|
Abstract
|
Every now and again in safety science, a book comes along that forces or invites us to change our thinking. The book might start with a powerful idea--the potential for an accident lies locked up in structural properties of the system; its interactive complexity and coupling, for example--even if its basis in empirical data is a bit porous. The book might also start with the empirical data itself, but take a fresh look at that data--a look that in turn gives rise to new ideas. Andrew Townsend's elegant little new book is in the latter category. And it seems to prove its own timeliness. Toward the end of the book, we find a graph that shows the relationship between increasing safety regulation and decreasing fatalities. What is most scary about the curve is that it is plateauing. It has become asymptotic. More regulations do not save more lives. We are clogging our industries with rules, and more rules, whose marginal returns have declined below any level of visibility. This, if anything, points to dying strategies. More of the same will not lead to something different.
In Safety can't be measured, Andrew Townsend works wonders with numbers, graphs, tables and figures. Here are some examples of what this can reveal. Increases in worker safety over the past decades, for instance, seem to correlate with the regulatory and managerial energy we have thrown at the problem. But it correlates at least as strongly with economic growth and prosperity. Are we blindly and arrogantly overinterpreting the reach of our own interventions? Perhaps Aaron Wildavsky was right after all: simply that "richer is safer" (Wildavsky, 1988). This, of course, challenges the bureaucratic entrepreneurialism that now characterizes much of our safety management systems. Increasing our safety bureaucracy is ultimately self-destructive: Ever-expanding systems of compliance will eventually lead to less compliance, because the job still needs to get done. A graph in Andrew Townsend's book shows the explosive growth of statutes and regulations since 1974. It is printed in ominous black and grey--a literal swell of rules, with the deeper paradox that these can actually harm safety. The history of industrial accidents, after all, shows that the potential for accidents brews at the core of the very processes and structures set up to prevent them (Pidgeon and O'Leary, 2000). The more these processes and structures are developed or enforced bureaucratically by those who are at a distance from the operation, the greater the risk becomes that these "fantasy documents" bear no relation to actual work or actual operational expertise (Clarke and Perrow, 1996). It helps create "structural secrecy:" a by-product of the cultural, organizational, physical and psychological separation between operations and safety regulators and bureaucracies, where critical information does not cross organizational boundaries, and mechanisms for constructive interplay are lacking (Vaughan, 1996).
In another myth buster, Andrew Townsend tells us that production pressure may not be the reviled ill that many make it out to be (and that has led to canonical ideas in the literature: that production and protection are always opposites, always to be traded off against each other). Townsend's data show that increases in production lead to increases in safety--the moderating variable may well be investment. Producing more costs more (even as it generates more revenue), but it can also mean investments in new technology, more training, better equipment. Thus, the canon about production versus protection in our textbooks may well be false, or at least without any data-driven nuance or sophistication.
The title is an oxymoron, for sure. If safety can't be measured, then how can we base our improvement of safety on evidence? It must be the kind of evidence that cannot be "measured"--evidence of a different kind. Lots of sciences work off evidence that cannot be measured, of course: ideographic social sciences, large portions of clinical medicine, humanities, and much more. So this is not such a strange idea. It is, however, in a direction opposite of what the safety field has been traveling in over the past two, three decades. Driven by the dogma that it takes lots of little things that go wrong to create an accident, the obsession of safety management has become to hunt down all those little "latent" ills and plug them up before they can cause real trouble. Of course, the "real trouble," as Andrew Townsend acerbically points out, is not necessarily the accident. It is that not doing something about those little upstream wrongs creates liabilities for managers, regulators and directors. They need to be seen to be doing something about the little wrongs and latent issues, even if that boils down to telling all the workers to be a little bit more careful, to watch out harder. When things do go wrong, at least the elite can claim that it said to its workers: "Told you so." A brief look at malleable terms like "suitable and sufficient", "foreseeable," "preventable," or "reasonably practicable" gets Andrew Townsend to conclude that the victor and the powerful are able to write the precise definition and have it turn their way in each specific case. The book's very first case study is about a worker prosecuted for being involved in an accident--a chilling case in point.
Also on the chopping block for sacred cows is, of course, the very bête-noire of safety science: Heinrich's triangle. Whereas the original insurance data that went into concocting the triangle is lost to history, Andrew Townsend shows nicely how none of the data we have available today comes even close to mapping onto the triangle. In fact, the inverse seems true. Fatalities as a proportion of all recorded injuries are going up between 1990 and 2010, for example. This is probably in part because safety managers have become smarter at hiding, fudging or re-classifying injury data--practices that Andrew Townsend neutrally calls "underreporting." Yet the persistent use of Heinrich's idea in a number of industries (where it is embedded in loss-prevention systems, safety management systems, procedures and protective measures) gets Andrew Townsend to remark how:
"…ordinary members of the public, managers and working-level safety advisors may gain the impression that safety is a 'science' and conforms to immutable physical laws. Nothing could be further from the truth: the original research upon which much of today's safety management is based is at best fragmentary; at worst it is spurious. Yet the original assumptions made in the management of safety have been in existence for almost a century. They have had time to embed themselves in the minds of industry and the safety profession; to challenge them now is almost an act of heresy. … What might have started out with high ideals had become a creeping malaise founded on inadequate, incomplete, misleading research fuelled by a naive belief in punitive enforcement, political correctness, vested interest, myopia, denial and dogma that few dare challenge."
Is Andrew Townsend a man with a bone to pick? A fellow with a chip on his shoulder the size of a small European principality? I don't think so. In my reading of his book, and some of the correspondence that helped lead up to it, I can only see Andrew Townsend as a deeply concerned, deeply ethical fellow traveller. The multiple "sins" of HSE, institutionalized and legitimated in how we insure, regulate, control and contract each other in safety-critical industry, have simply become too much for him to remain silent. Hubristically mistaking the power of our own interventions; always coming down on the little guy at the end of the day; rewarding the wrong thing; counting what we can count, but largely ignoring that which really counts; setting up huge bureaucratic machineries that become expert in the "the art of measuring nothing," as a 2010 article in this journal called it (Lofquist, 2010). Those are the things that have motivated him to put his arguments together in this book. And the data, as he sees it, is screaming the same message in our face. If only we look at that data honestly, earnestly. And more data may be needed, but then we have to start asking different questions. Questions, says Andrew Townsend, like what are the effects of inadequate models of accident causation on safety regulation? His is not iconoclasm for the sake of being different or irreverent at the end of a long career in construction and safety. His is a call for us to be who we say we are: safety scientists. Go back to the evidence--isn't that what distinguishes science from faith, from religious zealotry? Sure, much of the evidence can't be measured, and that which we measure ranges at best from "fragmentary" to "spurious," as Townsend has it. But evidence there is, plenty of it. Enough, in any case, to start thinking differently about how we do safety. |