Why Model?

What is Model thinking, and how can models and paters be useful in resolving problems?

Joshua M. Epstein∗


 Based  on  the  author’s  2008  Bastille  Day  keynote  address  to  the  Second  World  Congress  on  Social Simulation, George Mason University, and earlier addresses at the Institute of Medicine, the University of Michigan, and the Santa Fe Institute.

 The  modeling   enterprise   extends  as  far  back   as   Archimedes;   and  so   does  its misunderstanding.  I  have  been  invited   to  share  my  thoughts  on  some  enduring misconceptions about modeling.   I hope that by  doing  so, I will give heart to  aspiring modelers, and give pause to misguided critics.

 Why Model?

 The  first  question  that  arises  frequently–sometimes  innocently  and  sometimes  not–is simply,  “Why  model?”   Imagining  a  rhetorical  (non-innocent)  inquisitor,  my  favorite retort is, “You are a modeler.”  Anyone who ventures a projection, or  imagines how a social dynamic–an epidemic, war, or migration–would unfold is running some model.

But typically, it is an implicit model in which  the assumptions are hidden, their internal consistency is untested, their logical consequences are unknown, and their relation to data is unknown. But, when you close your eyes and imagine an epidemic spreading, or any other social dynamic, you  are running some model or other.  It is just an implicit model that you haven’t written down.

This being the case, I am always amused when these same people challenge me with the question, “Can you validate your model?”  The appropriate retort, of course, is, “Can you validate yours?”  At least I can write mine down so that it can, in principle, be calibrated to  data,  if  that  is  what  you  mean  by  “validate,”  a  term  I  assiduously  avoid  (good Popperian that I am). The choice, then, is not whether  to  build models; it’s whether to build  explicit ones. In explicit models, assumptions are laid  out in  detail, so we can  study  exactly  what the entail. On these assumptions, this sort of thing happens.  When you alter the assumptions that is what happens. By writing explicit models, you  let others replicate your results. You can in fact calibrate to historical cases if there are data, and can test against current data to the extent that exists. And, importantly, you can incorporate the best domain (e.g.,biomedical, ethnographic) expertise in  a rigorous way. Indeed, models can be the focal points of teams involving experts from many disciplines.

  ∗  Senior Fellow in Economic Studies and Director of the Center on Social and Economic Dynamics, the Bookings Institution, and External Professor, The Santa Fe Institute.   I thank Ross A. Hammond for insightful comments.

Another advantage of explicit models is the feasibility of sensitivity analysis.   One can sweep a huge range of parameters over a vast range of possible scenarios to identify the most  salient  uncertainties, regions of  robustness, and  important  thresholds. I  don’t see how to do that with an implicit  mental model.  It is important to note that in the policy sphere (if not in particle physics) models do not obviate the need for judgment. However, by revealing tradeoffs, uncertainties, and sensitivities, models can discipline the dialogue about options and make unavoidable judgments more considered.

Can You Predict?

 No sooner are these points granted than the next question inevitably arises: “But can you predict?” For some reason, the moment you posit a model, prediction–as in a crystal ball that can  tell the future–is  reflexively presumed  to  be your goal.  Of course, prediction might  be  a  goal,  and  it  might  well  be  feasible,  particularly  if  one  admits  statistical prediction in which stationary distributions (of wealth or epidemic sizes, for instance) are the regularities of  interest.   I’m  sure  that before Newton, people  would  have said  “the orbits of the planets will never be predicted.”  I don’t see how macroscopic prediction–pacem Heisenberg–can be definitively and eternally precluded.

Sixteen Reasons Other Than Prediction to Build Models

 But, more to the point, I can quickly think of 16 reasons other than prediction (at least in this bald sense) to build a model.  In the space afforded, I cannot discuss all of these, and some  have  been  treated  en  pass-ant  above.   But,  off  the  top  of  my  head,  and  in  no particular order, such modeling goals include:

