More on The Org

by Henry Farrell on April 11, 2013

Tim Sullivan responds to my post at _OrgTheory._

bq. The point of the AA story, though, was not that organizations are perfectly efficient but that organizations face tradeoffs, and it can be useful to acknowledge those tradeoffs explicitly and to understand the economic architecture of organizations because it makes the situation of the average employee, manager, executive more comprehensible. In the AA case, they had a terrible website (which reflected plenty of other dysfunction within the company), and yet to do the job that AA aspired to (that is, flying people and stuff all over the world), you have to build a big, complicated organization that does lots of things all at once – managing fuel contracts, negotiating with pilots and flight attendants, setting prices, and so on. And organizing all of this involves a lot of tradeoffs. … Ray and I aren’t suggesting that orgs can’t be full of politics, power plays, bad managers, ridiculous HR departments, and so forth. They clearly are — but you have to accept these realities when you decide that there’s something that you want to do that will be best accomplished as a group of bosses and employees. The trick is not to ignore them or pretend they don’t exist, but to understand how and why they are produced, to recognize that sometimes apparent inefficiencies are the result of being organized, and understand the difference between tradeoffs and the _truly_ ridiculous and pointless aspects of organizational life.

I think that the nub of the disagreement is best summed up in one half-sentence here, where Tim suggests that “you have to accept these realities when you decide that there’s something that you want to do that will be best accomplished as a group of bosses and employees.” The point of the alternative perspective I set out is that there _isn’t_ any moment when a collective ‘you’ of bosses and employees, with a common interest in getting something done, decides this. The actual ‘you’ who makes the decisions is a very specific ‘you’ with a very specific set of interests. It is the ‘you’ who is in charge (or, if you want to get all old-style, the ‘you’ who is a capitalist). There is a literature of course in organizational economics, which talks about ‘team production functions,’ and how teams might rationally, if they wanted to get stuff done and minimize shirking, assign oversight to a hierarchically empowered actor. But in an economy which is not organized around cooperatives, very few private enterprises will originate in this way. Instead, they will originate with decisions by owners of capital, who will empower managers (a group which may, or may not, overlap with the owners of capital) to hire workers. The logic will be different, obviously, in non-profits and the government sector, but less different than you might imagine, as both these sectors become more and more like private enterprise.

The obvious consequence is that the owners of capital will not have an incentive to design organizations to solve problems. Instead, they will have an incentive to maximize the returns to their investment. This will mean that they will, as a general rule, prefer organizational rules that are lousy at problem solving but that increase their profits to organizational rules that are less lucrative and more efficient. Similarly, managers will prefer rules that increase their personal returns and protect their jobs to rules that are better at furthering organizational goals but likely to put their economic benefits at risk. And if the owners of capital, and, to a lesser extent, the managers, are _setting the rules_, then these aren’t problems that one can design around, or even decide that you reluctantly can accept. The actors doing the designing – the “you” of Tim’s phrase – are the owners or the managers. From their perspective, the problems aren’t actually problems. They are not bugs, they’re features. They are part of what the Org is supposed to do, from the perspective of its most important designers.

This conducts towards an _entirely different_ way of thinking about organizations, one in which many of the unpleasant features of organizations are not quirks of bad managers or the background noise of office politics, but instead are baked into the cake. This perspective understands the organization not as a cooperative venture plagued by pesky inefficiencies, but as a power struggle, a distributional game played by actors with different interests, each of which is striving to maximize its particular benefits, if necessary at the expense of everyone else. When Walmart plays with the tasks allotted to its employees to try to get those with potentially expensive health problems to quit, this is, in a very obvious sense, efficient behavior for Walmart’s owners, even if it isn’t great for the employees, and even if it results in inefficiencies as employees are required to carry out demanding physical tasks that they are obviously unable to perform. The ‘you’ who is deciding here has a set of interests that are clearly at odds with the interests of the employees who are getting decided about. But it would be odd to see this as inefficient, or as impeding the goals of the organization, since the (informal) rules are doing what they are supposed to do.

