Monday, May 29, 2006

Another definition of science ...

"... a body of knowledge collected an nurtured by experts according to neutral, objective, and universal standards."
That's from an entertaining article with the staggeringly unoriginal title "The Management Myth" in the June issue of The Atlantic (which was kindly passed on to me by Brother Hrab).

Hmmm, doesn't strike me as a great definition. Who are these "experts"? And what are these "neutral, objective, and universal standards"? But most of all, I still think that science is fundamentally observational. To give some context, the author of the piece, Matthew Stewart, was discussing the historical development of "scientific management". (Full disclosure: I co-authored a paper in the journal Management Science a few years ago.) To my mind, unless there's an attempt to take careful observations, it's not science.

As an aside, I'd like to comment on an ambiguity in the word observational. Sometimes people distinguish experimental from "observational" methods. But of course observation is a component in experimentation; the real distinction is that in experimentation there is planned manipulation of conditions. Sometimes, to evade this distinction, people refer to "natural" experiments, namely observations with coincident variation in potential explanatory factors. But you can't get around the fact that these are not real experiments, and may well suffer from the usual shortcomings of non-experimental studies. And that would be my preference for terminology: experimental versus non-experimental studies. Sometimes people will insist that it's not science if it's not experimental, but this is going too far: it would rule out—among many other sciences—astrophysics and evolutionary biology. Of course there are many challenging issues in analysis of data from non-experimental studies. Consider, for example, the analysis of data from a case-control study. While this epidemiological design is indispensable for investigating rare outcomes and in cases where randomization is not ethical, the problem of confounding can bedevil analysis. Hey, science isn't always easy (certainly not as easy as the textbooks sometimes portray it).

Returning to Stewart's piece in The Atlantic, I think he's at his best skewering management fads and the associated vapid management-speak:
"On the whole ... management has been less than a boon for those who value free and meaningful speech. M.B.A.s have taken obfuscatory jargon—otherwise known as bullshit—to a level that would have made even the Scholastics blanch. As students of philosophy know, Descartes dismantled the edifice of medieval thought by writing clearly and showing that knowledge, by its nature, is intelligible, not obscure."
I end with a cartoon that, by its nature, is intelligible, not obscure:
Bookmark and Share

Wednesday, May 17, 2006

The trouble with models

Models are central to science. A scientific model is a kind of representation of physical reality, an analogy if you like. For example, the image on the left is of the "planetary model" of a nitrogen atom.

The power of analogy is that it lets us think about one thing in terms of something else—often something simpler or more familiar. An example of this is metaphor: when we say, for example, that "life is a journey", we are using the relatively simple model of a journey (with a starting point, a period of travel, a destination, etc.) to understand the more nebulous concept of life. The importance of metaphor was first revealed to me by the marvelous book Metaphors We Live By, now available in a second edition. (About 15 years ago, I picked up a copy at a used bookstore more or less by chance. It had a tremendous impact on my thinking; it was some time later that I learned that it's actually rather a famous book.) The authors, Lakoff and Johnson, argue not only that metaphors are ubiquitous, but that they in fact structure the way we think. Using what strikes me as a more general version of the same argument, Douglas Hofstadter suggests (in a very entertaining essay) that analogy may be the "core of cognition".

Analogies are indeed wonderful. But it is well to remember that no analogy is perfect, and sometimes an analogy can actually be an obstacle to understanding. The notion that "life is a journey" can be helpful, but it may lead us to overlook aspects of life that don't resemble a journey. For example it may cause us to unduly focus on trying to "get somewhere" in life. An alternative to the title Metaphors We Live By might be Metaphors We're Trapped By. Notwithstanding the fact that a metaphor can be very informative, strictly speaking, it is always a lie. Or as statistician George E. P. Box put it:
"All models are wrong; some are useful."
The trouble is, we often confuse models with reality. (Incidentally, that applies to fashion models too!) What is true about a model may not be true about what it represents. Sound familiar? It's very much like the objection I quoted in my recent post about definitions: it may not be valid "to draw conclusions about what is true about the world based on what is true about a word". (And what is language but a type of model? I believe this links rather directly to the field of semiotics, but unfortunately I don't know much about that!)

