Decide what your point is before you write

I have spent most of my time recently writing papers for publication, and I have come to a realization. It is one of those realizations where I knew it all along, but had forgotten, or had never considered how important it was. What I realized is this: every paper should have a point, expressible in a sentence or two, and everything in the paper should be relevant to understanding and evaluating that point. Analyze the data, read the literature, analyze the data some more, but before one actually starts writing, one should have a pretty good sense of what one's point is. I've taken to making the point of the paper also be the title of the paper. Here are three titles I've written recently:

• Post-fertile survival in comparative perspective: humans are qualitatively different
• Behavioral biologists don't agree on what constitutes behavior
• The grandmother hypothesis is supported, but only in humans

I don't know that these will be the titles these papers actually have when they get published, but they serve to remind me that there is a point I am trying to make, and I'm not just spilling out everything that I've done or found out or thought. In many cases the point changes somewhat once I start writing the paper, and then I change the title. But in those cases where I start writing not really having a point in mind, I end up in a morass, casting about, writing several pages and then deleting them because they don't really say anything, the bits don't go together into a single logical argument.

The null assumption of many paper writers is that one starts writing at the beginning of the paper, writes until one gets to the end, then stops. In some forms of writing, (e.g. writing a short essay for one's blog) this is probably the most reasonable approach. I have heard it suggested that in writing a scientific paper, one should write the sections in revers order. Compile the list of references one needs to mention, write the conclusion, then the discussion, the results, the methods, the intro and only very last the abstract. I am think that what works well for me in creating a first draft is more like this: Reference, title, abstract, decide what journal I hope to submit to, methods, results, discussion, conclusion, add more references and then re-write the abstract and then write the introduction last, putting in only that information necessary for readers to understand the rest of the paper. These are arranged into the document in the order the journal demands, but I write them in the order that I feel leads to an efficient writing process. Of course I then end up going back and reading it in the final order to make sure the document does not read as disjointed.

Now that I've come to this realization, and begun to implement it in my writing, I need to also impress it upon my students. I have more than one very talented student struggling somewhat in writing a paper for publication, and in some cases I think what the papers lack most is a clear and central point. We need to rectify that. New rule for the lab: decide what your point is first, then continue writing after that.

A little praise goes a long way

I gave a draft to one of my mentors, and it came back covered in red. Comments, suggested edits, deletions, additions, objections, pointing out "wordiness," "bogging down" and opportunities to make it "more fun to read." However the comments start with the statement, "this is a very valuable paper, and I expect it will be widely cited."

A teaspoon of sugar does in fact make the medicine go down in the most delightful way.

Academic constipation

The NYTimes has a recent article about the current academic job market. In two words, it sucks. Professors who thought they could afford to retire are staying on, so positions aren't opening. Even when they do retire, hiring freezes are leaving their positions vacant until the economy improves, so post-doctoral researchers aren't likely to find a tenure-track position, and remain in underpaid temporary jobs. This means there are extremely few positions (as faculty or post-docs) for recent grad-students, and the usual routes out of academia, industry jobs, also aren't available. But while the routes out for grad-students are limited, recent college graduates who can't find real jobs are apply to graduate programs in record numbers.

While the situation is bleaker for the humanities, which have long been in decline economically, the scientific community in the US is keenly aware that however this turns out will have effects that will be felt for generations. The closest comparison I can make is the tremendous expansion of enrollment in colleges in the 1960s and '70s. As the students flooded in, colleges hired large numbers of professors right out of grad school. By the 1980s there were extremely few positions opening up, almost every professor had been hired in the previous two decades, and there were almost no retirements. A whole generation of academics had almost no chance of finding a professorship. I know several people of that era whose careers were permanently put onto alternate tracks because they simply couldn't find a professorship. Then, just in the last decade, that flood of professors hired to teach the baby-boomers have been retiring in droves, and many people were expecting a new wave of hires. More than one person has told me, "When I finished grad school, there were no jobs. When you finish, there will be openings everywhere." Now, given the economy into which I am graduating, I feel lucky to be in a sub-field that has some post-doctoral positions for a few years. Perhaps in three or four years universities will start filling all those vacant positions, and a new wave of young faculty, hopefully including me, will be an impediment to the career aspirations of our younger colleagues.

Evolutionary challenges to gene engeneering a better human

My friend Terry is a bioengineer, as well as a part time futurist. Much of what people in his field work on (as judged by what Terry talks about when he puts on his futurist hat and has a couple of glasses of wine) is thinking about how to modify the human genome to increase our lifespan and healthspan (I just made up the word healthspan, but I bet someone out there is already using it). Much of my work is to understand how and why we evolved to have the lifespan and healthspan we currently do. My understanding of my work is not promising to my understanding of this part of Terry's colleagues' work. In my view, bioengineering a much longer lived human will be extraordinarily difficult for several reasons.

First, we have a great many systems that seem to fail at about the same age, and there are good evolutionary reasons why this should be. Why bother building a femur that last longer than your heart, or your brain, or your pancreas? So to engineer a much longer-lived human, one has to be prepared to make a large number of changes to see a small effect. Terry counters that there will inevitably be some low-hanging fruit, and I concede this point. Simply by editing out some of the purely harmful mutations in the human gene-pool, we can probably extend average life span by a few months or maybe a couple of years. But because so many things fail at similar ages, no one or two or 100 changes could give us healthy 150 year olds.