To put it another way (stealing from Jack Knight’s arguments) – one cannot have both (a) selfishly rational actors, and (b) collectively beneficial institutional outcomes in a world where these actors have clashing interests and different levels of bargaining power. In this kind of world, it may be better to see rules and organizations not as the product of any desire to solve collective problems, but instead as a consequence of power struggles between actors, and hence likely to perpetuate inequality and inefficiency rather than to remedy it. And the Org is just such a world – a world based on hierarchy, in which the most important forms of control lie with the owners of capital and with senior managers, actors who have their own individual sets of interests, which come into play when they set the rules. Jack (like me) is a lefty – but there is a right leaning version of this argument too, in which people like “Terry Moe”:http://journals.cambridge.org/action/displayAbstract?fromPage=online&aid=310907 and “Stephen Krasner”:http://www.jstor.org/stable/2010398 criticize liberals for their emphasis on collective benefits rather than power. This approach cannot explain everything – actors aren’t fully selfish, are sometimes able and willing to coordinate for the common good and so on. There is space too for the kind of liberal account that Tim and Ray Fisman so ably lay out. But I don’t think that the power-and-distribution approach can be easily subsumed into the liberal account, since it suggests that certain ‘inefficient’ aspects of the hierarchical organization are basic to its design, and, properly understood, its purpose.

{ 40 comments }

1

Tim Wilkinson 04.11.13 at 8:46 pm

1. A propos I’m not sure what, how teams might rationally, if they wanted to get stuff done and minimize shirking, assign oversight to a hierarchically empowered actor sounds very Hobbesian. I don’t know if this is remarked on in the lit.

2. organizing all of this involves a lot of tradeoffs. … you have to accept these realities when you decide that there’s something that you want to do that will be best accomplished as a group of bosses and employees sounds like a microcosm of the guilty open secret of the ‘market efficiency’ nuts. Comparing the realistic Soviet Central Planner to the ideal Walrasian Auctioneer, they find that the latter does better at a very particular ‘optimisation’ task. Then, having adopted the Market as the implementation of Walrasian Optimality, they will, if pressed very hard and persistently, mumble something about how leaving large corporations to do what they will with their enormous (quasi-)rents is just how it has to be in a market economy (piling on epicycles about social Darwinism and the rest of the ‘supply-side’ mythology as required).

2

Ana 04.11.13 at 9:17 pm

I think the key line here is “The trick is not to ignore them or pretend they don’t exist”… planting in your brain the notion that indeed you, dear reader, had done this, since you hadn’t anticipated the following paragraph already… this literature is all about setting up a “gee whiz why didn’t I think of that” sentiment, which thirty seconds of reflection or common sense dialogue reveals to be false…. this literature preys on the exact same behavioral biases that toaster insurance does…

3

Bruce Wilder 04.11.13 at 9:21 pm

Not to be too picayune, but “tradeoffs” is the wrong label.

A “tradeoff” is movement along a frontier of allocative efficiency. That’s not what bureaucracies are doing, for the most part; they are engaged in cybernetic control of processes, which is another dimension of efficiency altogether. In terms of the frontier of allocative efficiency model, they are trying to either reach that frontier, or move the frontier outward.

Where exactly they are on that frontier, and whether a tradeoff would improve their position is usually a trivial matter, and when it is not — e.g. the price of fuel just skyrocketed and a crash economy is required — it is a transitory situation.

There are conflicts. For the middle manager, there’s the stress of being tasked to control two or more outcome variables, simultaneously, which, in terms of cybernetic control of a system, is impossible. In any system, it is only possible to control one thing at a time. Being required to control two or more things is impossible and, therefore, stressful. From the upper management’s point-of-view, they are not requiring crazy; they are requiring the manager to seek as her only out, technical improvement. The stress is part of motivating the movement out to the frontier of allocative efficiency.

For the upper management, the challenge is making a coherent system of the disparate processes under control — the primary topic of the OP, I suppose. Creating a system of systems isn’t about “tradeoffs” per se, though. So, wrong metaphor.

4

Barry 04.11.13 at 11:08 pm

” Being required to control two or more things is impossible and, therefore, stressful. ”

Stressful, yes, impossible, no.

5

Patrick 04.11.13 at 11:23 pm

If the “you” is the creator of a business, the statement is wrong for the reasons you mention. But if it is the society that crafts the environment in which organizations exist and the structure of rules and relationships that make them operate, things get even weirder. If society chooses organizations as an instrumental tool for, say, the goal of distributing hamburgers, then McDonalds profit margin is an inefficiency, and we should be trying to minimize it so that those funds can be used to enhance burger distribution rather than fattening the wallets of investors. Record profits would be a sign of disfunction.

