You wouldn’t think that books discussing agronomics would have much to say relevant to Organizational Structure, IT Management or Knowledge Management. You’d be wrong, of course, but you can see how people would think that. I’d like to show how some of the ideas being debated in the agricultural industry’s fringes can illuminate our own issues.
James C. Scott, in Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed (The Institution for Social and Policy St), writes about the “scientific forestry” movement of Germany, where they clearcut the old growth forests, with their unpredictable wood output per acre, and planted rational forests of a single tree type, row upon row. I’ve seen this tree farms in Belgium: they are remarkable in their being so perverse. We went tramping through the woods two days after I got there with some friends of L’s. Other than getting to see Chimay and having my first Chimay Bleu, the weekend was a waste, partly because the trees were so darn boring. I’m colorblind and trees for the most part all look what you would call brown to me so having some patterns to make things visually chaotic is my only respite from nature’s ugliness. These perfectly ordered trees made me sick.
Monocultures such as the forest farms of Europe seem like a great idea to those who have to manage them. It seems that they will produce a standard crop in a standard amount of time, without all of that ups and downs so often produced in the “natural” world. Unfortunately, they also are prime candidates for tree felling (where an entire forest falls over in high wind), total devastation by disease or pests, and soil destruction. Something about the way the natural system develops and adapts (evolves, really) in place produces a more resilient ecosystem and even a more predictable harvest.
And this has severe implications for software, knowledge management and our management of people in organizations in general.
We have tended to create monocultures and agriculture is not the only place but it’s the easiest example. In the Malay village where Scott did research for several years, the old men all knew at least 80 different landraces of rice, each having a particular set of strengths and weaknesses. If you don’t raise rice, you might think that you would certainly need several but not eighty. Perhaps one for wet and one for less wet, one for sunny and one not.
But agriculture is actually more complex than that. The rice paddy may be flooded only part of the year; the soil may be more acidic; the soil may be acidic only part of the season; monsoons may bring rain at different times depending on the weather patterns (think El Nino). The actual needs are quite complex, a multi-dimensional problem, not a simple 2×2 table.
Joel Salatin also talks about this, in a different way and a different context in his Salad Bar Beef. Salatin was an early American champion of popular, hand-raised beef as an alternative for ranchers to the money-losing ideas of today’s cattle culture. The book is an easy read and is a great primer on how to use your God-given creativity to make a job more interesting. He’s left the corporate world of fatty beef and only sells grassfed. I could give you an interesting debate on grassfed and cornfed but that’s not what he’s trying to say.
One of his points is that it takes you awhile to develop the herd gene pool that will work best on your particular lot. Sure, he says, there are breeds that are better for certain general regions than others. But within that breed, you need to find the particular traits that will work best on your land. The only way to do that, he argues, is to put some head on it and see which ones graze best. The ones that do well are the ones to breed, and the others are either this season’s harvest or off to the auction block. Just because it is a great steer doesn’t mean that it is a great steer for you and your land.
This is a hard lesson for us to learn. Cattlemen go for the feedlot, with its heavy use of antibiotics and steroids, partly because it allows them to use almost any steer. Instead of adapting to particular land, cattle are bred for adaptability to the feedlot. Feeding them corn giving them bloats and sickness? Breed one that doesn’t react that way.
There is something more advanced about having something that works best for you and not for all people. Levis will custom make jeans now, and it is considered a step forward that they are going from mass production to custom production. The World Wide Web was hailed because it allowed producers to go from broadcasting to narrow-casting, from producing mass-oriented news and materials to sending you only what you wanted or needed. This is even the idea behind almost every Knowledge Management effort, to create access to knowledge that is directly relevant.
