Models and criteria

[John R Hutchinson <jrhutch@stanford.edu>]

Kris (MariusRomanus@aol.com) writes a nice clear post using meteorological 
models' seeming inadequacies to suggest that general models are unreliable, 
and models should be specifically tailored for each case, such as each 
organism.  He also suggests that my criteria for proving my work wrong are 
unrealistic or unfair.

In the case of the animal locomotion models that have been used for decades 
and repeatedly validated, I disagree that a general model is not reliable 
enough.  I know what all the parameters I am putting into the model mean, 
and which ones are least certain, and I know which ones I've left out, and 
what they might mean if I did add them in.  So in general, I trust the 
applicability of my models, but I'm never one to stop with a model and 
assume my work is done.  That's why I wrote a newer paper that goes into 
more detail, and that's why I'll continue to do so.  Also, I am indeed 
designing animal-specific models: 3D models of every individual muscle 
group (rather than harmonic means of muscles acting about joints) in a 
variety of animals.  That we do not have tyrannosaurs to dissect does not 
make modeling them a fanciful or meaningless exercise; it just depends what 
you put into the model and how carefully you examine the uncertainty in the 
model parameters.

To resolve this controversy, dismissing models or anatomy or any approach 
as not useful or reliable is not the way to go.  I won't talk any more 
about that; it's beating a dead horse and was covered 6 months ago.  Even 
if the model is wrong, the model can be used/modified to explain why, 
that's why models are useful.

To be pro-active, I'll outline a few examples (there are more, I am sure) 
of how I think the entire debate could be resolved:

A.  Falsifying our model:
        1. Showing that the model method we used actually does not apply 
to living animals in general.  I put this in as a rhetorical case more than 
as a good option.  It would take tons of work to do, and would basically 
show that some 30 yrs of animal experiments and modeling all are wrong.  It 
would have to show more than just that the numbers are a little off; it 
would have to show that they are way off, changing the conclusions drastically.
        2. Showing that the data we put into the model are inaccurate for 
Tyrannosaurus (because of scaling assumptions, parameters excluded, etc.), 
and using the model or a modification of it to show that more/better data 
give a fundamentally different result.  Such an approach would not only 
need to show that the parameters I put into the model were bad, in my 
opinion, it would also be best if it showed that the parameters we left out 
of the model on a conservative basis do not matter.  In other words, 
finding a solution for a very low T value that makes sense (i.e. not using 
1cm muscle fiber lengths). This is realistic; it could be done.
        3. Similar to above, developing a model that works much better 
than the one we used, and gives different results that totally contradict 
ours, yet still being validated for living animals. Very do-able.  I may 
even unwittingly do it with some of my newer models; who knows.  I'm not 
afraid to prove myself wrong.
        4. Discovery of an unambiguous 11-20m/s running adult tyrannosaur 
(or other huge theropod) trackway.  Quite conceivable.

B. Supporting Greg et al.'s hypothesis:  (Falsifying our model does not 
prove Greg's right)
        1. Developing a biomechanical model that shows tyrannosaurs could 
run 11-20m/s, and works for a variety of living animals and small dinos.  I 
know they're doing something like this, and I might stumble across the same 
thing and show that I was wrong.
        2. The trackway mentioned above.
        3. Showing that the anatomy Greg sees as being indicative of 
11-20m/s running really results in that speed.  Perrsonally, I think this 
would require a biomechanical approach similar to #1 above, but there might 
be another way (e.g. a good non-biomechanical scaling study) that would be 
convincing that I haven't thought of.

And I think the opposite of each case above (using other models to support 
what I've done, which I and others have already succeeded in doing, and Don 
Henderson's work suggests as well, or falsifying Greg's hypothesis with an 
alternative approach) is also important.

Sure, none of the things I've listed above are easy to do.  To me, that is 
satisfying, because it leaves plenty of work to be done for everyone involved.

To my mind, none of these things above have been done to my 
satisfaction.  Maybe I'm hard to please.  Thus I do not consider the 
controversy settled, which is why I continue to work on the problem and 
probably will for years.  If I thought it was settled, I'd work on 
something else and wouldn't respond to criticism.  Maybe someday I will be 
convinced that I am wrong, and then I will write papers using living animal 
experiments, models, anatomy, etc. to help explain why I was wrong, and 
jump aboard that old bandwagon.  But for now, I want to be convinced first.

Obviously one or the other side of the controversy must be wrong 
eventually.  There may be a middle ground that we're both missing, although 
the way we phrased the paper left a lot of shades of grey that I'd consider 
middle ground we covered.  I don't know if Greg thinks his hypothesis could 
ever be wrong, and that's what concerns me.  It would be interesting to 
hear him say that much; "Who knows, I could be wrong."  But the tone of his 
and some other folks' posts suggests to me that they see things in a 
black-and-white fashion: he must be right and I must be wrong, based purely 
on what evidence is available right now.  Whenever I hear (or think I hear; 
maybe I'm misrepresenting his thoughts) dichotomous thought like that in 
paleontology, I feel dubious and look for holes in the science.

That's what people are doing now with my work: looking for holes.  And that 
is fine with me; in fact I would be more worried if I did not hear of 
controversy!  Nothing is worse to a researcher than a field with all big 
questions already solved; that's why I love paleontology.  I will still 
defend my work as long as people don't convince me otherwise.  If they find 
something dramatic that changes the picture, then science has 
progressed.  And we'll move on to something else.  Likewise, if my work is 
supported, science has progressed.  It's all about learning through 
hypothesis testing and reliable methods.

--John R Hutchinson