More things in heaven and earth ... than are dreamt of in your philosophy

It seems there has been an increasing recognition that models lie at the heart of science (rather than "laws"). And models are necessarily imperfect. The planetary model of the atom was soon superceded by the Bohr model, which was followed in turn by even better models. At its best, science progresses by recognizing the shortcomings of models and substituting more appropriate ones. But the process by which this is achieved remains controversial. The interplay of deductive and inductive reasoning is part of it, but it may be far more complex than that, as Thomas Kuhn argued in The Structure of Scientific Revolutions.

Whether we like it or not, we're stuck with models, not only in science but in language and indeed in the core of thought itself. In an essay on "The Fall and Rise of Development Economics", Paul Krugman has a fascinating section on "Metaphors and Models" in which he writes:
"The problem is that there is no alternative to models. We all think in simplified models, all the time. The sophisticated thing to do is not to pretend to stop, but to be self-conscious -- to be aware that your models are maps rather than reality."
Bookmark and Share

Tuesday, May 16, 2006

Pompous rhetoric


Canada's Conservative government wants to extend the stay of Canadian troops in Afghanistan another two years beyond the current February, 2007 deadline. (See reports from The Globe & Mail and CBC.) After a 6-hour debate tomorrow, members of parliament will vote, and Canadians will have to live with the consequences. Pronounced Prime Minister Stephen Harper:
"What we are doing there is not just protecting our national interests, but providing international leadership and providing real advancement to the standard of living and human rights of the Afghan people."
Do these claims stand up? First, how is the Canadian military presence in Afghanistan protecting our national interests? No doubt there will be some grand words in the House of Commons tomorrow, but I'd like to hear a cogent argument, not just hot air. A national child-care plan would be in our national interest, but the Conservatives won't hear of that.

Next, does a foreign military adventure demonstrate international leadership? Perhaps we're supposed to believe this simply because the words "military" and "leadership" happen to be in the same sentence. Living up to our Kyoto commitments to reduce greenhouse-gas emissions would show real leadership, but instead the Conservative government wants to back out of that agreement.

Finally, is the Canadian military presence in Afghanistan advancing the standard of living and the human rights of the Afghan people? Perhaps, but what's the evidence? And could we achieve more by different means? I believe it would be much better for Canada to provide financial support and diplomatic interventions to nurture real democratic progress in Afghanistan.

I wonder if the Conservatives have anything more than pompous rhetoric to support this military adventure?
Bookmark and Share

Sunday, May 14, 2006

A defining moment


Definitions are wonderful and terrible. Wonderful because carefully chosen definitions bring thought into focus. But terrible because conflicting definitions are the source of endless misunderstandings. Try to define something carefully and it won't be long before someone suggests that "it's just semantics!" Translation: you're wasting your time on words instead of what really matters.

While I agree that it's invalid "to draw conclusions about what is true about the world based on what is true about a word" (I'm quoting from the Wikipedia page on semantics), that hardly makes semantics irrelevant! Semantics is, after all, the study of meaning. I don't imagine anyone has ever objected that "it's just meaning!"

The trouble is, some words carry around a lot of baggage. As I've just indicated, an unfortunate example is semantics itself! Lately, I've been thinking a lot about evidence. Not easy to define, but everybody seems to think it's a good thing. Similarly comments apply to science and research, both of which are favorites of advertisers—a sure sign that these words evoke powerful responses.

A couple of naive approaches to definitions deserve mention. One is simply to resort to a dictionary. This raises the (oddly recursive) question of whether dictionaries are or should be descriptive or normative. The fact that multiple definitions are often given for a word suggests a descriptive role. But I think that at the same time they try to be authoritative. The fact that different dictionaries sometimes offer strikingly different definitions suggests that this is an impossible task. Language is fluid in part because thought is fluid. Evolving concepts that are abstract and controversial don't submit to anyone's putative authority. So while dictionary definitions do provide grist for the mill, they're hardly definitive the last word beyond criticism.

In my most recent post on evidence, I cited the Oxford English Dictionary, focusing not only on the definitions offered, but also on the etymology of the word. Word origins can inform a definition, but it's naive to rely on them exclusively. Science, for example, derives from the Latin scientia, meaning knowledge. Yet science surely means something different from simply knowledge.