6

Main Street Muse 04.12.13 at 1:20 am

Is there anyone who works at an org who ignores “the realities” in the work of deciding that “something you want to do that will be best accomplished as a group of bosses and employees”? From all I’ve read and heard, AA (like many airlines today) sounds like a case study of an org filled with pointless aspects of organizational life.

What’s the point of this story about the guy who got fired for noting the bureaucracy preventing approval of a better website? To inform us of those pointless things? Or to showcase the obvious – that all work at orgs includes tradeoffs? And those tradeoffs might mean getting fired?

7

reason 04.12.13 at 8:31 am

Bruce @3 like Barry @4
I wonder exactly what you mean by saying being able to control two things at once is impossible. If I substitute temperature and humidity for the two things – I know that this is possible. Isn’t the problem a degrees of freedom problem – that the number of independent tools doesn’t match the number of targets?

8

Bruce Wilder 04.12.13 at 10:27 am

In a system under control, you can only control one variable: you control that variable, and let everything else in the system vary as it will, or as you need it to vary, to maintain control of the one thing you choose to control. When you are driving a car down the highway, you can control the speed of the car only to the extent that you are willing and able to manage everything else to maintain speed: fuel economy, steering around the curve in the road, avoiding (or killing) the pedestrian in the cross-walk, etc. You give up control of the vehicle’s speed, when you choose to control something else, by, say, braking for that pedestrian in the crosswalk, or maximizing fuel economy, by reducing speed.

We are so used to switching focus in the zigzag management of our pursuit of fuzzy goals that the formal logic of my assertion may seem strange, but it’s recognized, for example, in the formal analysis of, say, monetary policy. What’s more familiar to managers, I suspect, are iron trilemmas: e.g. “Fast, Good and Cheap — choose any two”. Managers are often given conflicting goals: they are required to conform to budget, meet schedule and achieve performance goals. I have a friend, who manages stores for a retail chain, with a lot of low traffic locations. Store managers are required to keep stores open during stated business hours, to meet budgets for staff (implying a certain number of employees and hours worked) and meet various performance goals. One rule says that there must always be two employees in the store at all times, but, of course, there’s never enough staff hours budgeted to conform to the rule. The rule only matters if the store is robbed; if the store is robbed with only one employee on duty, the manager is fired. So, managers are very careful about when they schedule a single employee to staff the store. I’m sure economists could construct a Principal-Agent analysis of revealed something-or-other to explain it as asymmetric information; I just wanted to note that it’s stressful to work under such a regime. (And, it’s not a “tradeoff” — it’s about how you get to the frontier, where a tradeoff might take place; the economy involved is a technical economizing on error or waste, not an economy of, say, substituting capital for labor (which could conceivably involve actual tradeoffs) in response to relative prices.

People in organizations are required to strive, while constrained in multiple ways. Typically, their role is embedded in systems they only partially and imperfectly understand; they have to follow rules, not their own judgment, except (as in the case of my store managers) when they have to break the rule, exercising their own judgment, at their own risk.

The systems can be quite complicated. Forbes had an interesting little article on the differing regulatory regimes governing eggs in the U.S. and Britain, and how these regimes could have rules, which seemed diametrically opposed on their face.
http://www.forbes.com/sites/nadiaarumugam/2012/10/25/why-american-eggs-would-be-illegal-in-a-british-supermarket-and-vice-versa/

Again, I don’t think “tradeoffs” cover the case. A system architecture is put into place, and rules and processes are developed in light of experience, and the system evolves or deteriorates, because control — especially social control, with all its reflexive gaming of the system — is dynamic and emergent.

9

Metatone 04.12.13 at 12:46 pm

Alongside Bruce Wilder’s points, I’d note that contra most of the simplifications common in the literature, “efficiency” isn’t even one thing, it’s a number of things depending on the unit of time used for analysis.

Allocating ill employees to lifting heavy boxes is “inefficient” in the short term, but possibly “efficient” in the longer term. Critical to note is that there’s often a conflict inside a traditional power bloc (e.g. managers or e.g. owners) over whether short or long (or medium) term efficiency is the priority at this moment in time.

10

Trader Joe 04.12.13 at 12:56 pm

“The obvious consequence is that the owners of capital will not have an incentive to design organizations to solve problems. Instead, they will have an incentive to maximize the returns to their investment. This will mean that they will, as a general rule, prefer organizational rules that are lousy at problem solving but that increase their profits to organizational rules that are less lucrative and more efficient. ”

I don’t think this is inherently true in most organizations or for the most important types of organizational problems. In most organizations, a good solution to a business problem is also one that maximizes profits over the long-term.

