The meaning of “pessimal”

January 7, 2013  |  Edward Tufte
3 Comment(s)

The following appears in The New York Times story on digital globes by Mark Vanhoenacker:

For digital globe engineers, the holy grail remains a spherical computer screen. Edward R. Tufte, the author of “The Visual Display of Quantitative Information,” is enthusiastic about the potential of digital globes to remind us of earth’s offline realities — “by forgetting about the 3D whole Earth, flatland economic optimizing leads to global pessimizing” — as well as the possibility that a company like Apple will someday soon roll out a Retina-caliber spherical display.

An inquiring emailer asked about “pessimizing.” My response:

Pessimal is a word my housemate (Ludwell Sibley) and I concocted around 1963 at Stanford; it is the opposite of optimal. It was my response to economisting (see my Beautiful Evidence on economisting) theories that claim some sort of optimal outcomes (usually in a narrow, theoretical sense). That, in turn, led to the mocking phrase “Pareto pessimal” to make fun at optimality claims, probably as a response to reading Buchanan and Tullock. Similar play to encourage a bit of skepticism about statistical estimation theory: instead of “Least Squares Estimators,” “Most Squares Estimators.”

It is also my story about how the Earth ends: that local optimizing (local maximization of local interests) eventually adds up, via a massive accumulation of externalities and unanticipated consequences, to global pessimizing. Global warming might work along those lines. Or when too many Dear Leaders go thermonuclear. Or perhaps some hints in the recent financial crash, as fiercely local optimizers aggregated to a big screw-up.

More grandly: locally competitive evolution and locally competitive markets, which, while producing locally optimal outcomes (at least in our theory-dreams), lead to globally lousy results. Seeds of their own destruction and so on. This is also an answer to the Fermi Paradox “Where are they?”, where they = extraterrestrial intelligence. Thus the Search for Extraterrestrial Intelligence (SETI) has been unsuccessful because ETI’s have blown themselves up before they were able to spread, conquer, or communicate successfully with other intelligence beings. So it goes.

These thoughts partly grow from maybe the best paper ever in social science, the sociologist
Robert Merton’s “The Unanticipated Consequences of Purposive Social Action” (by Robert Merton the father, not the economist son).

I was thrilled that pessimizing got into the Times, which now makes it officially a word, defined easily by being in parallel with optimizing. I sent the quote in an email to the reporter (probably after an interview) so as to get it exactly the way I wanted to say it, and the Times quoted it accurately.

Other words of mine: chartjunk, data-ink, data-ink-ratio, non-data-ink, sparkline, slopegraph, rugplots, ghostgrid (graph paper), economisting. At Princeton years ago, I taught a statistical graphics seminar with John Tukey, who made up his own very extensive private vocabulary (he coined the words bit and software) and was said to speak in Tukish.

Topics: E.T., Science
Comments
  • Brian Olson says:

    Can ‘pessimizing’ have moral connotations? If someone is optimizing towards evil ends, are they pessimizing? Specifically I’m thinking of gerrymandering, where a lot of work has gone into optimizing district maps for partisan gain or incumbent protection.

    ET response: Partisan gerrymandering seeks to optimize partisan values and thus compromises, perhaps even pessimizes, equity values. But to perserve the going-to-the-limit idea of

    optimal

    and

    pessimal,

    I have included “seeks to optimize” and “compromises, perhaps even pessimizes.” Obviously one person’s optimizing can be another’s pessimizing, but that’s just a bit flip. As you suggest, the tone of optimal is toward a favorable claim, and pessimal not. Unless used ironically, as in Pareto pessimal or most squares estimators.

  • Andrew Pollack says:

    I was thrilled to see the article, and especially so after seeing your quote.

    Regarding your assertion that “flatland economic optimizing leads to global pessimizing”… Certainly the local focus (whether geographic or institutional) is bound to lead to Prof. Merton’s unanticipated consequences.

    But whether the 2, and not 3, dimensional representation is to blame, I’m not so sure.

    Consider this: Wassily Leontief’s input/output tables — totally 2D and flat — nonetheless attempt to portray a complete picture of an economy’s flows.

    Be that as it may: Imagine such an I/O table converted to 3D and displayed on one of these globes. Now picture a group of representatives from each economic sector projecting a display of their industries’ needed inputs and projected outputs onto the globe: picture trucks whizzing across it, factories being built (and torn down), hospitals rearranging their clinic sizes and locations — and the globe’s computer continuously updating the picture as a whole with each new input, the global (pun intended) output from which are then adjusted in an iterative fashion.

    And as all this goes on, a program running in parallel displays the changing climatic impact…
    The stuff of science fiction? Perhaps not for long.

    ET comment: I was using flatland in its original Abbott sense here to mean not display surfaces but locals or flatlanders (for whom the Earth is flat) as opposed to cosmopolitans (for whom the Earth is a common sphere for all of us), again to make a Mertonian distinction. I like you have some hope, but how can hope depend on technology outpacing and curing (without its own unintended consequences) the unintended consequences and externalities resulting from burning soft coal in Ohio and 10,000 similar, the dispersion of nuclear weapons, and catastrophic modeling failures as in the financial casino? My life window is both optimistic and not: I hope to live long enough to see the discovery of extraterrestrial life (and perhaps having beachfront property at my farm at 600-feet elevation in upland Connecticut thanks to global warming!), but not long enough to see the end of the Earth. The end, because selfish optimizing (or even just satisficing) will forever remain local and will never be able to go global.

  • David Brower says:

    The 1984 paper, Pessimal Algorithms and Simplexity Analysis, (http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.116.9158) offers an alternate definition of pessimal. The result is not net-negative as per-Tufte, but is positive in the least possible way.

    “… we must look for an algorithm that does indeed progress steadily towards its stated goal even though it may have very little enthusiasm for (or even a manifest aversion to) actually getting there.”

    cheers,

    -dB

    ET comment: That is wonderful, “positive in the least possible way.”

Contribute

Leave a Reply

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