Today I shipped the final revision of a manuscript I started thinking about in 2013 (the first conversation about the idea I had for this work was at dinner with Ryan Chisholm while at the 2013 SIAM meeting in San Diego – he does not remember, we were at a Mexican restaurant sipping Margaritas, maybe that's why), doing the analyses for at the end of 2016 (I believe I was in Santiago de Chile visiting the Universidad Católica when I started with coding), began writing in early 2017, and submitted to journals for publication in late spring 2017.
The title, more descriptive than prescriptive, is: Estimates of vital rates and predictions of population dynamics change along a long-term monitoring program. In the paper, I try to figure out, using one of my model systems, how many years of monitoring we need to make reasonably accurate predictions on the future number of organisms in natural populations. Briefly, models of population dynamics use estimates of life-history and demographic processes and traits to make predictions on the future state of natural populations. Data coming from monitoring (for example, organisms can be tagged and followed through their lifetime to see how their size, weight, shape, and position change over time, along with when they die) are fed to statistical models, which give back estimates of those processes and traits; if the estimates used in the model change, the predictions are also expected to change, and more accurate estimates are expected to lead to more accurate predictions. More data should increase the accuracy of the estimates of processes and traits, but time, money, the importance of the research question, and the accuracy needed for taking decisions when the scope is applied and not theoretical are also in play when defining optimal stopping rules for monitoring programs.
In theory, and maybe in a better world, it could be a helpful paper; most field activities in biology start when researchers can get some money from funding agencies and end when there is no more money coming in. Not much time is spent beforehand on estimating for how many years the monitoring program should go on to get the information, estimates, or knowledge the researchers are looking for. Two years? Five years? The usual answers are "something like that" or "it depends". Ten, maybe. And why not twenty? A partial explanation for the dearth of analyses on the minimum or optimal duration of monitoring programs is that the scientific goals are only one of the goals of these projects, the others being funding the lab and attracting Master's and Ph.D. students and post-docs, along with the requests for collaboration and prestige coming from running long-term monitoring programs.
I am also confident that the paper that took me so long to write and revise will be read by a maximum of twenty people all over the world, and maybe five of them will put some serious thought into it. It is not a matter of being jaded, cynical, or nihilist; I have seen how many people have downloaded some of my previous papers. Quite disappointingly, not enough to warrant a celebratory cocktail despite my well-known passion for Negroni.
Academic writing (I am referring to publications in peer-reviewed journals and not to "academic" in the sense of "pedantic", although the two categories largely overlap) is valuable in the context of an academic career, where publications are the currency for positions, grants, and reputation. You send the paper to a "journal", whose prestigiousness typically depends on some measures of popularity of the papers it publishes, editorial board, and history. The article, when not "desk-rejected" by one of the associate editors, is reviewed by other academics who are working on the same topics, and after 2 to 12 months you receive an e-mail from the handling editor saying that the paper has been rejected, accepted, or most likely when not rejected, that you need to revise it before it is suitable for publication.
Outside of the ivory tower, academic writing is not read, considered, or relevant for anything of professional and personal value I can think of. Academic papers are boring to write, difficult to find, and hard to read. And yet, it's hard to let the unfinished ones go, even after leaving the academic world. More or less one year and a half after leaving academic research for better money in the tech world, I still have the following half-baked manuscripts in my soon-to-be-shut-down pipeline:
- A paper on morphological differences in fish (that is, differences in shape) in populations of the species I worked on for 13 years and changes. Differences in shape, really? It should not be difficult to let it go. Well, unfortunately, it's not so easy. I took approximately 3000 pictures of fish in September 2012, which I then selected, digitized, and analyzed with the help of a student (heroic effort!). I was being helped by my girlfriend at the time (feelings!). It is good fun to see that you can actually tell the populations apart by looking at fish shape (meaning!). The analyses are mostly done, I am so close! (regret!). An emotional tour de force. Why should not I finish writing it then? Well, and it is enough of a reason, you can bet that almost nobody will read it and it won't advance any career I can think of. Time is a finite resource for all of us. There are other feelings, meanings, and, surely, other regrets.
- A manuscript on the effects of extreme events on natural populations. The simulations took a few weeks to run on a cluster – that I remember. I opened years ago a fruitful research line on the effects of extreme events on population dynamics, risk of extinction, and natural selection, but it has been a few years since I started thinking about the problem and I barely remember what I was looking for at the time. I already wrote half of the manuscript anyway. It'd still require way too much effort to complete it. No way I am going to do it.
- A blog post that could be turned into an academic paper on the usefulness of models of mathematical biology for modeling marketing, competition, and viewership predictions problems in entertainment. This paper would be a good read for a broad audience; the main problem here is that I was using Netflix examples, I am not working at Netflix anymore, and without those examples, the paper would be much more theoretical than I would like it to be. I have tried to get box office data for theater releases of movies, but I should pay more than $1000 for the data, and it is not going to happen. Next!
- A paper on the spatial correlation among extreme rainfall events recorded in Slovenia. I paid $600 to some freelancers on Upwork in 2014 for scraping the data from the website of the Slovenian Meteorological Service. The negotiation was quite fun, she told me she had a team and the money was enough, then she asked for more money because she had medical problems – who knows whether it was true or not –, I gave her the money, and she told me I was a most excellent man. A mutually enjoyable collaboration – compliments are always well received and sometimes we have to pay for them. I have a valuable dataset for a critical research question; extreme rainfall can cause landslides and flash floods that can, in turn, wipe out populations of fish living in streams (if this were an academic paper, I would write "stream-dwelling fish populations"...). By estimating the spatial correlation of extreme rainfall events, we can make assumptions on the risk of multiple populations going extinct at the same time, and then set up conservation strategies that reduce the probability of extinction of the whole species. Of course, conservation strategies that will never be applied. It is too bad that I have neither the time nor the strength of character to resurrect data, code, and words. And it is too bad that I worked so hard for getting the data and the compliments. Oh, well.
- A few others. A paper on the genetic structure of fish populations living in adjacent, but isolated, river basins defined by using Single Nucleotide Polymorphisms data. Another on the spatial genetic structure emerging over time within a single population due to habitat fragmentation (I presented some preliminary results at the above-cited 2013 SIAM meeting). A paper on the effects of fish movement on body growth and viceversa. All with half-written manuscripts sitting somewhere in my Dropbox folder.
I have decided that I will let those unfinished papers go, peacefully. Like a boat, leaving the shore to get lost in the foggy waters. On the one hand, I love writing, and I believe there will be quite a bit of writing in my future. I enjoy communicating my thoughts, and I have always loved the written word, both in Italian, my mother tongue, and in English. On the other hand, seriously, no more academic writing. Regret can motivate for a brief period, but long, exciting work requires aspirations for something, not flights from bad feelings and looming regrets for unfinished business. In any case, it is comforting to think that nobody on their death ever said: you know what, I regret not having written that paper on the fast estimation of hierarchical growth models using the Empirical Bayes method and parallel computation (another half-written paper, by the way). Five people would have skimmed it!