Something quite unique happened last week in my home state of Minnesota. Following the release of the police file on the Jacob Wetterling case, the county sheriff—a guy named Don Gudmundson—did one of the most remarkable things: he publicly acknowledged that his department had botched things.

This is all big news here, but a little bit of background for the rest of you: Eleven-year-old Jacob Wetterling was abducted by gunpoint back in 1989 from a sleepy, country road while walking home from the store with his brother and a friend. It was the stuff of nightmares for parents. I was sixteen at the time and remember it vividly.

In 2016, after more than two decades of false leads and dead ends, one of the initial suspects—who was in jail on child porn charges—confessed to the murder and led investigators to the body. It was heartbreaking for many who held out the slimmest of hope that the worst hadn’t happened. Details of the confession suggest Jacob was murdered shortly after he was taken.

What I found incredible—and am still trying to process—is the press conference Gudmundson held. He had gone through the entire file and was not happy with what he found. He presented a powerpoint that gave a point-by-point breakdown of what went what wrong and what the lessons from it should be. It you want to see something truly remarkable, watch it.

I bring this up for one reason only: We must find a safe and productive way to talk about failures, especially government failures. There is a lot there to talk about.

This is not going to be an anti-government screed. The opposite, in fact: I want to defend government today. But I want to defend them by first acknowledging that governments, especially city governments, do a lot of things wrong. Devastatingly wrong.

Just recently, I spoke with a city manager who told me that a street project they are doing is a disaster that is going to set the city back decades. Will that city manager speak about this publicly? No chance.

As with many public projects, the path to getting it approved was difficult. Factions lined up, compromises were made, nobody is fully happy. Even though there has been turnover in the city council and many of the project boosters have lost their seats, there is an overwhelming sense that the good-soldier thing to do is to be a booster for the project. We need to put a positive spin on this, Chuck.

Here’s the problem: We have decades of really bad development practices we need to speak openly about. At the community level, this is going to be a neighborhood-level conversation. And that means it is going to get personal.

There is another part of this that’s equally important. We need to try new things. We can’t keep doing business the way we have been. When you try new things, you sometimes fail. That’s how wisdom is created. We need to invite government to strategically fail more often.

We—American society—have a dysfunctional relationship with local government. Many of us have unreasonable expectations of what government is capable of doing. We expect lots of services and low taxes and are ready to call out anyone who suggests otherwise. Many expect local government to do things we should be doing ourselves, like talking to our neighbors to work out conflict. Yet others expect local government to solve complex, systematic social problems, even when City Hall has not demonstrated basic competence in paying pensions, maintaining streets and collecting the trash. You know, the prerequisite stuff we all expect them to do.

Local governments all too often respond like… governments. Someone sends the wrong notice for a meeting; why not create a three-person signoff process for all future public notices? Someone complains about the placement of a parking bench; sounds like it’s time to form an executive committee of stakeholders to create a process for signing off on future park bench placement! Two neighbors argue over one leaving our their trash can too long after pickup; clearly we must create a ten-page ordinance and an enforcement regime to ensure all garbage cans are retrieved in a timely fashion.

There is a chicken and egg aspect to this. If people living in a city weren’t so demanding and unreasonable, cities wouldn’t hunker down. If cities didn’t hunker down and act so dumb, people wouldn’t get so unreasonable.

In an ideal world, we’d run governments a lot like (I have read) doctors handle surgeries. Or like we used to handle missions when I was in the Army. In both instances, there is a debrief afterwards. In the Army, rank went away during the debrief, and everyone was invited to talk openly about what worked, what didn’t work, and how to make things better. My understanding is that the same applies for doctors in post-op.

I know some cities do this, formally and informally. I think the tough part here is how to include elected officials and, most importantly, the public in this process of reflecting on both successes and failures. I think smaller projects will help. There are two reasons.

First: statistically, large projects are more likely to have cost overruns and critical failures. There are entire books on this; one of the best is Megaprojects and Risk: An Anatomy of Ambition. There is a truism about project size: The bigger the project becomes, the harder it is to hold one person accountable. There is a CYA aspect to large projects, in which committees of people report to other committees of people, and the location where the buck is stopping is removed from the technical details. By the time the project is over, the entire team has turned over two or three times. The megaproject structure defies good after-action review.

Second, however, is the opportunity on smaller projects to make the case for experimentation. To open up a dialogue with the public that brings them into the decision-making process, not to react but to co-design.

We tolerated Google giving us a beta version of Gmail because we knew that was part of the improvement process. They told us it was, and we see it change and get better each iteration. We are co-designing the next version of Gmail with Google right now, as we use the product. None of us is writing code—that’s not our part of the co-creation process—but it wouldn’t work as well if it were just delivered to us as a finished product, with no iterative steps along the way.

Local government needs to try more things, and that means they need more options to fail—and specifically, to fail small. To bring about a Strong Towns revolution, Americans need to become more tolerant of small, productive government failures that are quickly recovered and learned from. That will happen if, and when, government starts to deliver improvements iteratively and demonstrates the capacity to learn and improve with each iteration.

Sheriff Gudmundson is the interim sheriff. He was not around when the investigation took place. He is about to retire and has nothing personally to defend here. That puts him in a unique position. Here is something he said in a follow-up interview with the Minneapolis Star Tribune:

My hope is that in some small way that shining a light on the failures of law enforcement in this case will grant healing to victims, families, and the community,” he wrote. “I did not set out to find failure and errors in the file. They were glaring almost from Page 1. The truth is in those files from long ago and it pained me to see them …

“It was hard because for the first and only time in my life I publicly criticized other cops. I tried to present the facts in the case in a measured, careful and precise manner.”

We must get to the point where it doesn’t require brave people stepping out of the accepted norm for us to learn from our mistakes. We have too much learning to do.