The Tao of Setting Goals
Recently, as my business has grown, I have been thinking a lot about making goals. Why do we do it? Is there a way to do it more effectively?
After toying with a number of options, I have settled on the following basic model for goal setting as the baseline:
Goals as a form of Executive Control
In a business context, that means goals are a way for management to set expectations, and thereby influence the behavior of workers. It’s part of the top-down model:
In one of the most famous papers ever published in the field of economics, Ronald Coase once asked why economic planning exists within firms, even if a free market prevails in the larger economy. A market economy is based on free transactions between buyers and sellers. But within a firm, managers don’t negotiate individual behaviors on a day-to-day basis. Instead they typically enter a long term relationship in which the employer agrees to pay the employee a fixed amount and the employee agrees to follow directions.
Coase get’s right to the point with this conclusion:
The main reason why it is profitable to establish a firm would seem to be that there is a cost of using the price mechanism. The most obvious cost of ‘organizing’ production through the price mechanism is that of discovering what the relevant prices are.
Corporate goals need to be understood in this context. They are an alternative to a free market where every transaction has a price. The reason they are used is that measuring the output of employees is desirable, but difficult. Dilbert illustrates the concept well:
If employee behavior were simple and easy to measure, we wouldn’t use goals. In fact, we might not even have employees. We would just hire contractors to do piece work.
So goals exist in this world where executives need to exert control over the behavior of their employees, but the desired behavior is complex and interdependent. The situation is also complicated because, unlike the hierarchical diagram above, in many cases the employee has information about the quality of the work that is not easily available to the employer.
Personal Goals
One way to understand personal goals is to just make an analogy to corporate goals. Our conscious rational mind is the management, and the unruly rest of us is the labor. The ability to exert rational control over ourselves is called Executive Functioning. Effective goal setting is a part of executive functioning.
Interestingly, executive functioning is particularly important when things are complex. From the Wikipedia article linked to above:
The executive system is thought to be heavily involved in handling novel situations outside the domain of some of our ‘automatic’ psychological processes that could be explained by the reproduction of learned schemas or set behaviors.
A common example of a personal goal is to lose weight. We need goals to lose weight because eating is such an automatic process and given the abundance of food available to most people, it can be natural to overdo it.
However, unlike the corporate goal setting context, our rational mind can’t really threaten to fire our basic desires if they don’t fall in line. So with personal goals I usually have to resort to tricks. By tricks I mean doing ‘easy’ things that commit me to doing something harder. Our rational mind might have more control over certain behaviors (like buying groceries) than it does over others (like opening the fridge and grabbing a snack). So, we plan out and perform all the easy behaviors that will help us make the hard decisions.
Toxic Goals
I guess I have always known that setting and sticking to goals was a hard thing. But more recently I have learned that goal setting can actually be harmful. This can happen both in the corporate context and the personal context.
One of the most obvious ways that corporate goal setting can go wrong is that it can establish metrics that motivate the wrong behavior. The Dilbert comic above is great example. Another example comes from the history of Soviet command-style economy from the book Entrepreneurship and Economic Progress:
There are many stories about problems in the economy of the former Soviet Union that arose because output was measured in physical terms rather than in terms of value. For example, a factory making nails was given a quota in terms of tons of nails produced, and produced only very large nails because it is cheaper to produce a ton of big nails than a ton of small nails. Planners, recognnizing that there was a shortage of small nails, changed the factory’s quota to produce a certain number of nails rather than a certain weight, with the result being that they produced lots of small nails but no large ones.
Harvard Business School did a study of all the potentially harmful effects of goal setting called Goals Gone Wild. Some of the main problems they found resulted from when goals are too narrow (like the Dilbert bugs or the Soviet nails) or too challenging (because they can result in risky or unethical behavior).
I think these issues are pretty closely related, and both go back to the inherent paradox underlying executive goal setting: goals are used in a context where you need to incentivize inherently complex behavior.
Of course, the other problem with goals is that they can be costly to create and observe. Goals can be viewed as one way to put a value on the work done by employees, and as Coase mentioned, there is a cost to using a price mechanism.
