The Question You Avoid Asking Is The Public Relations Coup

The recent salmonella outbreak has sparked a "spirited" debate on one of those NYT bullshit blogging heads deals. Here's the problem: not one of the experts in their invited comments talks about the structural problems with factory food, and that a complex system with lengthy supply chains and distribution links will always fail in a geographically large scale manner. Not one blurb about the problems with capital markets and foodways with derivative trading on everything from salt to eggs as a possibility. Not one mention of the MBA model of cost/benefit food analysis where a few hundred dead folks (and a few million dead chickens) means a marginal risk at best. Not a single meaningful criticism of the very notion of an "egg industry." The fight for food security is a local fight and I am glad to see one of my alma maters, UTK, sparking interest in local and market gardening among its students, even if revolves around a boutique food. Lord, what's next? A tomato that hasn't traveled three thousand miles?

Comments

[Hayduke]

Efficient is the opposite of robust. We've been working on efficiency for a long time in our food system and fault tolerance was a natural casualty. I don't know what it would take to get Americans to start valuing reliability over price, or if it's even possible without sustained pattern of disasters and shortages. In his Long Now Foundation talk, Social Collapse Best Practices (http://longnow.org/seminars/02009/feb/13/social-collapse-best-practices/), Dmitry Orlov described how when the inefficient (and unreliable) Soviet system collapsed, people already had contingency plans and arrangements with nearby farmers who were already providing about half their food. To the extent we still have a competitive market environment, we might be a bit better off, but I wonder how many places that supply web narrows to a cutoff point caused by a natural monopoly.

[lobbygow]

Efficient is the opposite of robust.

I agree with the spirit of that assertion, but not the letter.

I guess it depends on how you define "robust," but I've been consulting on two dimensions of system performance for years: effectiveness and efficiency. I define "effective" as the degree to which a system (process, organization, organism) meets its goals (profit, safety, reliability, etc.) and efficiency as how effective a system is per unit resource consumed (labor, materials, energy, etc.). These dimensions are not inherently in conflict. The problem is one of focus. If only efficiency is focused on, effectiveness tends to diminish because there is no feedback on that dimension. No feedback, no adaptation or improvement. No improvement in effectiveness, no future.

Now, if "robust" means "free of fault," then I would say that from a system standpoint it's fairly rare for all but the simplest of systems. In manufacturing, you can refine the system to be virtually fault free, but that's because 1) The goals are specific, simple and static (i.e. this widget must measure exactly 1.75 mm) and 2)Feedback is abundant and the system of variables influencing the result is relatively simple.

So I guess I would rephrase your first statement as "a continued focus on efficiency alone suboptimizes the system, perhaps fatally." And by "alone," I mean that's where the majority of the attention, feedback and rewards are centered.

Another problem I wrestle with is asynchronous nature of consequences for being ineffective or inefficient. In nature, effectiveness is nearly always the short term focus, regardless or resource utilization. I either get food or I don't. I either find a mate or I don't. I either escape a predator or I don't. You can't even get to the point of worrying about how much energy you spend being successful until you are successful. Therefore, effectiveness is the dominant short term feedback and thus drives system behavior more strongly than efficiency. Being inefficient can ultimately be deadly, but being ineffective is automatically deadly.

This is true for business too, but the managers usually don't manage that way when it's a publicly held corporation. Wall Street replaces the natural selection of the consumer marketplace, and decisions are driven by variables that influence stock prices, not consumer demand or overall business health. I hear "shareholder" nearly three times more than I hear "customer," especially when I'm dealing with corporate management, who are completely insulated from the real-world behavior of consumers and their product/service. This is why I'm so skeptical about "big business" management. They only manage what they can "see," and what they can see isn't the right stuff.

How did I get off on that tangent?

[Hayduke]

I'm thinking of "robust" in terms of being able to absorb and deal with changes in the operating environment and contrasting that objective with Thomas Princen's observation that an "efficient" system is necessarily optimized for a limited number of parameters. As a system becomes efficient it becomes simultaneously less sensitive and more susceptible to the ignored variables. Imagine a mortgage industry optimized for profit, but ignoring the limits of actual non-speculative housing demand.

I hadn't really thought about effectiveness, but I'm not sure I buy the boolean nature of natural outcomes. There's a continuum of satiation (I'm always hungry and do nothing but search for food --> I'm done eating, so let's do something else) and procreation (we brought one child to maturity after years of failure --> I cannot count the fruit of my loins). But of course you're right below a certain level of success.

