Showing posts with label models. Show all posts
Showing posts with label models. Show all posts

Wednesday, April 11, 2012

Charles Tilly's Poetry, and the Use of Models in the Social Sciences

In a 1991 review essay discussing several books by James C. Scott, the late Charles Tilly gets cranky with rationalistic explanations of behaviour, in verse ("somehow I find the point easier to make in verse"):
Rationalists  imagine  life  as lightning chess, 
each  individual respecting well-known  rules, 
instantly sketching alternative scenarios 
that start from possible  moves,  comparing 
their merits, and choosing  the wisest means 
to  maximize the  probability of  victory, 
only to see  that the  opponent  is playing 
the selfsame  game. In such  a  caricature, 
few of  us  can  recognize  the  improvisations 
of  our  cluttered lives. Are  we  inadequate? 

Let's find  a betaphore,  a better metaphore, 
for the  expedients  by which we  rush through 
the  traps and troubles of  an ordinary day. 
We  resemble  kayakers, knowing the  long stream 
in which we  ride, more  or  less,  but never sure 
when  its course will bend, speed  up,  narrow, 
or thrust us upon  rocks and splintered trees 
savage storms have hurled into  the streambed. 

In fast white water we  often  cannot  tell 
whether we will founder, flip over,  crash 
into  the  bank, or  hurtle against obstacles 
within the  current. When  the  river slows, 
we  become  sentient  driftwood, silently gliding 
at the  pace  of swans. Or we  dig our  paddles 
into  the  depth -  one  side,  then the  other 
in order to  propel  our own  course  past fish, 
past tributary brooks, past fields of  flowers. 

The  riverbanks exist, our boat  exists, we  exist, 
the  current's force  exists, the  boulders  exist, 
and yet we  improvise, combining these  elements 
in  quick inventions, and sometimes  run aground. 

Still, life  as a surging kayak ignores the  fact 
that makes the  race worth  running: the sociability 
that ties  us to  other  humans and their poor  proxies 
dogs,  cats,  and faded  photographs - in strands 
of  gold,  or silk, or steel,  or yet  barbed wire. 

Metaphor  gives way to  metonymy, for our true model 
could  be walking through crowds, alone  or  in pairs, 
silent  or  in  earnest  conversation, at once  scanning 
faces  and facades, feet  moving in two-four beats, 
signaling our  approach with glances  and swaggers, 
knifing between  oncomers  who seem  separate or separable. 

We  follow  those  pioneers  who  find an  open  path 
across the  traffic, follow  even  when  another path 
would  be shorter, or faster, or  easier on  the  feet. 
We  spot  a penny on  the sidewalk, a  gown  in  a window, 
or  a broken hydrant while  the  interior monologue  hums, 
the  exterior dialogue swirls, the  frantic tinkering 
of  every day continues. A  cameraman above shoots  film 
and charts human beings  as molecules  in a  channel, 
lawfully accelerating and slowing according to  density, 
flowing symmetrically around those  talkers who stop 
precisely in midpath, walkers miming viscous  fluids 
whose  laws they do  not  know. Meanwhile we  pedestrians
dream, improvise, weave, stumble,  curse, above  all, hope.  
("Domination, Resistance, Compliance... Discourse," Sociological Forum 6(3), p. 602). To which my first reaction was: WTF? Also: is this the only use of a poem to make an argument in sociology or political science? Are there others? (James C. Scott apparently promised a poem in "his next review" of Tilly's work - does anybody know if the promise was kept?).

I don't have a lot to say about the quality of the poem - let's say it's better than some, but it's no Dante. (I like the  "strands of  gold,  or silk, or steel,  or yet  barbed wire" image, for what it's worth - it brings to mind other "strands of gold" images in ancient poetry and nicely reframes them). As for the points Tilly is making - roughly, that most problems of everyday life are computationally intractable, so we "satisfice," and that our intrinsic sociality affects the patterns of social action that we observe - they strike me as unobjectionable in the abstract. But I get the feeling Tilly misunderstands the purpose of models in the social sciences. (I say this with some trepidation - it is far more likely that I am wrong about this than that Tilly misunderstood anything).