 1. Explain (very distinct from predict)

2. Guide data collection

3. Illuminate core dynamics

4. Suggest dynamical analogies

5. Discover new questions

6. Promote a scientific habit of mind

7. Bound (bracket) outcomes to plausible ranges

8. Illuminate core uncertainties.

9. Offer crisis options in near-real time

10. Demonstrate tradeoffs / suggest efficiencies

11. Challenge the robustness of prevailing theory through perturbations

12. Expose prevailing wisdom as incompatible with available data

13. Train practitioners

14. Discipline the policy dialogue

15. Educate the general public

16. Reveal the apparently simple (complex) to be complex (simple)

Explanation Does Not Imply Prediction

 One  crucial  distinction  is between  explain  and  predict. Plate  tectonics surely  explains earthquakes, but does not permit  us to  predict the  time and  place  of their occurrence. Electrostatics explains lightning, but we cannot predict when or where the next bolt will strike.   In all  but certain  (regrettably  consequential) quarters, evolution  is accepted  as explaining  speciation, but we  cannot  even predict  next year’s flu  strain.   In  the  social sciences,  I  have  tried  to  articulate  and  to  demonstrate  an  approach  I  call  generative explanation, in  which  macroscopic explanation–large  scale  regularities such  as  wealth distributions, spatial settlement patterns, or epidemic dynamics–emerge in populations of heterogeneous software individuals (agents) interacting locally under plausible behavioral rules1.  For  example,  the  computational  reconstruction  of  an  ancient  civilization  (the Anasazi) has been accomplished by this agent-based approach2.  I consider this model to be explanatory, but I would not insist that it is predictive on that account.  This work was data-driven.   But I don’t think that is necessary.

To Guide Data Collection

 On   this  point,  many   non-modelers,  and  indeed   many  modelers,  harbor  a  naive inductive that might be paraphrased  as  follows: ‘Science proceeds from observation, and then models are constructed to  ‘account for’ the data.’  The social science rendition–with which I am most familiar–would be that one first collects lots of data and then runs regressions on it.  This can be very  productive, but it is not the rule in science, where theory  often  precedes  data  collection.   Maxwell’s  electromagnetic  theory  is  a  prime example.  From his equations the existence of radio waves was deduced. Only then were they sought…and found!  General relativity predicted the deflection of  light by gravity, which was only later confirmed by experiment.  Without models, in other words, it is not always clear what data to collect!

illuminate Core Dynamics: All the Best Models are Wrong!

 Simple models can be invaluable without being “right,” in an engineering sense. Indeed, by  such  lights, all the best models are wrong.  But they  are fruitfully  wrong. They are illuminating abstractions.  I think it was Picasso who said, “Art is a lie that helps us see the  truth.”  So  it is  with  many  simple  beautiful  models:  the  Lotka-Volterra  ecosystem model, Hooke’s Law, or the Kermack-McKendrick epidemic equations. They continue to form the conceptual foundations of their respective fields.  They are universally taught: mature  practitioners,  knowing  full-well  the  models’  approximate  nature,  nonetheless entrust to  them the formation  of  the student’s most  basic  intuitions.  And  this  because they capture qualitative behaviors of overarching interest, such as predator-prey  cycles, or the nonlinear threshold nature of epidemics and the notion of herd immunity.  Again, the issue isn’t idealization–all models are idealizations.  The issue is whether the model offers a fertile idealization. As George Box famously put it, “All models are wrong, but some are useful.”

1See Joshua M. Epstein, 2006.  GenerativeSocialScience:StudiesinAgent-Based Computational

Modeling (Princeton University Press) and the review: Philip Ball, “Social Science Goes Virtual”

Nature,Vol448/9 August2007.

2  See Axtell, RL, JM Epstein, JS Dean, GJ Gumerman, AC Swedlund, JHarberger, S Chakravarty,

RHammond, JParker, and M Parker, “Population Growth and Collapse in a Multi-Agent Model of the

Kayenta Anasazi in Long House Valley).  Proceedings oftheNationalAcademyof Sciences,Colloquium

99(3): 7275-7279, and the review: Jared M. Diamond, “Life with the Artificial Anasazi,” Nature419: 567-


From Ignorant Militancy to Militant Ignorance

 To me, however, the most important contribution of the modeling enterprise–as distinct from any particular model, or modeling technique–is that it enforces a scientific habit of mind, which I would characterize as one of militant ignorance–an iron commitment to “I don’t know.”  That all scientific knowledge is uncertain, contingent, subject  to revision, and falsifiable in principle.  (This, of course, does not mean readily  falsified. It  means that one can in principle specify observations that, if made, would falsify it).  One does not  base  beliefs  on  authority, but  ultimately  on  evidence.   This, of  course,  is a  very dangerous idea.  It levels the playing field, and permits the lowliest peasant to challenge the most exalted ruler–obviously an intolerable risk. This is why science, as a mode of inquiry, is fundamentally antithetical to all monolithic intellectual systems.  In a beautiful essay5, Feynman talks about the hard-won  “freedom to  doubt.”   It was  born  of  a  long and  brutal struggle, and  is essential to  a functioning democracy. Intellectuals have a solemn duty to doubt, and to teach doubt. Education, in its  truest  sense, is not  about “a  saleable skill  set.”   It’s  about freedom, from inherited prejudice and  argument by  authority.  This is the deepest contribution  of the modeling enterprise.  It enforces habits of mind essential to freedom.

5 Richard P. Feynman, “The Value of Science.” In Feynman, R. P. 1999.  The Pleasure of Finding Things

Out.  Perseus Publishing





Leave a Reply

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