I’d fully agree that information asymetries, managerial priority conflict and communication barriers may prevent finding the necessary solution – but this isn’t the same as saying an org isn’t designed to solve problems – most are, they just aren’t able to always achieve a perfect batting average for a variety of reasons.

Indeed in the AA example – having an awesome website is clearly a tool that can drive business volumes and create profit maximization, thousands of businesses have figured that out. If your premise is correct AA would have taken the necessary incremental steps (particularly since they were laid out by the fired employee as being fairly easy) to make the profit maximizing decisions. The fact that they did not do this suggests some barrier to making good decisions or a failure to recognize what the profit maximizing decision would be – not a purposeful failure to design an organization for problem solving.

11

Barry 04.12.13 at 2:25 pm

Bruce: “In a system under control, you can only control one variable: you control that variable, and let everything else in the system vary as it will, or as you need it to vary, to maintain control of the one thing you choose to control. When you are driving a car down the highway, you can control the speed of the car only to the extent that you are willing and able to manage everything else to maintain speed: fuel economy, steering around the curve in the road, avoiding (or killing) the pedestrian in the cross-walk, etc. You give up control of the vehicle’s speed, when you choose to control something else, by, say, braking for that pedestrian in the crosswalk, or maximizing fuel economy, by reducing speed.”

That’s odd, because I’m quite capable of varying the speed and direction of a car simultaneously.

12

Lee A. Arnold 04.12.13 at 3:41 pm

I may start a campaign to restrict the word “efficiency” to mean only the energy-saving in a technological improvement, and to restrict the word “rational” to the calculating of ratios. For the remainder, we can use phrases like “profit maximization”, “self-regard”, “ecological vacuity”, and the like. Then we could combine them. For example, for “allocative efficiency”, which is one of the most useless and misleading phrases ever concocted, we can instead use “profit maximization + self-regard + ecological vacuity”.

13

Bruce Wilder 04.12.13 at 3:47 pm

You can walk and chew gum at the same time, too, Barry. The important qualifier is, in a system, one system, under control. So, yes, your car has a steering wheel and an accelerator, tied into different systems. What constitutes a single system can be somewhat a matter of analytic convenience. If you don’t moderate your speed when you change direction, you may find you lose control of the vehicle.

14

Lee A. Arnold 04.12.13 at 4:01 pm

Well in that sense, the automobile is a set of polycentric subsystems, each having a separate intention. The fact that automobiles can be made to work, might answer Henry’s objection that “one cannot have both (a) selfishly rational actors, and (b) collectively beneficial institutional outcomes in a world where these actors have clashing interests and different levels of bargaining power.” Because I think you can, indeed we must. The solution begins by understanding the ways in which all organization is always the same, whether it is group organization or individual self-organization.* Next, then allow for polycentric institutions, e.g. there will be market organization, plus there will be social spending. Finally, observe that the key to the lock is total social preference, which can be changed. The only people who don’t seem to get the program are the ones who ran off with believing that Milton Friedman is God.

*See here, after the ponderous scrolling intro: http://www.youtube.com/watch?v=GrVsLdTtepM

15

mpowell 04.12.13 at 4:20 pm

I think you’re definitely confusing layers of decision making here. The management of companies is an extremely curious thing because it is absolutely not true that the are controlled by their capitalist owners. They are controlled by their corporate board and chairman, who typically hire an executive CEO to do most of the management. There are lots of conflicts of interest here and mutual backscratching, but I think you would find that the general goal of the board is to maximize share value and medium to long term profits (though these are sometimes at odds with one another and also these goals may not align with employee interests in large classes of cases). And sometimes the CEO will attempt to undermine the board’s governance though information hiding. But this is really just the first example of the process of fiefdom building which comes not out of the goals of the board, but through the problems with delegation. It’s not as if a bunch of guys get together and someone gets to be CEO and a bunch of others get to be vice-presidents and they go about forming a company and deciding on how it will be run. You can’t really claim that the mismanagement that occurs at this level represents fundamental organizational goals. Most of the rules are set up by the CEO. When a VP punishes an employee for making him look bad (and this action comes at company wide expense) this is not something the CEO generally approves of, but they may recognize it as a close to unavoidable cost of the necessary process of delegation of authority.

