The Burlington Vermont Police Department was stuck using an information system that was expensive and time consuming to use. The department needed a change but finding a new solution that fit their needs seemed almost impossible. In the middle of this, Burlington Police Chief Mike Schirling asked a critical question: What if we built this ourselves, based on our own needs and experience?
It has led him and the rest of the criminal justice system in Vermont down a unique path.
I spoke with Captain Schirling about his experience.
Rich Nadworny: What types of information are we talking about?
Mike Schirling: Contemporary policing involves managing information. Calls for service and crime reports come in to computer aided dispatch and records management. There is a vast amount of digital evidence that we need to enter and access - photographs to audio and video, statements, reports, and affidavits.
RN: Why did you ultimately co-develop the new system?
MS: The first thing we did was look around to see if there was anything that stood out in the marketplace that met our needs, and there wasn't. I talked with a couple of people who were involved in software development, bought a book called "Getting Real" about agile software development.
We decided to learn more about the process of agile software development so we hired a Ruby on Rails developer in Colchester Vermont to train us. He taught us, among other things, how to create "user stories" from our various specialized groups. We brought our team together and started making various stories or paths for the way we needed the system to work.
RN: How did you find someone to build this for you?
MS: We put out two very streamlined requests in the form of an RFI and later, an RFP. The responses that came back were expensive. Some of them were as much as $3 million and required up to three-year build period. At that point, I was thinking it probably wasn't going to work, because the risk was too high.
We tried one more thing: I found a wireframing, prototyping application and rapidly developed a working model based on a simple paper form called a "crime sheet." Working with the basic information that we collect on every response, coupled with the user stories and workflows that we developed, we wireframed a workflow for the software.
RN: Once you had something that people could respond to, how did that change your conversations, both with developers and among your team?
MS: We had a conference call with each of the respondents and we walked them through the wireframes. So we said, "Listen, what you're proposing is too big, bulky, and expensive, let us walk you through the simplicity we want to achieve." By the end of two weeks of conference calls, the company that ended up getting the job -- CrossWind Technologies --had built a working Web prototype that we could log in and use.
Once the software coding began, we brought in our team that created those user stories to do the rapid development. Rapid development was new to everyone, but the advantage that we have as a police department is weird: we've got computer crime and forensic background and we have a building full of investigators who are used to going through methodical process to come to a conclusion. We found striking the parallels between the software development processes to what we do on a day-to-day basis when we're working on major case. We have people trained to do investigative work, and they just had to translate that to an agile software development process.
Of course, when we shared the wireframes with our larger group they said, "No, that doesn't belong there, that belongs here, what if you did this?" Once you start to show people what it looks like- when you can see it on a computer screen, it's not a piece of graph paper, and you can push a button, and actually a dropdown will occur, or it'll take you on a hyperlink somewhere - people react to it right away.
They start thinking, "Well, what if you did this and put that there?" One of the most interesting parts of the process was that the further we got into the actual development and the coding, the more ideas people had.
Once a week we had a live call with the developers where we tested and gave feedback on the working prototypes. In the intervening week, they would push at least one update, sometimes more.
After a few months, we brought in other agencies from around Burlington, from Winooski, Colchester, South Burlington and the University of Vermont who had not been involved in the development process to help us with the testing.
After nine months, we flipped the switch and it's been live ever since.
RN: How much did all of this cost and who's using it?
MS: In Vermont, there are 31 police departments and over 100 public safety agencies using this system.
The total initial cost of development was about $80,000, which is about 20% less than the yearly licensing cost for the previous system. This was possible only because we had become co-developers of the system. We've cut the time needed for staff to manage information in half. We had a specific savings goal we were aiming for over 10 years, but we now anticipate reaching that goal in 5 years.
Our co-developer Crosswind has the rights to the source code outside of Vermont. We provided the subject matter expertise, they provided the programming expertise. Think about it, there are 18,000 police departments across the U.S. who could use this system.
RN: And how does it work, out in the field?
MS: This is a solution tailored to the 21st century: We're working toward evidence-based or data-driven policing, and data-driven decision-making. The future version of that is predictive policing - running algorithms against information to figure out where things are most likely to happen in the future.
Now we do paperless case reporting to the prosecutor's office. Once we put in the entire case file, they can access it. It's the only system we're aware of in the world that does paperless electronic discovery to defense attorneys as well.
It has enabled us to cut down the amount of work for doing discovery -- "I need this CD, I need this recording, I need that photograph, I need this report" -- it's all gone, because you log on and you see everything we see.
They get the entire report.
We never get to build anything, so this was pretty cool for all of the folks who worked and continue to work on it. It worked, and it demonstrates that the agile development process can be a viable model for government to approach software or IT projects.