For example, I know a number of quotes from Shakespeare's plays, but that kind of knowledge isn't science. The Wikipedia entry on science notes that science is often used informally to mean "any systematic field of study or the knowledge gained from it". For the sake of clarity let's call "the knowledge gained from it" scientific knowledge. But when we say science, do we really mean "any systematic field of study"? I would prefer to use the word research for that purpose. For example, I don't consider theology to be a science, but have no problem with the term "theological research".

Science, as I understand it, is fundamentally observational. I've made the same point about evidence, and I think the two are clearly related (but more of that in another post). An interesting test case is "computer science", sometimes known as "computing science" since its focus is really on computing more so than computers ("Computer science is no more about computers than astronomy is about telescopes." - Edsger Dijkstra). But is it a science? Well, the scientific method can be applied in computing science, in the design and analysis of empirical studies, but most of computing science is about mathematics.

And mathematics, while strongly associated with science, isn't a science. I believe that science is a good thing, but not all good things are science. I also believe that science gets at truth (and that's a challenging notion), but it's not the only thing that gets at truth. That the Earth revolves around the Sun is a truth established scientifically. Pythagoras' theorem in Euclidean geometry is a truth established mathematically. Both are objective truths in that anyone can, in principle, verify them. I suspect that there are also subjective truths that may be established using entirely different methods. The relationships between these different truths and the methods used to establish them strike me as profoundly interesting.

I still haven't offered a definition of science, and in fact I suspect that ultimately what may be most important is the pursuit of a definition rather than a definition per se. I would provisionally define science as the practice of the scientific method—clearly a bit of a dodge! The Wikipedia entries on science, the scientific method, and the philosophy of science provide useful background, but no clear consensus. I find myself in agreement of much of what is written in those entries, but not all of it. (For example, Karl Popper's perspective on science is prominent, but I have my doubts.)

Science, technology, and evidence-based medicine

Notably, discussions of science often focus exclusively on natural science. But the essence of science is careful observation and reasoning from observation, which clearly can be applied to any observable phenomena. This obviously includes human behaviour, notwithstanding the formidable obstacles that arise. The scientific method can also be applied in evaluating technology. Evidence-based medicine is an example of this. While natural science is of central importance in guiding the development of healthcare technologies—be they drugs, devices, surgical techniques, what have you—the evaluation of their performance is not a matter of natural science. The application of good science may lead to a technology that turns out not to work, or to have unanticipated adverse effects that preclude its adoption. Conversely, useful technology need not result from the application of science. Many drugs, for example, have come to us by non-scientific routes. Ultimately, regardless of their origins, healthcare technologies need to be evaluated to determine how well they work, and what adverse effects they may have. Why they work the way they do is of secondary importance.
Bookmark and Share

Tuesday, May 09, 2006

Cool optical illusions

I'm a big fan of optical illusions. And I just found a blog devoted to them, from which I copied the remarkable image shown here. I got there from Blogs of Note, which features daily links to "Interesting and noteworthy Blogger-powered blogs, compiled by the Blogger Team." Their archives go back more than 5 years—great for semi-random web surfing!

By the way, for more cool optical illusions, I'd recommend the book Incredible Visual Illusions by Al Seckel.
Bookmark and Share

Monday, May 08, 2006

Disclaimer

The views and opinions I express on this blog are mine, and not those of my employers, clients, family, or friends, past or present. This blog is in no way affiliated with my employers, clients, family, or friends, past or present.

The information and content on this blog are provided "as is" with no warranty of any kind, either express or implied.

Links

I cannot guarantee that webpages I link to will work all of the time and I have no control over the content of linked pages. I am not responsible for the contents of any linked websites and do not necessarily endorse the views expressed on them. The fact that I link to a given website does not mean that I endorse the contents of that website.

Comments

I welcome comments, but reserve the right to delete comments as I see fit. Comments represent the views and opinions of those who post them, and I do not necessarily endorse these views and opinions.
Bookmark and Share

Sunday, May 07, 2006

Miscellany

It's Sunday evening, and rather than coming up with something thoughtful myself, here are some miscellaneous items of interest. A cornucopia if you will, as illustrated at left.

First, something important—an urgent appeal from Amnesty International Canada:
On May 9th the U.N. General Assembly will elect the members of the new UN Human Rights Council.