That being said, I’m not sure it makes sense to try to analyze a corporation in a, “this is what corporations try to do” way. There are different people trying to do different things, from the owners to the managers to the employees and hardly anyone fits perfectly under a single one of those labels. It is just too complicated to break down to a simple narrative.

16

William Timberman 04.12.13 at 4:27 pm

In management as well as in economic policy, defining which are the independent and which the dependent variables is always a matter of choice, and as often as not the choices made are — or appear to be — completely arbitrary. Those inside the belly of the beast have no idea who made those choices, or on what basis they were made. What they do know is what any Dilbert knows, or, to put it in Soviet terms: They pretend to pay us, and we pretend to work. Given who and what we are, it’s hard to imagine any large organization working any other way, at least not for very long.

17

Sam Penrose 04.12.13 at 5:03 pm

Venkat Rao’s wonderful Gervais Principle series is an interpretation of The Office through this lens. The latter parts are directly relevant to this post, but start from the beginning anyway:

http://www.ribbonfarm.com/the-gervais-principle/

18

Bruce Wilder 04.12.13 at 7:35 pm

mpowell: “When a VP punishes an employee for making him look bad (and this action comes at company wide expense) this is not something the CEO generally approves of, but they may recognize it as a close to unavoidable cost of the necessary process of delegation of authority.”

The Cossacks never work for the Czar. If the Czar only knew what was being done in his name . . .

mpowell: “It is just too complicated . . . ”

So many things are.

19

mpowell 04.12.13 at 7:52 pm

BW @ 18: That comment certainly seems unduly snide, not to mention carries a completely unjustified analogy. The view that all of the senior management work together cohesively to screw the remaining employees is just as naive as the one where they all work industrously for the betterment of the firm. Information assymetries are a well understood and explained problem in organization management.

20

Bruce Wilder 04.12.13 at 7:53 pm

Lee A. Arnold: “. . . the automobile is a set of polycentric subsystems, each having a separate intention.”

I don’t think the automobile, or any component of it, has any intention at all, but it is a system of systems: individual components of subsystems, like the engine, are recognizable as mechanically controlled processes. The driver, presumably, provides the intention.

21

Bruce Wilder 04.12.13 at 8:13 pm

mpowell: ” Information assymetries are a well understood and explained problem in organization management.”

Well, no, but I like your spelling.

22

Mao Cheng Ji 04.12.13 at 8:49 pm

From what I’ve seen, I believe Bruce is mostly right: they tend to identify one parameter that they believe is problematic, and fix it. Then another one. Then the next one. By that time, the first one is out of whack again.

Payroll is too high, we need to get rid of employees! A few years of layoffs, the place is full of consultants. Oh, shit, the place is full of consultants who come and go, no institutional memory, no one know what’s going on! Let’s get rig of consultants, hire some people who can bring quality and consistency! And so on. Kind of a campaign mode. What was extremely important yesterday will be completely forgotten tomorrow; you can bet on that.

23

Coulter 04.12.13 at 9:08 pm

“The obvious consequence is that the owners of capital will not have an incentive to design organizations to solve problems.”

Well, I guess I see much alignment in maximizing profits and solving problems. AA does a pretty good job solving the problem of getting people from point A to point B and getting paid to do it, no?

24

Barry 04.12.13 at 9:18 pm

Bruce Wilder 04.12.13 at 3:47 pm

” You can walk and chew gum at the same time, too, Barry. The important qualifier is, in a system, one system, under control. So, yes, your car has a steering wheel and an accelerator, tied into different systems. What constitutes a single system can be somewhat a matter of analytic convenience. If you don’t moderate your speed when you change direction, you may find you lose control of the vehicle.”

In the end, you’ve made a claim, and not backed it up (not being able to control all things is not the same as being able to control only one thing).

“What constitutes a single system can be somewhat a matter of analytic convenience.”

In this case, a matter of what you wish to be a system.

” If you don’t moderate your speed when you change direction, you may find you lose control of the vehicle.”

It depends on the speed, and my whole point was that I was capable of doing both simultaneously.

25

mpowell 04.12.13 at 11:20 pm

BW @ 21: Great, make fun of my spelling. But I hardly imagined that this would be so controversial. The principal agent problem can exist at many levels in an organization.

26

Tony Lynch 04.12.13 at 11:40 pm

May I say that I too quite liked your spelling. It made me laugh. But not at you. At the spelling.