[lobbygow]

I'm thinking of "robust" in terms of being able to absorb and deal with changes in the operating environment

OK, I had forgotten that use of the term from my ecology days. I've taken to calling that "adaptiveness," but robust is less clumsy. The simplest form of this in my work would be scalability -- i.e. the ability of a system to maintain effectiveness (and to a lesser extent, efficiency) in the face of fluctuating demand on that system. It's an elusive goal given how most people approach design.

As a system becomes efficient it becomes simultaneously less sensitive and more susceptible to the ignored variables
That's essentially the point I was making about focus and feedback. Unfortunately, most of us think first about the operationally important structural elements of the system, and how adaptive that structure is to changes in the operating environment. There are whole conferences devoted to the concept of "enterprise architecture" in my field, which causes no end of eye-rolling on my part. I am something of a heretic in that I don't believe the structure of the operating system is nearly as important as the quality of the feedback system, which is about as easy to separate from the operating component as mind from body, but you get the drift. In my personal experience with processes and organization structures, the better feedback system almost always wins regardless of how the system itself is organized to accomplish its work.

One more thing -- I don't think I believe that there are goal directed systems where "efficiency" is the goal. By my definition, the raison d'être ( I suppose it could be raisons d'être) of a system are what defines the degree of effectiveness (there's your Boolean point). Efficiency is merely an attribute of ANY system that can only be understood once you know what the definition for effectiveness is. When someone says "Brent is very efficient," that may be viewed as a good thing, but my first question is "efficient at what? Serial killing? Brain surgery? Blogging?)

I think the goal of profitability is often misunderstood as pursuit of efficiency, because reducing resource utilization is one obvious way to increase profit. In fact, I would say our food distribution network is extremely inefficient once you take the dreaded externalities into account. It's not efficient at all when you look at total resources consumed. The Chief Operating Officer of a large agribusiness firm might believe she has made operations more efficient, but that is most probably an artifact of how the metrics have been defined -- in other words, the feedback has ignored variables by design. I'm not just talking exclusive focus on one or two metrics. I'm talking about how the metric itself is defined. "Unit Cost" is easy to measure because it typically only includes the costs within the control of the person measuring it. You can't measure everything, but subtle changes in the information included in feedback can have huge results on system performance without changing the underlying components or their obvious relationships very much at all. Of course, the interesting stuff is happening in the relationships we cannot see or even imagine.

'm not sure I buy the boolean nature of natural outcomes.
Not sure? Either you buy it or you don't.

All models are wrong, some models are useful.

Boolean was a convenient way to express how effectiveness works compared to efficiency, but it's not how it really works. While it's true that you can't "almost" escape a predator or "almost" have viable offspring, feeding and many other goals can be measured along a continuum. Some meals offer more nourishment than others, etc.

Thomas Princen's observation that an "efficient" system is necessarily optimized for a limited number of parameters

You can't be Segovia AND B.B. King.

I do think it's true in general that the more you "perfect" something, the less room you have to change when the definition of "effectiveness" changes - although that implies you are perfecting the structural elements of the system itself and not just the feedback.

Knowing when to perfect vs. when to change altogether is an important decision. It's the kind of choice that requires real judgment by business managers, especially in the corporate setting. Of course, some products and services themselves are so robust in terms of appeal, that it's hard to imagine much adaptation once you've hit on the right formula. Many of them start with "b" - barbecue, beer, blowj... but I digress.

[CAFKIA]

I would probably use the phrase "fault tolerant" when thinking/speaking of robustness.

[lobbygow]

CAFKIA, I like "fault tolerant," but it's not the whole picture. I think "robust" also applies to the ability to be effective in varying environmental circumstances -- which I wouldn't categorize as fault tolerant per se.

I think "ability to persist," is another way of thinking about it. That covers fault tolerance and effectiveness.

Part of my bias comes from an ecological background. I find myself always thinking of living systems, which may be a special case. Perhaps a system that has no apparent purpose can persist forever -- in fact there is no reason why that should not be the case. So, really the system just has to persist -- a hairy question in and of itself when speaking of a dynamic entity. Can the identity of a system change once it changes sufficiently? Seems that question can only be answered by a subjective observer. When is a hurricane formed and when does it die? Who makes up those rules? How do we know when a system ain't what it used to be or ain't anything at all? Tough call, it would seem, but we make it every day.

[metulj]

How about "resilient?"

[Lobbygow]

"Resilient" is the best so fat. Robust implies some sort of a priori judgement, whereas resilient is very definitely about demonstrated performance over time.

Thanks.