Second, many individual genes have an enormous number of different effects in a wide range of systems, tissues and traits. When a gene has more than one effect, this is called pleiotropy, and we are chock full of pleiotropies, most of which we don't yet know about or understand. DNA is not like a blueprint, where you can just erase one wall, re-route a few wires, and draw in a new door. It is more like a vastly sprawling and disorganized system of interacting computer applications, add-ons, duplicates, and operating systems (only without any comprehensible order, annotation or easily understood compartmentalization). Something which functions as part of an unnecessary application may also be used in several disparate parts of the operating system. Modify a line of code and all sorts of unintended things can happen. Evolution has fine-tuned this system of interactions through millions of generations of trial and error, with emphasis on the error. Our best computer simulations are barely able to comprehend the folding of a single amino acid string into a protein, let alone a whole cell or organ or human, and animal models only go so far. So the process to modify evolution's optimization would not be fun, fast or clean. Our various bits are tuned to work together, and most potential single modifications can only move us away from that local optimum.

Terry counters that in many cases what evolution was tuning was utility in the form of health/strength/life vs. cost in the form of calories. A large part of the theory of life-history evolution is based on models where developing organisms have limited nutritional resources to invest in important tasks like growing, healing and reproducing. If one assumes unlimited calories are available, one can theoretically grow, reproduce and heal maximally all at the same time. And in Terry's view (which I can't help but see the wisdom in) anyone who can afford to play with the human genome can also afford plenty of potato chips. For the relevant population, calories are no longer limiting. In fact, we go out of our way to burn extra calories now. Spending calories lavishly to buy a few extra years of life or more garish secondary sexual traits is a win-win. The bioengineers of the future will have the advantage over evolution, because they won't have to worry about one of the main constraints evolution was dealing with, calorie restriction. So we may have to change a few things at once to make it all work well together, but we can do that. We can, in my imagining of Terry's thinking, reengineer the organism to its new environment.

It occurred to me last night that there is third, bigger and more insurmountable barrier to re-tuning. One that is not just a technological limitation: Breeding. Humans have been known to breed with each other, and in doing so they mix their genomes. You have half the genome of your biological father and half the genome of your biological mother. Imagine if your uber-mench father had a carefully altered suite of genes, and your mother was a good old-fashioned non-GMO woman. What do you get? You get half a carefully altered genome mixed with genes they were never designed to interact with. Chances are, you have all sorts of wacky health problems, and greatly reduced longevity. It would be like taking half the code of Mac OS 9 and half the code of OS X and expecting a stable operating system.

This means that every change and group of changes would have to be carefully designed to be back-compatible. The alternatives are gene altering the entire human population (which would never ever ever ever work (and I very rarely use that many "ever"s in a row)) or engineering the longevous new humans to be incapable of interbreeding with the old model. They'd have to start by separating off one population as a seperate species, Homo terrii, and only thereafter get serious about reengineering.

So suppose the engineers decide they want to make everything back compatible?
I'm not convinced this would work either. Most mutations are bad for you not only because they break a piece of the system, but because they make a new piece that doesn't work with what is already there. Requiring back compatibility means we have to have every piece work with not only the old set of genes and the new set of genes, but every possible combination of old and new. Evolution, largely free from constraints of time, funding and ethics, accomplishes this by letting those individuals who have bad combinations die out until there are very few harmful combinations possible. To extend the computer code analogy, this would be like trying to write OS XI in such a way that if one blended the code with OS X, it would still work. It is possible to do, but XI would end up looking an awful lot like X, too similar to be more than a service update.

This leaves only the option of creating a population incapable of breeding with normal humans and altering their genes extensively to try to overcome a large number of age-limiting factors at once. Again my understanding of evolution suggests a major difficulty. To do this successfully, one would need a large population all gene-altered simultaneously, to avoid inbreeding effects. One can't start a new population with just a few individuals and expect that species to have a decent chance of surviving well. Even if the species does make it through, there is likely to be an extended period of decreased lifespan and healthspan while the inbreeding kinks work themselves out and the population increases in size and genetic diversity.

Without doubting that bioengineers will continue to make things that seem impossible become projects of undergraduates, I consider it highly unlikely they will achieve any very significant advances in human longevity in the next few decades.

(NOTE: I sent this to Terry for comment or objection some time ago but he has been busy with 'job' and 'editing the book.' I take his failure to offer a substantive reply as evidence that in some basement deep under campus, his department is already failing to build an immortal human.)

Moving the last rotifer

For much of the last year my life and schedule have revolved around daily rotifer census. How often I go to campus, at what times, when I have time for anything else and the energy and time I have for anything else have all depended on lab work. When I could rely on my students to take care of it, I could do other things. Frequently, very frequently, my supply of dependable students was not up to the demands of taking data on and caring for several hundred animals each day. Even when my students are scheduled to do everything, it is rare for a day to go by without questions, problems or scheduling issues. If I am not in lab for a day or two both the quality of the data and the survival of the animals begins to decline.

So it feels like a big deal that my lab work will be done this week. Thursday. I've told my students that after that they are free to continue working on their side projects, but I'm not going to be in the lab. I'm not going to spend hours moving rotifers. I'm not going to be harassing them about keeping the lab organized and the rotifers' containers clean. I'm not going to be on campus six or seven days a week. I'm going to be at home, writing a thesis, and will come to campus on Wednesdays and Thursdays. And I'm taking my desktop (the lab's erstwhile main computer) home.

I like my students, and the rotifers are fascinating, and microscopes are fun. But I really like the idea of not needing to be in the lab every morning at 8. And the prospect of being able to have whole days to work on writing my thesis is positively thrilling.