Remember, the goal is to test your proposed solutions as efficiently as possible, not build something that’s bomb-proof. Nevertheless, you’ll want to be thoughtful about minimizing risks. A few tips:

1. Keep it simple. Test one or two hypotheses at a time, starting with the most critical.

1. Use volunteers. Don’t compel anyone to take part in your experiment.

1. Make it fun. Think of ways to gamify the experience.

1. Start in your own backyard. That will minimize the number of permissions you need and the risk that someone tells you to stop.

1. Run the new in parallel with the old. Don’t blow up the existing process until you’ve validated the new one.

1. Refine and retest. Create an expectation that this will be the first of many experiments.

1. Stay loyal to the problem. Don’t fall in love with your solution. If it doesn’t pan out, search for other testable hacks.

As Jim’s experience illustrates, the experimental method does not necessarily entail an orderly sequence of steps in which one side project leads logically to a next. Instead, small probes are often fragmentary and spontaneous, driven by unexpected opportunities and dynamic situations. Jim’s wife got pregnant and Alaska was out. What next? A different kind of experiment. Like Ben and Carol, Jim went for variety in designing his experiments. But the trend is clear: Small wins may be scattered, but what counts is that together they amount to a sense of progress—away from the stifling situation we are trying to escape.

Psychoanalyst Adam Phillips observes that “people tend to flirt only with serious things—madness, disaster, other people.” When we craft experiments, we are flirting with our many selves, a serious endeavor because it matters so much to us. The stronger the attraction, the more vulnerable we are to biases that affect how we perceive alternatives. Since we are not neutral about which outcome we prefer, we can fall into the trap of evaluating our experiments with a positive bias, one that encourages us to escalate commitment, even when we have evidence that it would be better to abandon ship or put the pet project on hold. A related danger is inadvertently putting a current work situation at risk. The exploration feels risk free, because we hide it from work associates. But the project becomes all-consuming, and it becomes obvious to everyone around us that our attention is divided.

The lesson was an important one: the things you choose not to do often matter as much as the things you choose to do. The real test of a person is the degree to which they are willing to nonconform to do the right thing.

When the stakes are high, and there are no take-backs, you want to decide at the last moment possible, and keep as many options on the table as you can while continuing to gather information.