27

Dr. Hilarius 04.13.13 at 2:16 am

mpowell@15 is on the right track. One way to think about organizations is levels of selection. Classical group selection hypothesized that group members would lower their fitness in order to increase the group’s overall fitness. Doesn’t work, other than under very unusual circumstances, as individuals give priority to their needs with immediate needs getting priority even when in conflict with longer-term needs.

I’ve worked for dysfunctional companies where top management told middle management to achieve certain goals. The managers push for those goals by every available method, often without concern for legality or the long-term health of the company. Line level workers in turn do whatever is needed to keep supervisors happy, with the same disregard for the longer term. The larger the organization and the more mobile the decision makers (at every level, not just the top) the greater the chance for real dysfunction at the organizational level. Corporate officers do deals for immediate bonuses knowing they are going to take the money and run. Fast-food employees ignore food safety rules to compensate for being deliberately under-staffed by management.

It’s a Jungle out there.

28

Peter T 04.13.13 at 5:54 am

Coulter

Well, I guess I see much alignment in maximizing profits and solving problems. AA does a pretty good job solving the problem of getting people from point A to point B and getting paid to do it, no?

Airlines as a whole have made no profit at all since 1945 (good years being more than offset by bad years). Yes they mostly do a good job of getting people from A to B, but being paid for it consistently seems to be problematic. Ditto passenger railways. So yes, there are significant sectors where the simple story of profit as motivation seems a bit thin.

29

dk 04.13.13 at 12:13 pm

I usually enjoy Bruce Wilder’s comments, but he is way off base here. Control of multiple target parameters using multiple control parameters is a central part of modern control theory, and such control is both theoretically and practically possible under a reasonable set of conditions.

30

William Timberman 04.13.13 at 1:43 pm

dk @ 29

I get your point, but we’re talking about organizations here. Human beings make for very unpredictable servomechanisms, and that sets unknown limits on the precision of your multiple control parameters. To put it another way, was Lehman Brothers bad management, or a bad dream?

31

Luis Sanchez Campo 04.13.13 at 3:23 pm

Interesting article. I think that many companies do not raise good administrative organization, nor staff organization. This leads to lower profits or even losses several, as we are seeing in recent years.

32

Bruce Wilder 04.13.13 at 3:53 pm

Barry, I certainly did not make clear what it means to be a system under control; I thought maybe people would already have some familiarity with the concept, and I could just invoke it to make my small point.

dk, my target was the concept of a tradeoff; I wanted to build from an analogous atomistic meme. Driving a car might a nice example of the problem of multiple target parameters and multiple control parameters (but shouldn’t it be target variables and control parameters?). On that level of analysis, the phenomena of impossible demands shows up as the iron trilemma. You still cannot drive fast and save on gas at the same time, and you can drive to destinations, serially, but not simultaneously in opposite directions, and you have to moderate speed to make it safely around a sharp turn.

mpowell, I see I was being mean. The problem with principal-agent analysis — which every bright group of students sees immediately when first introduced to the idea, in my limited experience — is exactly that it deliberately ignores precisely the problem Henry is analyzing in the OP: realistically, the subordinate (the “agent”) is as interested in the economic purpose of the organization and the outcome of cooperation as the manager or capitalist (the “principal”), but the analysis is only about how the principal prevents the agent from cheating the principal, and the more practically important problem is the principal shirking or cheating the agent. There’s a power struggle, but the analysis is strangely, and defectively, one-sided. The problem with Old Russia is not that the peasants are bloody-minded shirkers and the Jews are Jews; the problem is that the Czar is an ignorant, greedy incompetent ripe bastard with all the power, and no sense; so, sure, let’s analyze how to prevent the peasants from shirking, and how the Cossacks could be nice, and still do their jobs.

33

Barry 04.13.13 at 4:56 pm

Bruce Wilder 04.13.13 at 3:53 pm

” Barry, I certainly did not make clear what it means to be a system under control; I thought maybe people would already have some familiarity with the concept, and I could just invoke it to make my small point.”

My point is that you seem (to me) to be in effect defining a system as ‘that of which you can control only one thing’.

34

Barry 04.13.13 at 5:10 pm

Bruce: “…but the analysis is only about how the principal prevents the agent from cheating the principal, and the more practically important problem is the principal shirking or cheating the agent. There’s a power struggle, but the analysis is strangely, and defectively, one-sided. ”

I don’t think that this is an accident for traditional economics – the viewpoint is from those who have power and money.

