The Hawthorne Effect is a basic idea in social science research that I first encountered in William Whyte’s Organization Man. I remember making a note of it at the time but never got around to thinking more about it, until it came up again in a recent conversation.
It is a sort of Heisenberg Principle for social science. The effect hypothesizes that the changes in participants’ behavior during a study may be mostly caused by the special social situation and treatment they receive from management.
I think the principle has far greater significance than people realize.
The original effect was discovered by productivity researchers at the Hawthorne Works of the Western Electric Company in the 1920s, who found that just about any intervention seemed to improve productivity, from minor changes in lighting to the kind of food provided. The researchers speculated that just being the focus of well-intentioned management efforts to improve their lot helped the workers’ morale (by leading them to trust management’s genuine concern for their welfare) and led to higher productivity.
Or to put it another way, given the same set of laborers, the management model that involves paying more attention will do better. Within limits, the model itself does not matter.
In an unrelated thought trail a few days ago, I was trying to dig deeper into the meaning behind the “people over process” idea, which is especially common in information work. The principle is usually stated as “good people working with a mediocre process will beat mediocre people working with a great process” (if I recall correctly, this particular form of the statement is in Good to Great but I’ve seen at least a few other versions of the idea).
A second apparently unrelated idea came up in some excellent essays on military strategy that I’ve been reading. Many great military leaders in history have claimed that unless there is a radical technological breakthrough involved (like guns vs. swords), differences in weapons don’t matter. They make the claim that they would have won even if they’d exchanged roughly equivalent weapons. The Israeli Air Force commander, General Mordecai Hod, remarked after the victory over the Arabs in 1973 that even if the two sides had swapped planes, the Israelis would still have won the air war.
Thinking about these three apparently disconnected ideas (management experiments on workers, people/process trade-offs in information work, and people vs. tools in skilled physical work), it struck me that the key point in each case is the effect of paying mindful attention to the means being employed.
In traditional industrial thinking, management is the tool user and (relatively unskilled) labor is the tool. In information work, educated humans are the tool users and abstract management processes are their tools. In skilled work, like fighting with a class of weapons, the tool is literally a physical tool.
The phenomenology in each case appears different, but it isn’t really. When you pay attention to tools, you trigger deliberate practice and mindful learning.
Of course, you cannot take this too literally. The human instrument, abstract process, or physical tool have to have certain minimum capabilities. Assuming General Hod was correct, you still cannot assume that the Israelis would have won with Wright-era biplanes. Even the low-skill workers of the industrial age who replaced the higher-skill craftsmen had to have certain minimal skills.
So we can formulate a generalized version of the Hawthorne Effect: the effectiveness of a tool depends almost entirely on the amount of mindful attention being devoted in its use, not the specific form the attention takes.
I haven’t fully thought this through, but I figured I’d share what I have so far.
I believe there is a necessary condition here. For a tool to behave this way, it must be fundamentally finite and limited in its intelligence, or be used in a way that restricts its intelligence (or to make it more geekily precise, it cannot be an unfettered Turing-complete tool that might start thinking for itself in open-ended ways). Indefinitely increasing returns in the effectiveness of its use must come from the user. The tool needs to embody a certain minimum amount of intelligence (i.e. the design of a screwdriver embodies some minimal thinking about screws), but it cannot be generally intelligent. The reason is simply that generally intelligent instruments might have their own conflicting intentions about what to do, like Potterverse magic wands.
The original Hawthorne Effect is interesting precisely because the tools are not naturally finite or limited, but were made artificially so. For general attention to have positive effects, the tools must be somewhat dumb. The speculative psychology suggested as the mechanism is consistent with this idea: trusting management’s concern for its welfare makes labor more compliant and willing to be used in dumb-tool ways. They voluntarily shut off their critical thinking and become submissives.
Computers are today’s labor: capable of vastly more than they are normally constrained to do. More on that in a minute.
What about abstract processes that fill the pages of business books? Increasingly, I am turning into a radical minimalist as far as these are concerned. I think there are only two “minimum necessary tools” for all pure business thinking: written language and book-keeping (in the former, I include informal white-board level drawings and photographs used for appreciative purposes, but not formal, prescriptive diagrams or algorithms).
From what I’ve read of his management style, Jeff Bezos seems to operate by a similarly radical philosophy (he demands briefings in narrative form and bans PowerPoint apparently).
So that suggests a certain difference between physical tools and abstract process tools. A high-precision CNC milling machine really is a more intelligent tool than a flint axe. The full complexity of the tool plays a functional role in the work.
But formal abstract processes (information work algorithms and prescriptive process diagrams) that constitute management “tooling” seem on much shakier ground. If language and book-keeping are really the only minimal tools needed, then anything more complicated can be viewed as a non-functional thought starter, the stone in the stone soup.
Of course, this is a very valuable function, but it is not the advertised function. It’s as if all the complexity of a CNC milling machine were merely about getting you to think about submicron-level precision, with flint axes being sufficient to actually make modern things. In fact the advertised function is actively dangerous. If the point of a formal process model is to get you thinking about breaking beyond its confines, thinking of it as a prescription to follow religiously makes it dangerous. It would be the equivalent of trying to make stone soup without suckering the villagers into contributing additional ingredients.
Minimalism is motivated precisely by concern over these risks. The fewer and simpler the tools you use, the fewer the chances that you might buy into dangerously over-complicated delusions.
If this radical position is in fact correct, then every abstract management “process” is basically a strawman thought starter/stone-in-stone-soup. The effectiveness comes from the catalysis of mindful management thinking. Or to cast it in a more positive light, good management models are necessarily appreciative views of management matters, not instruments for execution. Execution thinking can be done with basic writing, drawing and book-keeping skills. If I ever taught a minimalist MBA course, that’s all I’d teach.
A good test is this: all management can be conducted entirely using whiteboards, spreadsheets and email. If you think you need something more, you are either confusing some specialized variety of engineering with management, or you are buying into some sort of management fad. This is why PowerPoint is so dangerous: it is a Trojan horse medium that allows you to sneak in all sorts of other crud. It is a steroid booster to faddish thinking. These days, I find that I increasingly make minimalist slide decks as a result.
Computers present an interesting challenge. Like the industrial laborers of a century ago, they are mostly used in severely hamstrung ways today. A class of autonomous “information worker” computers has not yet arisen except in narrow areas like high frequency trading, where they already do what humans cannot.
Increasingly, we will need a theory of such “white collar computers.” I think all we need to do is teach them to write, draw and do autonomous management accounting. If we want to prevent the Singularity from happening, we need to make sure we teach these white collar computers all the management “theories” of the last 60 years. That’ll keep ’em spinning their wheels for a while, buying John Connor some extra time. Skynet applying Lean Six Sigma to its operations should be easy to defeat.
Looking back, I’ve been getting slowly radicalized this way for a while now. This is why I believe the Turpentine Effect is a pathology: it is often motivated by the desire to make the wrong sorts of over-intelligent tools that attempt to capture generalized intelligence instead of attempting to create things that catalyze mindful attention.
Good toolmakers intuitively understand the generalized Hawthorne Effect, and realize that minimalist tools are the safest.
Or maybe I am just a management Luddite with a romantic view of pre-modern management methods.