We all want to write papers that synthesize, make connections, and fill in gaps. But what if sometimes we need to do the opposite?

A standard template for a scientific paper is to identify a gap in the literature, and fill it. But, as sociologist Kieran Healy astutely remarked yesterday, the most important papers do the opposite:

Now I’m wondering if this point generalizes.

Many ecology papers seek to synthesize, unify, integrate or draw connections. For instance, what if instead of thinking of ecology and evolution as two separate things that happen on different timescales, we treat them as interconnected things that happen on the same timescale? Viola: eco-evolutionary dynamics! As another example, I’m old enough to remember when one of the big motivations for studying the biodiversity-ecosystem function relationship was to unify community ecology and ecosystem ecology. As a third example, Matthews et al. (2011) called for integration of evolutionary biology and ecosystem ecology. As a fourth example, SESYNC is an entire research center dedicated to synthesizing ecology and the social sciences. I could keep going, and I’m sure you could too. You cannot throw a rock in ecology without hitting an opinion/perspective-type paper seeking to Integrate Things with Other Things. And it’s not just ecology; you could cite similar examples from all scholarly disciplines.

It’s easy to understand why. The burgeoning literature on, well, everything, pushes us to specialize. It takes some conscious effort to push back.

And yet. Is it always the case that the optimal amount of synthesis or integration or or whatever is “more than we currently have”? How come nobody ever writes a paper about how we should separate research on topics P and Q, or stop trying to integrate disciplines X and Y? After all, presumably disciplines exist for a reason! Maybe, just occasionally, the reason why nobody’s working at the interface of X and Y is that X and Y have nothing to do with one another, and so there’s no useful work to be done at their interface.

Ironically, there’s a good example of “de-synthesis” in the greatest scientific synthesis ever: Darwin’s Origin of Species. In Darwin’s day, questions about heredity and questions about development were seen as inextricably linked (at least, that’s my understanding, happy to be corrected!) That is, the question “Why do offspring resemble their parents?” was seen as inextricably linked to the question “How does an embryo develop into an adult?” In the Origin, Darwin de-links these questions, by setting aside heredity. He basically says “We don’t know why offspring resemble their parents, and for my purposes we don’t need to know.”

Another good example of “de-synthesis” that comes to my mind is also from a synthetic book: Mark Vellend’s The Theory of Ecological Communities. As I noted in my review, Mark sets out to synthesize theoretical and empirical work on “horizontal” (i.e. within-trophic level) community ecology. But as a side effect of doing that, he implicitly decouples “horizontal” community ecology from “vertical” community ecology–the bits of community ecology focused on trophic structure, trophic cascades, etc.

A third example that comes to mind is philosopher Gregory Cooper’s book The Science of the Struggle For Existence. In that book, Cooper sets out to define the field of ecology–and ends up settling on a definition that, by his own admission, excludes entire subfields such as ecosystem ecology. (My review is here.)

I’m sorely tempted to generalize from those three examples, and suggest that the only time ecologists (or philosophers of ecology) are ok with de-synthesis is if it’s in the service of synthesis. That is, you’re only allowed to break (or decline to make) connections between concepts or disciplines if it’s in the service of making connections between other concepts or disciplines. So that, on balance, you’re still increasing, or at least not decreasing, the total number of connections in the world.

In the comments, suggest things that ecologists link/synthesize/integrate (or want to), that should actually be delinked/disaggregated/disintegrated.

7 thoughts on “We all want to write papers that synthesize, make connections, and fill in gaps. But what if sometimes we need to do the opposite?

  1. I’m going to go a little meta on you, because I think it would be productive to separate two kinds of papers that you’re trying to synthesize/unify here. Kieran’s tweet was about a paper that creates a gap in the literature. Your post is about papers that separate or divide two questions and point out that they might be better addressed separately. I don’t think these are the same thing. Separating two things does not automatically create an interesting gap between them. I think Kieran’s kind of paper would be one that points out “we thought we understood this, but here’s why we really don’t, and I can’t explain this yet”. Many good examples from physics, of course.

    As for *your* type of paper, very early in my career I published a review of key innovation hypotheses (https://www.tandfonline.com/doi/abs/10.1080/10292389509380518, or https://www.researchgate.net/publication/236240517_Key_evolutionary_innovations_and_their_ecological_mechanisms). We pointed out that the kinds of things people point to as key innovations (traits that drive diversification) are really three fairly different things, which, with stunning lack of originality, we called Types I, II, and III. From the Abstract: “We group ecological mechanisms of diversification in three major classes. Diversification may be spurred by innovations that: I) allow invasion of new adaptive zones; II) increase fitness, allowing one clade to replace another; or III) increase the propensity for reproductive or ecological specialization. Key innovations in different classes are likely to produce different evolutionary patterns, and therefore may be supported by different kinds of ecological evidence.” This paper has been cited quite a bit given that we published in a pretty obscure place, but I don’t think it’s changed the discourse all that much 😦

    • “Kieran’s tweet was about a paper that creates a gap in the literature. Your post is about papers that separate or divide two questions and point out that they might be better addressed separately. I don’t think these are the same thing. ”

      You’re absolutely right. Unfortunately, I didn’t see that last night when I wrote the post. I only realized it this morning, just before I saw your comment.