35

Barry 04.13.13 at 5:12 pm

Note – ‘good, fast, cheap – pick any two’ is in fact a system where two things can be controlled. And given that it’s frequently not a binary choice,….

36

Barry 04.13.13 at 5:24 pm

Bruce, what I will gladly concede is that in a large, complex, real-world system (like a large corporation), there are a limited number of things that the leadership can practically control. I’ve seen advice to new CEO’s that the best they can do in a few years will be only one thing, so pick the most important thing.

37

Bruce Wilder 04.14.13 at 3:37 am

Barry: “My point is that you seem (to me) to be in effect defining a system as ‘that of which you can control only one thing’.”

Roughly, yes. A simple system creates functional relationships between variables; you can pick one to be the dependent variable, and manipulate the other variables in the system to control the value of the dependent variable. I can see that the possibility of complex systems, composed of multiple systems, alongside the possibility of strategic creation of systems, created a distracting confusion, when I stated my premise emphatically, but without sufficient qualification or elaboration. (It certainly wasn’t my intention to hijack the thread.)

My point — hopelessly lost, now, I fear — was that bureaucratic hierarchies, which exist as both creators of, and part of the structure of, controlled systems of production, assume a characteristic social character as a result. We ought to interpret that character in light of the efficiency values of controlled systems, and not muck up that interpretation with misplaced terms, like “tradeoff”. Logically impossible, because contradictory and conflicting, demands on subordinates, including subordinate managers, was my intended example of seemingly irrational and unpleasant organizational behavior. But, if you think that bureaucratic organizations, being capable of doing many things at once (which is certainly true), are not constrained by their own commitments to particular controlled systems, by the nature of controlled systems, you lose the essential flavor of my argument. The circular irony of motivating subordinate managers to realize efficient control, to move toward the frontier of allocative efficiency, by demanding that they control multiple variables, when control of one variable conflicts with control of the other, is lost. As is, I fear, what that sort of domination feels like to the people involved: the contempt, the stress and the resentment, in particular. And, the clumsy arbitrariness of any particular system of control.

This my little hobbyhorse: the misapplication of, and uncritical use of terms and concepts borrowed from a benighted economics to mis-describe and mis-understand the political economy around us. I’m always a little afraid it will make me a troll, as it seems to have, quite inadvertently, on this thread.

38

Bruce Wilder 04.14.13 at 4:04 am

Barry: “Note – ‘good, fast, cheap – pick any two’ is in fact a system where two things can be controlled.”

Well, no. Trying to jump out of the hole I had dug for myself, I made a huge, Atlantic-sized leap, by introducing the iron trilemma. It was a stupid gambit on my part, to put the focus where I wanted it, on the experience of people stressed by the contradictory demands of economic life in bureaucracies.

The iron trilemma is a meta-level observation about the inherent limits in strategic choice of systems of control. “good, fast, cheap” are qualities realizable from control schemes (in project management), not variables in a system of control. In fact, given enough time and investment, in a context other than project management, an elaborate, well-controlled and complex system can deliver all three. That’s what mass-production often accomplishes, after all: high-quality product, produced quickly, and cheaply. The mass-produced automobile is a testament. Bureaucratic hierarchies can produce extremely impressive results and “efficiency” realized in total factor productivity gains sufficient to dominate alternative ways of doing things — there’s no tradeoff, say, with more labor-intensive, less capital-intensive methods, such as economists might imagine.

39

Bruce Wilder 04.14.13 at 4:10 am

(The truth of the project management adage about the iron trilemma is in the deep, sunk-cost investment in product development and manufacturing systems design, which mass-production requires. It doesn’t take long to produce a car, once you’re up and running, and the process is neither capital-intensive nor labor-intensive, normalized to units of output, but . . . )

40

mpowell 04.14.13 at 5:19 pm

BW @ 32:

I appreciate your follow up. I didn’t mean to imply that the classical principal-agent problem is the entire story. The point I was making is that the dynamic that is present in these kinds of relationships can and typically will be present at multiple levels of an organization. And in many cases, management of managers is much harder than management of line level employees. Sometime the cossacks are not working for the czar, whether the czar is mean and stupid or intelligent and far-sighted. So I think that analysis which relies on grouping people into ‘management’ and ‘labor’ runs the risk of getting the wrong result for this reason.

Comments on this entry are closed.