Canada, which is standing for election, has a critical role to play in ensuring that the Council's mandate of protecting all human rights for everyone is not a hollow promise.

Send an email to Canada's Minister of Foreign Affairs, the Honorable Peter MacKay, and urge him to ensure Canada demonstrates clear leadership in the respect and fulfillment of human rights norms and standards.

Take action before May 9th!
If you're in a science-and-public-communication mood, this post is worth reading.

Next, something fun (via Antonella Pavese's interesting blog): a very entertaining manifesto on how to be creative by advertising executive and blogger Hugh MacLeod.

Incidentally, Antonella Pavese recommends this cleverly-named, free, online, task manager: remember the milk. Ok, I haven't actually tried it, but their symbol is just too cute.
Bookmark and Share

Wednesday, May 03, 2006

A view of evidence

In the last month I've discussed various aspects of the evidence-based debate. I'd now like to present some of my ideas on the subject. As I've suggested previously, the definition of evidence seems to lie at the heart of the dispute. But before I attempt to define evidence, I think it's useful to consider the word itself.

Evidence is the noun form of the adjective evident, ultimately deriving from the Latin evidens from ex- (out, forth) + videre (to see). The Oxford English Dictionary gives the following definitions for evident: "1. a. Conspicuous b. Obvious to the sight. 2. Clear to the understanding or the judgement; obvious, plain. 3. Indubitable, certain, conclusive. —1653." And the primary definition of evidence is simply "The quality or condition of being evident."

So a very literal interpretation would take evidence to be what can be seen, or "shown forth". It seems natural to extend this beyond vision to sense perceptions in general. Further metaphorical extensions bring us to the notion of clarity, certainty, and conclusiveness. And already the problem becomes apparent: we have moved from sense perceptions to things that might be considered similarly certain and conclusive. Like what? Deductive arguments? Accepted theory? Long experience? Expert opinions? Religious precepts?

Let's consider each of these in turn, starting with deductive arguments. A valid deductive argument is water-tight—provided that its premises are true. But how do we establish them? Perhaps we can depend on accepted theory. But history is littered with theories that were once universally accepted, but are now discredited or superseded. Of course there's always long experience. Experience is the cumulative product of personal practice and observation. But it is notoriously subject to selection bias (and perhaps other biases too). And when it comes to rare events, no amount of experience is sufficient. Can expert opinion step into the breach? The opinion of an expert represents a synthesis of many different sources of information, usually carried out over many years. While there may be many good reasons to trust an expert (such as his or her qualifications, intelligence, experience, and good standing in the community), well-meaning experts have been spectacularly wrong any number of times. Ultimately trusting expert opinion is an act of faith. Which nicely brings us to religious precepts as a source of certainty. These are perhaps the original "self-evident" truths—that is, to the believer, but perhaps to nobody else.

It is, of course, true that sense perceptions can be misleading too. A classic example is the straight stick that appears bent when placed in a glass of water due to the refraction of light. Any number of other illusions and hallucinations make interpretation of sense perceptions a thorny philosophical problem—one of the fundamental challenges in epistemology. Without wishing to minimize these, I will set them aside, by simply asserting that we know the external world through our sense perceptions. They are the only raw materials at our disposal, and they are the closest we can get to certainty about the external world. Evidence is observational.

But that doesn't mean that observation is all there is to evidence. For instance, a scientific experiment involves action (manipulating or controlling conditions) as well as observation. But if there's no observation, there's no evidence. To consider a particular example, this means that opinion doesn't count as evidence, but observation of opinion does! That is:
  • Opinion about X isn't evidence about X.
But
  • Observation of opinion about X is evidence about opinion about X.
I certainly don't mean to imply that opinion is worthless—that would be a bit self-defeating, now wouldn't it?—but it just can't be counted as evidence. In any case, regardless of what I or anyone else has to say, opinion will always guide us.

Ultimately, I think that opinion is most convincing when it is backed up—whether by deductive argument, theory, experience, evidence, or some combination of these. And I hope that my opinions have some of these supports.

While I haven't yet given a definition of evidence, I have presented what I think is a crucial qualification: evidence is observational. But there's a lot more to it than that!
Bookmark and Share