      Another difference between the sort of gap-creation work Kieran’s thinking of, and the sort of gap-creation work I’m thinking of, is that after someone writes his sort of gap-creation paper, other people should try to fill in the gap. Whereas, when someone writes the sort of gap-creation papers I’m talking about, the gap *shouldn’t* be filled in.

      • I think I’ll screenshot the “You’re absolutely right” part of your reply. Not something I hear very often in life 🙂

        Your “another difference” does make sense to me. (“You’re absolutely right.”). Your kind of paper would be successful if someone else sliced it more finely or differently, but it would not have been a success if others just glued the synthesized bits back together 🙂

  2. I would propose one general class of papers that create gaps in the literature: papers that propose new ways of calculating things. I think of this because I tend to write such papers sometimes, and feel like they open up holes that were previously not accessible. Examples: showing how to calculate the sensitivity of population growth rate to changes in demographic parameters created a gap in the literature with the form “what do those sensitivities look like? are there patterns? what are they related to? does the effect of this thing on sensitivity depend on the value of this other thing?” and so on.

    I think of this in terms of theoretical developments because that’s what I do, but maybe it would apply equally to developments in statistics and data analysis, or to new methods for empirical measurements.

    I suppose that sometimes the gaps created by the feeling that “no one has looked at this before but now we have the ability to do so” are fruitful, sometimes they wither away.

    • Yes, this all makes sense.

      Now I’m wondering about a comparative study of which papers that say “we couldn’t measure this before; now we can” get taken up by others, and which don’t. I’d guess that one big driver is data availability.

  3. To some degree, identifying massive information gaps is simply an outgrowth of filling gaps that people don’t risk doing. I think this is true whether we are talking about empirical research papers that seek first to answer a focal question and then shine the light on related unknowns, or review-type papers that seek to provide conceptual frameworks for understructured areas of science and typically take the next step of trying to see how well the evidence aligns with this new organization knowledge.

    The typical formula is to lean heavily on what we can now conclude very proximally from the solid ground of empirical knowledge and established theory, and only at the end to speculate more broadly, and to do so sparingly. Scientists are, by nature, training, and for many logical and scientific reasons, conservative about evidence and what can be inferred from it, and so the speculative portions of research papers, relegated to the final paragraph or two (at best) are typically very conservative in the way they forecast research needs and the degree to which they speculate about possibilities. And I think this is a shame because some of the most fundamentally important contributions that scientists can make aren’t just in the empirical evidence that supports or refutes hypotheses, but in the generation of ideas about the way the world MIGHT work.

    Where extreme creativity and competence meet, one can provide enormous service to the scientific community simply by organizing, articulating and speculating about the possibilities of any one area of science in a highly competent manner that is rooted in existing knowledge and theory. I think there are several reasons why we don’t see this kind of work done more expansively and lengthily in publications. First, journals don’t allow space for it. Second, scientists are commonly self-conscious about speculating like this in ‘public’. However, I also think that, possibly as a consequence of the first two, we have a culture of science that just plain undervalues this type of knowledge. And so, it seems to be papers don’t open knowledge gaps as much as they close them is largely an outgrowth of these issues.

  4. One scientific gap I’d like to create is related to your blog’s discussion of Price’s equation in 2020 (about the special Review issue on it in Proc of Roy Soc 2020 ) .

    First I’d like to ‘fill in the gap’ mentioned in your blog post on how to use the Prince equation –though it appears Gardner, Frank, Queller and others are doing that in one way (Frank’s 2012 paper in J Evol Biol explains why critics like van Veelen are wrong.)

    You have 4 ‘forward looking questions’. I think all 4 are possibly amenable to my solution which sort of come from physics, math, logic and maybe more places. I can’t actually solve these but just have ideas about how to go about solving them.

    Especially the part about partitions and non-zero sum dynamics.
    The techniiques include Feynman-Kac path integrals, renormalization group from many-body / nonequilibrium statsiical physics, statistical number theory / combinatorics, and logic /computability theory (godel incompleteness, NP-completeness, maybe transfinite set theory.

    I think the question of whether long standing questions about Price’s equation can ever be resolved are either ‘yes’ or ‘no’. Similar to quantum theory—likely one can resolve questions in that field like how to use it, but ones about what it means and why it exists, poissibly not, and there will be some open questions that still can’t be solved because they are too complex.

    I view this as as like filling a gap between an island and the mainland by building a bridge. This is what Frank, Gardner, etc are doinig –normal science. The techiques i discuss mostly are possible a faster way. But the ‘non-normal science’ starts when you get to the island –thats when you get to the great unknown which likely requires new math just being developed.
    This is because the island is not a usual one.
    .
    When you ‘open up’ Price equation by appliying it to get to the island, its like you’ve opened Pandora’s box or ‘hit the jackpot’.

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.