Monday, August 23, 2010

Creating a self-reporting crime tracker?

Over the last few weeks, I've taken a keen interest in crowd sourcing and data visualization. Until the recent events in Chile, Haiti, and Pakistan (all examples of where these technologies are being used to better serve the population), I saw little value in mapping data in a visual way instead of the traditional approach of spreadsheets and graphs. Now, as I watch the use of the technology half a world away, I can think of several reasons why these systems far outpace anything a spreadsheet or other historical data presentation format could provide.

Let's take a fleet management system, for example. While a spreadsheet or statistical chart can show you where vehicles have been, you're always playing catch up, trying to calculate more efficient routes, eliminate redundancies, and coordinate pickups and deliveries to be as efficient as possible. If you use a real-time mapping system, that dynamic changes. With real-time mapping, you not only have access to historical route and location information, but you also have access to real-time data that tells you where each of your fleet is at this moment. That allows you to make on the fly changes, create tighter deadlines, and eliminated wasted time by knowing exactly what is happening when and where.

A few nights ago, while watching the news, I began thinking about crime and how real-time visualization coupled with historical data might help police in better allocating their already stretched to the breaking point resources. In many ways, police play the same catch up game the fleet companies I mentioned earlier do: they are always working from historical data and most have little to no access to real-time, citizen generated, data. Add to that the fact that many communities have a tenuous relationship with the police at best, so many crimes go unreported and, thus, don't make it to the end of month, quarter, or year data.

Over the last year, we've seen a massive rise in technology being used in crisis management. From the earthquake in Haiti to the floods in Pakistan, we're seeing real-time systems being used to better schedule resources and to track and plot vulnerable individuals in need of help. Using nothing more than a stack of open source software, mobile phones, and the Internet, volunteers using this technology are having an enormous impact in disaster situations and self-reported, real-time, statistical mapping information is right on the forefront.

Taking things a bit further, and out of the crisis management arena, this same technology is being used to monitor elections in Kenya, allocate healthcare resources to HIV positive mothers in Mozambique, and monitor human trafficking in the Congo. The uses for the technology are endless, limited only by the dedication and creativity of the project managers and developers.

So, while I was watching the news a few night ago, all of these projects all over the world began drifting through my head. How could the technology be used to reduce the burden on police and enhance their ability to meet citizen needs? As the idea began to take shape, I also realized that this could extend far beyond service delivery but could also provide valuable and detailed statistical information for crime researchers who could use the aggregate data to create better policing models and gain better insight into criminal activity and flash points.

With the goal of the system resolved, the next question I asked was what such a system should look like? Could the system (and should it) rely solely on internet connected citizens reporting crimes or will that run the risk of disconnecting a large subsection of the poor and elderly who might not have computers or internet access? What should be the options for reporting a crime and how would should the system go about validating incoming reports.

It seems obvious to me that the system would need to accept reports in the widest possible way. This should include SMS, MMS, internet, email, and Twitter. Call-in reports should also be integrated at some point but that doesn't have to be a priority right away.

Next, we come to the question of report verification. Is there an efficient way to verify the validity of each and every report? No. That means we're going to have to take most reports at face value. If it's reported, then we map it. If we later find that the event didn't actually happen (maybe through the use of crowd sourcing) then we remove it from the map. Not the best system out there, but one that requires the least human intervention.

The system should also have broad police support. It would be great if police would use it in real-time to augment their call-in reports. For example, if they see a sudden increase of breaking and entering reports in an area, they could proactively increase patrols, increasing the perception of police involvement in the community. This enables police to work smarter and citizens to be better protected. It's a win-win situation.

Now, for the tools. As a developer, my mouth begins watering when I start to think of the possibilities available for designing such a system. But the practical side of me says 'keep it simple and maintainable' so the system has to be kept to an absolute bare minimum that will allow expansion and functions according to expectations.

Thankfully, there are many open source tools available that just have to be cobbled together to make this system work. After reviewing a few of them, I believe this is all that is needed to create a fully functioning system:

Using the components above, a fully functioning crime reporting system could be created in less than a week. The system would allow seamless reporting regardless of user location or the method in which they choose to communicate. Since it's a web application, tying police agencies into the system is a simple matter of providing them an account on the system that gives them access to more detailed data while the public facing website provides only generalized aggregate data.

In the end, a functional system that serves an entire state could be built for under $1,000. Best of all, the system could be expanded easily and tied into other systems in other states or even a national system to provide a broad overview of criminal activity anywhere in the country, or even the world. The possibilities for such a system are endless and the costs are incredibly small.

Personally, while I'm looking at this system through local eyes, I could see the value of it to police in other countries, especially those where the government fights to keep control. Think for a moment of deploying this system in Mexico and allowing citizens to report drug activity and police to track it and respond to it in real time. Or perhaps in Iran where police are often the aggressors against the people and giving them the ability to report quickly and quietly would give humanitarian and amnesty groups a jump start on monitoring these oppressive regimes.

Giving citizens the power to self-report is an incredibly empowering thing. You don't have to wait for anyone to come to your door and ask questions, you don't have to fill out complicated forms. You just have to be able to send a text message or use the internet.

That is the true power of people in action.