Upstream: How to Solve Problems Before They Happen by Dan Heath

The key ideas here involve three kinds of ghost victories. 1) Your measures show you are succeeding, but the cause of your success is something external that you didn’t do. 2) You succeed in your short-term measures, but they don’t align with your long-term mission. 3) Your short-term measure become the mission and undermines your work.
An example involves fixing sidewalks in Boston. They used 311 phone complaints and the number of repairs as measures. Unfortunately, most of the 311 calls came from wealthy neighborhoods where repairs were easy. The short-term measure looked good, but the repairs that would make the most difference weren’t getting done. This is also an example of when faced with a difficult question, “where are the worst sidewalks?”, we substitute an easier one, “who is complaining?”
When you create short-term measures expect people to game them. An example is asking schools to increase the graduation rate. This was an easy game for some principals as the just told the teachers to grade easier. Essentially the measure didn’t match the mission. Better measures would have been more students taking AP courses, attendance, and better SAT/ACT scores. When you set up short-term measures you need to pre-game them by asking how the measures could be misused. There are also examples involving crime reduction and health care here.

Just about any innovation comes with unintended consequences. A classic example is the government of India offering a bounty on cobras to reduce the cobra population. This resulted in some people farming cobras. When the government picked up on this they stopped the bounty and the cobra farmers let their cobras go, thus increasing the cobra population.

What you need to do before instituting any system change is to have everyone involved consider negative consequences that might result. Try to zoom out so you see the big picture, not just the part of the system you are tinkering with. Dan doesn’t use the term pre mortem, but I think it applies. You also need to set up a feedback loop so you will know in near real-time the impact of your change. Also, be sure to look at how easy it will be to undo a change if it turns out to be harmful. Examples used here include the impact on face to face interactions in open offices and efforts to ban single-use plastic bags in stores.

When it comes to health care we pay very little for prevention and a lot for dealing with problems once they present themselves. Preventive efforts have radically increased life expectancy, however. If no one gets sick they will cut the budget. If doctors get paid for something, like ordering an MRI, they will do more of it. Some systems are experimenting with ideas like paying doctors on a per-patient basis no matter what tests and procedures are ordered.
Here we are introduced to the single pocket concept. The example given is a restaurant owner who is getting complaints about cold waffles. She finds out that her staff was making waffles ahead of time as the waffle maker couldn’t keep up with lunchtime demand so she bought another waffle maker. The cost came out of her pocket and increased revenue went into that same pocket.
Now consider the wrong pocket concept. The example here is a program that assigns a nurse to a poor woman who has just become pregnant. The nurse mentors the person until the child is two years old. This results in reductions in smoking while pregnant, preterm births, infant mortality, child abuse, criminal offenses by the mother, closely spaced pregnancies, and food stamp costs. Unfortunately, the winners here include Medicaid, the penal system, and the food stamp program, not the agency providing the service.

Book Summaries