Jack Vinson recently blogged about the joy of discovery in research, and others have written about the need for chaos to produce serendepity in scientific research, most entertainingly done by Connie Willis in Bellwether although I’ll not slight Tom DeMarco’s Slack: Getting Past Burnout, Busywork, and the Myth of Total Efficiency. It’s the definitive guide for this in the computer programming world, although what he says works across the knowledge-worker fields. We need time to play, to let our brains wander. It is in disruption, in working outside our own field, that we make the great discoveries, combine two unlikely ideas together to see something new. We can’t do that if we always get exactly what we want to see.
Information Technology has become a monoculture, both technically and culturally. We have adopted a single platform since, let’s face it, the Wintel platform has conquered the world. This leaves us vulnerable to a single exploit felling our entire server farm, much the way that tree farms with a single species all the same age would all be felled in a single storm or devastated by a single pest.
Mark Stamp argues similarly in a recent backpage column in Communications of the ACM, “Risks of Monoculture“. He believes that there is no benefit for going against the grain and choosing something other than market leaders. If we adhere to Nicholas Carr’s principles from his HBR article, “IT Doesn’t Matter (HBR OnPoint Enhanced Edition)“, then, yes, it offers no advantage. Except of course that Carr was shortsighted and it offers about the only advantage.
Think about the cost of losing business for a day or two. The MyDoom virus was so successful that Cisco is even using it to sell their products in their latest commercial, where various CEOs are at an expensive restaurant and all complain about how they lost two days and how their competitors will eat them now, except the guy who bought Cisco (of course). Fact is that Cisco is part of the problem, as a market leader.
By using a non-standard technology, you take on a greater risk but that risk has a payoff of being able to beat your competitors. If you are wining by using BeOS, it is highly unlikely that your competitors will follow suit. They will wait until a product is ported to their Wintel platform. If you go even further and gain competitive advantage by deploying a whole range of platforms, getting them to interact in creative ways, the complexity of the system will be impossible to replicate.
Which is why old firms with Mainframe apps find it almost impossible to move off them, since any application that you develop in a fresh field will not be able to have the complex richness of the spaghetti code developed over a decade or two. Or three.
I’ve heard a talk at a recent ISSA meeting in Chicago that coincided with what Stamp wants. He says that since no one will use nonstandard equipment, the only salvation is to deploy the technology now used in viruses, whereby they not only replicate but they replicate very differently looking children that behave in a similar way.
If all software installed a slightly different program at each site, viruses would have a harder time successfully attacking an entire platform, just like a diverse, multi-species forest is less susceptible to attack by pests.
The ISSA presenter (whose name and company I’ve forgotten) argued that the way to protect the network was to have the network know what was using it, what it was passing around. It makes the network stack obsolete, since it was developed by network engineers not by security folks. It’s an interesting argument but it leads us to having different environments in different areas. And maybe what we need is something more organic that allows for serendipitous occurrences.
Of course, that’s part of our problem. In order to outpace our competitors, we have to have an unpredictable or at least chaotic environment. You can’t simply have a skunk works, because that never works. You have to have allow for slack, for unpredictability, for variance. It’s interesting that we still attempt to manage as if we were a manufacturing economy and not a knowledge economy. Producing the same product a thousand times is great but it doesn’t produce good knowledge work. Indeed, it produces the lackluster knowledge work that so many consultants produce on a regular basis.
- Mark Stamp, “Risks of Monoculture”, Communications of the ACM, 47(3):120.
- James C. Scott, Seeing Like a State: How Certain Schemes to Improve the Human Condition Have Failed, Yale University Press, 1999.
- Joel F. Salatin, Salad Bar Beef, Polyface, 1996.
- Tom DeMarco,
Slack: Getting past Burnout, Busywork, and the Myth of Total Efficiency, Broadway Books, 2002.
- Connie Willis, Bellwether, Bantam Books, 1997.
Image credit: “Aberdeen Angus im Gadental an den Steilhängen des Muttawangjoch oder auch Mutterwangjoch genannt. Im Hintegrund die Südflanke des Feuerstein 2271m.” © 2013 Böhringer Friedrich (CC BY-SA 3.0 AT). Via Wikimedia Commons.