Fixing people vs. fixing processes

Only one is a realistic course of action

I have a colleague who is keen to find fault with various people in the organization, identifying ways in which they are falling short of what they should be doing. I find his analysis valuable, but I usually disagree with his proposed solutions. His preferred approach typically involves changing the character of the identified people. He seems to experience these problems as moral issues, with the perpetrators being defective people who need to be repaired. He doesn’t describe them as sinners who need to be saved, but if he had a different cultural background, I think he might.

The limits of reforming people

I also want to find ways to have the organization work better, but I’m more pragmatic than my colleague. I see at least a couple of flaws in his view of these situations:

1. It’s essentially never possible to reform people in the way that he wants. Even when people genuinely want to change, it’s sometimes surprisingly hard. The idea that we will get to tweak our co-workers’ attitudes to conform with our managerial preferences seems outlandish.

2. Even if we had an occasional success in the reformation efforts he envisions, it seems unlikely that we would achieve our overall goals.

To elaborate on the second point, let’s suppose we work on reforming people to deal with some problem X, and let’s further suppose an impressively high success rate of 20%. That immediately means that our effort didn’t necessarily make a big difference even to that one problem. Worse is that there’s no reason to believe a priori that our expensive investment in reform will also help us with some subsequent different problem Y.

The summary here is that we invested considerable time and effort to make only a small improvement for one pressing concern, while making no difference at all for subsequent problems. If our problems were few, and stable, this kind of investment might make sense; but in a highly dynamic environment, this approach isn’t workable.

My colleague is certainly not the first person I’ve encountered who is in the grip of this reform fallacy. Over time, I’ve come to use the phrase “trying to reupholster souls” to refer to this form of wishful thinking. There are at least three problems involved in trying to reupholster other people’s souls:

1. Reupholstery is inherently hard work, even assuming you are an expert at the task and have ready access to all the relevant supplies;

2. No one knows how or where to locate the relevant person’s soul, so that it can be replaced; and

3. The person in question almost certainly doesn’t grant their consent or access for the activity.

The role of process

It’s with this context that one can develop an appreciation for processes in organizations. The most admirable processes may be honored as “rule of law,” while grudgingly accepted processes may be denigrated as “bureaucracy.” What they share is the attempt to remove individual character differences, or personal circumstances, from how the work happens. In a business context, what we want is for the “right thing” to happen regardless of whether the people involved are “good people.” This kind of impersonal rule-following resembles how computers work, but organizational processes are much older. Indeed, the origins of bureaucracy are intertwined with the origins of writing. In its oldest extant examples, writing is used for prosaic recordkeeping… not literary self-expression. It’s kind of wild to realize that bureaucracy is ancient, long predating many aspects of modern culture.

It can be helpful to compare the organizational version of a process with the computational version of a process. In both cases there’s some notion of inputs, outputs, and steps to be taken to produce the outputs from the inputs. A computational process is fully specified, and happens in a completely mechanical way. That utter lack of discretion is, for better or worse, a core part of what it means to execute a process on a computer. In contrast, part of what’s necessary in an organizational process is some role for discretion. As noted previously, the world is dynamic and unpredictable. Those characteristics mean that judgment and flexibility are likely to be required from time to time. Refusing to apply sensible judgment or flexibility (working to rule) is a form of protest, about which I’ve written previously.

My usual advice to my reform-minded colleague is that we should look at process modifications instead of trying to fix our co-workers. Can we introduce processes, or extend or modify processes, to address the problem? In principle, we can then confine ourselves to the questions, “can our co-workers follow these processes?” and/or, “do we need different co-workers who will follow the processes?” rather than viewing our problems as somehow rooted in the moral failures or personal limitations of particular people.

Is process oppressive?

The usual problem with a focus on process is the potential for dehumanization. It’s worth thinking a little about why being human is in some sense opposed to being part of a process. If things get to be very small and nitpicky, with little connection to a larger whole or to meaningful choice, most people will feel oppressed. Likewise if we have systems that demand ever higher “productivity” in terms of the number of tasks accomplished, or ever-shortening deadlines. This is the problem with an assembly line, for example, as well as the kind of fine-grained time-on-task job design advocated by Frederick Taylor and his followers. (Interestingly, neurodivergent people are probably less bothered by this task-granularity issue than neurotypical people… but it’s not clear exactly how to take advantage of that insight without causing other problems.)

All of that acknowledged, there is usually a relatively large design space of acceptable task sizes and timing. If we’re in that realm of reasonable work, introducing or improving a process is likely to be a more effective path for organizational improvement than reforming people.

I’m not saying that process change is a panacea. It’s still better to have the right people doing the work than having the wrong people doing the work. Additionally, there are many ways that people can improve that are both reasonable and sensible: different forms of training or education, for example, often have implicit or explicit goals of improving skills and capabilities. Sometimes performance is impaired by something happening in a worker’s life outside work; understanding and support may be what’s required in those cases. But when there’s something that seems more like a group of workers “don’t care enough” or “don’t pay enough attention,” I doubt that “fixing” the workers is the right approach.

Choices

Instead, the choice seems to be between (1) improving the process somehow, (2) replacing the worker, or (3) accepting and living with the problem. All too often, a “bad seeming” person is oppressed by a badly designed process or no process. Particularly if the work environment changes from Theory X (people are basically shirkers and must be compelled to work) into Theory Y (people basically want to do a good job and should be supported in their work), the improvement in attitude and morale can be startling, even though it’s the same people doing the same work.

If we think in terms of a simple two-element four-state model, we are essentially dealing with where we are in the space between the best [right people, right process] and the worst [wrong people, wrong process]. Within this model, my colleague is tending toward an assessment of [wrong people, right process] while I’m leaning toward [right people, wrong process].

When considering where we are in this design space and how to get where we want to be, we must remember that we can probably improve the process more easily than we can improve the people. In general, people can be swapped in and out of roles as a way of getting the right people, but the cost of such changes is relatively high. The high cost arises because there is unavoidable disruption, uncertainty about the likelihood of success (the replacement may or may not be better), and uncertainty about the length of time required to establish success. Accordingly, the people-swapping approach only starts to look appealing (1) for situations where there is a severe mismatch between worker and job, and the mismatch looks unlikely to improve, and (2) for situations where process changes have been attempted without success.

My colleague (and others with similar ideas) imagine that they can make “small adjustments” to improve our existing co-workers. They expect these changes will yield better performance without replacing anyone or changing any processes… but I think that’s unlikely to be successful. If the actual choices available are only to change processes or replace people, process changes are usually the most effective way to address concerns about failures or breakdowns.