As Kevin A. Clarke and David M. Primo have argued in a recent book (and article) models are maps of phenomena. (David Schmidtz has made a similar claim with respect to theories of justice, and I think both draw on the work of the philosopher Ronald Giere). And the most accurate maps are not always the most useful, though it is important that maps be sufficiently similar in some relevant respect to the phenomenon they image for them to be useful. To this I would add that models are also toys ("toy models") and analogies ("metaphors" - Tilly alludes to this in the poem above). They enable certain kinds of rule-constrained "inferential play" (as toys) and disclose or conceal connections to other phenomena (as analogies or metaphors). (They are rhetoric too, qua metaphors, and hence serve persuasive purposes). But let's stick with the map imagery right now. Consider, to use an analogy Clarke and Primo point to, two maps of the London Underground:

From Knightsbridge.net

From A Welsh View

Is any of them more "correct"? They are certainly useful for different purposes: they provide different forms of orientation with respect to the tube, and each of them has characteristic failure modes when used for other purposes (to which they may not be well suited). The first map is useful for people who are actually using the train - it helps you visualize the transfers you will need to make, as well as the approximate number of stops left until your destination. But it is not a very good guide to actual distances, and it does not provide any information regarding the urban or natural context of the stations. The second map, while being a more accurate description of the physical organization of the underground and of the urban and topographical context of each station, is much less useful to commuters, who are likely to find it too "busy." Neither of them, it is worth stressing, is a perfectly "accurate" representation of the tube, though both are "similar" to it in some significant respect, enough so that we can speak of them as "representations" of the underground.

Similarly with social science models: economists and (some) political scientists tend to prefer more "abstracted" maps of particular social contexts (like the first map above), since they tend to use such maps for purposes that would not be served by the more apparently "accurate" models that may be favored by sociologists or anthropologists or (other) political scientists (like the second map below). The major methodological disputes in the social sciences thus tend to be (sometimes covertly) about the legitimacy of the purposes for which these maps are used (and only secondarily about whether any particular map does serve any given purpose, though that sort of debate does happen too).

In the poem above, for example, Tilly seems to be saying that the "chess" map of the rationalists (which would surely include economists) is not a good map for making sense of social action because we are not like that. (It is interesting that Tilly also seems to allude to the physicist's map, which depicts social action on the analogy with fluid dynamics, whose laws the fluid particles do not understand but which also makes no reference to individual psychologies, unlike the economist's map). But the "chess" model - the model of rational agents - is not generally a description of our psychology, though it does describe a psychology that is in some respects similar to our actual psychology and in other respects dissimilar. To the extent that rationalistic models of human life are useful (and they may not be, certainly not for every conceivable explanatory purpose) they are not useful because they describe our psychology accurately (though they will be more useful the more the similarities to our psychology in the model are relevant to the explanatory purposes to which it is put), but because they may provide insight into how human action can aggregate into larger patterns (e.g., how markets can sometimes produce efficient outcomes, or how conventions can be self-enforcing), or make certain kinds of predictions (e.g., about when certain norms break down), or identify potential puzzles about social action, or even simply to point to  long-term forces pushing social systems in certain directions rather than others. The resulting picture of human action will tend to look (to the anthropologist or the sociologist) like a stick-figure drawing, but that is precisely the point, at least so long as the stick-figure drawing tells us something about human action that is difficult to see in the hyperrealistic map of the anthropologist or the somewhat broader frame of the sociologist.

Social explanation at its best is the art of selecting the right map for orienting ourselves towards some question. In some cases, that map better be quite detailed; if I am interested in getting a real feel for how people distant from me live, or how they can be motivated to rebel against injustice, I am often better off consulting the anthropologists' map than the economist's map. But in other cases, the thickly descriptive map just gets in the way of the particular type of understanding I may be seeking. Even given a certain kind of question, however, some maps will still be better than others. (Some maps give bad directions, or have inconvenient lacunae). In economics, for example, it seems that maps with explicit microfoundations are worse for predictive or policy purposes than maps without such microfoundations, though "microfounded" maps are not thereby useless. In political science, detailed understanding of the politics of particular countries is not necessarily very useful for predictive purposes, though it is certainly very useful for many other purposes. And further problems arise, of course, because fights about methods are also fights about resources and status. The  theoretical pluralism of "multiple maps for multiple purposes" tends to break down when certain mapmakers are marginalized, or when there is a perception that particular kinds of maps are being used for purposes to which they are not well suited while serving to attack the status of makers of alternative maps. Somehow I find it easier to make this point in haiku:
maps are different
but mapmakers are prickly
and love their maps best
(Ok, not a very good haiku. I'm sure you can do better.)