morning at the Island Lake Rec Area, searching under eaves of picnic structures and outhouses. Mike will be using the wasps to study the evolution of facial recognition. Below is a picture of a wasp immediately after collection:

On Friday, I paid a follow up visit to see how the wasps we collected were faring. Below, you can see a member of the European invasive, P. dominulus, in the lower right and native paper wasps, P. fuscatus, on the left. They’re feeding on rock candy, their staple in the lab.

In addition to rock candy, water, and the occasional worm, Mike also supplies his wasps with (guess!) paper for nest construction. In my humble opinion, Mike could be having much more fun here. Mike says they prefer “pulpy” paper, like construction paper. The dyes apparently survive whatever processing the wasps do, resulting in slightly gaudy constructions:

I think it would be terribly symbolic if the wasps were instead supplied journal articles about themselves. Wendy inquired about using wrapping paper at lunch on Friday. I see a niche market for wasp decor among artistically inclined biologists (or biologically inclined artists).

Update (10/4): Last Saturday, Mike forwarded me a link to an article about a lab in Illinois with similar, if slightly cliché, ideas for the nests. Check out their photo. Execution is very good.