Personal goals can also be toxic. One of the biggest advocates for being careful about is Jamie Smart, who I think coined the term toxic goals:
Smart’s basic idea is that we shouldn’t use goals as a means of delaying happiness. His prototype of a toxic goal is one that looks like “I will be happy when…X”. In most cases, people set goals like this because they convince themselves that X is a prerequisite for being happy. But in other cases, people actually punish themselves for missing goals in a manner similar to how a manager might punish an employee for missing a quota.
The Tao of Goals
Goal setting should be viewed as a way of finding balance between two opposing forces. We have the rational, measurable, and strategic (but ultimately oversimplified) Yang, along with the complex, intuitive, and inexpressible power of Yin.
Goal setting would be simple if the best results came from maximizing the degree to which your goals were measurable and rational. It would also be easy if you could just do away with metrics and live free. But a true Master sets goals that are specific and measurable…but not too specific or too measurable.
One way to accomplish this is for the boss to simply make their wishes inscrutable, described by the great Deangelo Vickers :
Goals are a conversation between management and labor, simple and complex, rational and intuitive.
OKRs
There are a number of goal setting frameworks out there, but one that I’m using as a baseline for my business is called Objectives and Key Results (OKRs). The basic model of OKRs is to choose an overall objective and then identify 3–5 measurable key results that you think will lead to your objective. The method comes from a Silicon Valley engineering culture that tends to put a lot of faith in our ability to precisely define and systematize behavior.
For example, if your goal were to lose weight, you might choose some measurable intermediate results like walking 10,000 steps, or limiting calories to some specific daily threshold.
OKRs can be understood in the context of traditional corporate goal setting where a manager would sit an employee down once a year, rate them on how they met existing goals, and set some new ones. One problem with this kind of traditional goal setting is that annual reviews don’t necessarily provide enough course guidance along the way.
In some cases, an employee may not know exactly whether they are on the right track, but keeping them guessing isn’t always such a bad thing. In the words of Robert California, another great management Guru from The Office:
If I invited you to lunch I think you’re a winner, if I didn’t, I don’t. But I just met you all. Life is long, opinions change…Winners prove me right. Losers prove me wrong.
Honestly, this is not always a terrible management philosophy. To go off on a bit of a tangent, there is this fascinating form or artificial intelligence called a Generative Adverserial Network (GAN) in which two networks are trained at the same time. One of them is trained to perform a task, say, to create an image of a person. The other is trained to differentiate between fake pictures and real pictures.
The result is that one network gets really good at generating fake images (see, for example, the fake people generated on the website below).
The other network gets really good at determining if the image is fake. The ground truth training data is basically just binary knowledge about whether an image is fake or not.
So in this analogy, a boss who provides very high level, binary feedback (i.e., good job or bad job) is essentially training employees in two ways: one to do the task and the other to guess whether the task they are doing is the right one.
This is especially useful when it is really hard to explain what the task is (like trying to explain how to create an arbitrary fake picture of a human). It forces the employee themselves to figure out what they should be doing, and sometimes employees are actually in the best position to do that.
Anyway, OKRs are not like GAN. They don’t really rely on the employee to figure out the process themselves. Or at least they try to be a little more directive. Of course, at then end of the day every task is like a fractal that just gets more and more complex the closer you look at it.
Since setting goals and providing feedback is costly and can potentially be harmful, one reasonable goal setting philosophy would be to provide the least amount of guidance necessary for an employee to understand (or learn to understand) what they should be doing.
But again, that’s not really what OKRs are about. I think OKRs are born from the insight that sometimes its useful to be very clear about what kind of things create value. And they are useful precisely to the extent that value can be quantified.
For example, this website provides some neat guidance for implementing effective OKRs (courtesy of Google’s corporate OKR philosophy). One of the traps they point out is to choose low value OKRs:
OKRs must promise clear business value — otherwise, there’s no reason to expend resources doing them.
I am pretty sympathetic to the OKR perspective. For me, it is comforting to try and reduce a business problem to an engineering problem. Plus I probably just don’t have the balls to go full on Robert California. So for me and my business, OKRs it is. But we will doing them with the knowledge that the desire to rationalize and systematize can sometimes result in unintended consequences, and that it needs to be balanced by a recognition of the role of intuition and uncertainty.