VISTA Enterprise Network - Successful Implementation, World Class Support

Wednesday, February 1, 2012

Four Problems with Master Plans: (1 & 2) Precision & Imprecision, part one: Verstand

On 19 August 2009, I began a series of posts on Christopher Alexander's writings about organic order from his groundbreaking and underappreciated book, The Oregon Experiment (http://vistaexpertise.blogspot.com/2009/08/book-recommendation-1-oregon-experiment.html). My series ran until 24 November 2009 (http://vistaexpertise.blogspot.com/2009/11/principle-of-organic-order-our-journey.html), when I summarized the series to date before turning to an exploration of fluxus quo.

Christopher Alexander (photo by Jerry Telfer)
In a typical rejection of the idea that old is bad and new is good, I'm turning back the clock to resume my series.

We left off in the middle of a discussion of four problems with master plans. We have discussed (1) precision and (2) imprecision. Before continuing on to (3) alienation, we need to take a closer look at why master plans are both too precise and also not precise enough.

The problem is a pressing one, because ever since the passage of the Clinger-Cohen Act (1996) the main federal VISTA adopters (VA and IHS) have done most of their VISTA development through contracts awarded and administered through the U.S. General Services Administration (GSA). This method of software development requires that the federal agency in question specifies at the outset the outcome for which they are contracting. Every federal VISTA-development contract begins, de facto, with a master plan.

So why do the plans that VA and IHS try to use to develop VISTA lead them to both too much precision and too little?

In The Oregon Experiment, Christopher Alexander answers thusly: because master plans reduce developers, users, planners, and project managers to prophets, to predicting the future, to trying to decide in advance how things will turn out. We suck at prophecy. Any system of organization that requires us to be prophets steers us toward our weaknesses, makes failures of us, because we rarely know at the outset how things will turn out in the end.

Alexander's right, and that's the easiest way to explain what's wrong with master plans, but there is another explanation, another way to look at the problem.

Hegel (steel engraving by Lazarus Sichling)
In Die Philosophie des Geistes [The Philosophy of the Mind/Spirit, 1816-1830], German philosopher Georg Wilhelm Friedrich Hegel makes a distinction between two different kinds of thinking that produce different results. In English we can't easily make this distinction, because we lack the words to express the difference, but in German Hegel was able to show just how different these two kinds of thinking are. One of these kinds - Verstand - is used in the production of abstract master plans; the other - Begriff - is used in concrete, user-driven development.

As Hegel defines the terms, Verstand is a left-brain process of analysis, in which the whole is analyzed into parts. Verstand sees the whole as merely the sum of its parts, attempts to comprehend it through identification of those parts, attempts to control it through the manipulation of those parts. This fragmenting perspective on wholistic, organic systems invites Verstand to try to deduce knowledge about the whole through a process of logical derivations from a priori assumptions (such as what the parts are that supposedly make up the whole). It's like trying to understand the world if your only tool for investigation is a knife, good only for chopping things into pieces and rearranging them on the table.

Toaster, made of parts (photo by Conavan Govan)
The problem is that neither life nor living systems are logical syllogisms, nor are they puzzles. Organic order cannot be understood through dissection. Living systems are not composed of components, of discrete parts; you cannot remove your arm and then put it back on without doing terrible violence to yourself. It only looks like a part, and only to the Verstand process of thinking, which sees all things as abstract, manipulable parts. A cat is not a toaster, but that's how the Verstand mode of thinking sees a cat, or a house, or a software system, or a human being.

With Verstand we cannot model living systems without doing such violence to them that we deeply miscomprehend them. The grotesque inadequacies and distortions of the clumsy, clanking mechanisms by which Verstand insists on trying to understand organic order creates false distinctions between what are only aspects of a vast, organic whole. Further, while attending to these false distinctions it also misses whole swaths of vital meaning that do not present themselves in the Verstand-friendly forms of parts or facts or axioms or deductions. From the resulting homunculus, Verstand then tries to draw logical inferences about the optimal future state, then builds plans to try to reach that false future through mechanistic means and logical deduction. Even before the plan is written, the battle has already been lost because the planners are not dealing with reality but with Verstand's blinkered, mechanistic conception of reality. It's as though the master-planning process begins by putting on one-dimensional glasses, and then drawing up plans based on the linear, cartoon  world that results.

"Keep that up and I'll bite one of your parts."
As a man is, so he sees. With Verstand, we see a logical puzzle, a mechanism, which no living system is. In other words, it isn't just that when we draw up master plans we are looking at the real world and unable to foresee its future. With the Verstand processes of thinking, we are not even looking at the real world. We are looking instead at the world of abstract ideas, which is awash with distortions and falsehood.

This is why master plans are paradoxically both too precise and not precise enough. The key to avoiding this paradox is to avoid using Verstand thought processes when dealing with complex, organic order.

Fortunately, human beings are also equipped from birth with an alternative mode of thinking, with Begriff. In part two we'll show how to distinguish the two and how to use them in their proper spheres.

2 comments:

Rick Saling said...

It occurred to me that the toaster is not JUST made up of parts, but is also part of a much broader electrical ecosystem (and that ecosystem might be complex enough that top-down abstraction doesn't suffice to understand the whole), and the electrical system is part of the Earth's ecosystem, with varying reciprocal effects.

Rick Marshall said...

Yes, even though a toaster can be disassembled and then reassembled to create the same toaster again, it still doesn't do a damn thing of value until you connect it to a system of dynamic flow that bring it to a mechanical semblance of life. Organic processes underlie and power even the most mechanistic of mechanisms.

Of course, the more the complexity and feedback loops in a system grow, the less like a mechanism it becomes, until it is simply impossible to disassemble it into its parts and then reassemble it again - because it no longer truly has parts, only aspects of the whole that appear to the Verstand processes of thinking to be parts.

Frankenstein is perhaps the best known allegory about the problem with trying to treat organic systems as though they are mechanisms - we create something other than what we expect, not life but death.

And yet, VA and IHS are hand-cuffed by the Clinger-Cohen Act into repeating over and over the pretense that VISTA is a mechanism, that contracts that specify in advance exactly what constitutes successful completion of the contract can possibly work with a complex, organic, living system like VISTA. The necessary iterative process of discovery and adaptation without which organic order cannot be managed is forbidden by the GSA contracting process, which requires predefined outcomes, and which therefore with VISTA produces sadly foregone conclusions over and over and over to the tune of billions of dollars wasted - and still counting.

The GSA contracting process is the hammer, and the federal government is the child to whom all problems look like nails.

Fortunately, there is a better way.

Unfortunately, short of repealing the Clinger-Cohen Act, I have not yet discovered how the federal government will ever be allowed to return to that better way. In the pursuit of government efficiency, we have created grotesque new scales of inefficiency.

Where there's bitter irony, there are sure to be human beings at work.

Perhaps VA CIO Roger Baker will be able to use the new open-source EHR custodial agent, OSEHRA, to escape from Clinger-Cohen's shackles, to try once again that better way, which I hope to describe in future posts, but that Christopher Alexander has already described from an architectural perspective in The Oregon Experiment and from a biological/mathematical perspective in The Nature of Order.