A good summary of some scientific infighting

Maybe you haven’t noticed, but there’s been some fascinating arguments online about the proper way to analyze evolutionary trees. I’m not even going to try to dig into the details, but I thought this summary was clear and interesting. Go read that.


  1. says

    I’m a pragmatist. I would side with people who use multiple methods, including parsimony, and not with the purists who demand only parsimony.

  2. John Harshman says

    I thought the summary was interesting, but only in a pathological sense. Anyone reading it for information would get a very false impression of all the methods. Of course, this controversy has been dead for a good 20 years or more. Cladistics and the Hennig Society are fossilized remnants of the losing side. Nothing to see here, folks.

    The biggest problem with the explanation is the misunderstanding of the algorithms involved. Parsimony and likelihood are optimality criteria that allow you do choose among trees; in that respect, they’re identical. In both methods, you look at lots of trees and pick the one (or sometimes many) that’s best by that criterion. Neither should be confused with a distance clustering method (mentioned and equated to parsimony) that just builds a single tree. Bayesian methods are something else that’s much harder to explain, though they too rely on models of evolution and examine many trees.

  3. Crip Dyke, Right Reverend Feminist FuckToy of Death & Her Handmaiden says

    So, the non-evolutionary trees, are those the Gingkoes? The Wollemi Pines?

    …I’m so confused and dispirited. You just know that because of this post Ken Ham is going to say that PZ has announced that at least some trees were deliberately created through intentional effort.

  4. Rich Woods says

    @Crip Dyke #4:

    Oops. I first misread your ‘intentional effort’ as ‘international effort’, which if repeated would immediately bring down the ire of Ham et al for very different reasons.

    Still, give them a rope…

  5. Joseph Felsenstein says

    I have some responses to the uproar at Twitter (under the tag #parsimonygate) over the Cladistics editorial. Not having a Twitter account, I put it on the Facebook account I use for support of my PHYLIP program package. It will be found here.

    The real issue is not whether you use the parsimony method, but whether you justify it by a philosophical argument that insists that it is the only logically sound method. That kind of rigid approach has become obsolete, as most evolutionary biologists have concluded that methods should be judged based on statistical criteria, not declared true by definition. John Harshman and I have been part of that movement, resisting the dogmas of the Willi Hennig Society.

  6. biogeo says

    Very interesting controversy, and weird as an outsider. The idea that parsimony is the automatically preferred method seems very out of place in 2016. As a biologist, but not one who works in systematics, I would assume some type of maximum likelihood approach would be closer to the contemporary gold standard. (My prior is towards Bayesianism but I am receptive to new evidence.) At the same time, that so much objection was raised over the word “philosophical” also seems bizarre to me. Obviously, any time we have an argument over methodological issues and what the strengths and weaknesses of specific methods are, from an epistemological perspective, we are having a philosophical argument. These are always part of healthy science. The argument of the editors at Cladistics seems to be that parsimony is obviously the best method for phylogenetic analysis while all others require justification, which does seem like a silly position. But the idea that the use of particular methods should be justified on philosophical (i.e., methodological or epistemological) grounds doesn’t seem that outrageous or uncommon; I encounter complaints that a particular method can’t be used to support a paper’s claims during journal clubs all the time. The idea that an in-depth defense of all methods other than parsimony is required for every paper, though, is clearly pretty extreme.