Getting the police to think like a startup

Thief-catching involves a lot of technology — but as an organization, the London Met isn’t very good at adopting new tech.

Back in 2013, I in Hackney, London. As a bit of a nerd as well, in April, I got involved with .

Image for post
Image for post
Yes, thats a Lotus Exige police car. No, they didn’t let me drive it. Instead, they tended to hand me the keys to a lovely Vauxhall Astra. It’s… Not particularly glamorous.

It was an interesting challenge; I wasn’t much of a coder (and, frankly, I’m still not — although I do know my way around a programming language or two), and they don’t usually let me near any of the felt-tip pens. Nonetheless, I figured I could use what I know about the police, and about the methodologies of building user-first software, to contribute in my own way.

The Met is a great organisation, but it has a few challenges around technology. The key one, in my mind, is that they are spending a staggering amount of money for old tech — again and again — so I decided to demonstrate the concept of a rapidly iterating Minimum Viable Product to the management of the police.

One example where there is huge costs involved, is the Mobile Data Terminals (MDT) that are installed in police cars. Though various Freedom of Information requests and conversations with police officers at the hack day, it turned out that the hardware alone costs between £4,000 and £6,000 per unit. There are at least 700 vehicles in the Met that operate MDTs — that’s £2.8–4.2m worth of hardware installed in Met vehicles. Which, given what the MDTs actually do, is pretty outrageous.

Replacing these units with iPads would do some interesting things. At £370 per unit, measuring 700 iPads in millions of pounds becomes a silly-looking figure: It’s £0.26m. I don’t know how long a police car stays in service, but I’m willing to bet that the huge difference in cost would be noticeable in budgets very quickly.

There are several steps to cutting cost; Using commodity hardware instead of custom-made MDTs is an obvious first step, but there are also enormous savings to be had in relation to processes, especially in re-considering how software and systems is developed for the Met, for example

Prototype fast. Fail faster. Succeed earlier.

Since I’m a big fan of lean start-up methodologies, I decided that my demo for the wasn’t going to be a finished product (how could it be, in 48 hours worth of coding). The product itself was never meant to be complete — instead, it was a process demonstration of how to use design thinking and rapid iterations to build a testable prototype of a product.

The goal was to create a scenario that front-line officers are extremely familiar with: being on the street, needing to check details on a person or vehicle. The point was to do a demonstration of how you can gain valuable feedback and information from poice officers by developing a Minimum Viable Product (MVP). Think of it like a highly advanced and very visual , that gives the illusion of being a working product. As such, when you are looking for feedback and validation for an idea.

Why would you bother with a prototype?

To understand why prototyping is important, it’s useful to know how government procurement often works in situations like this.

What the Met would usually do is to spec a full MDT system. 18 months later, it is delivered. Then it gets deployed, the officers are trained, and everything is gravy. Except it’s 18 months and several million pounds later. Some organisations are willing to accept that sort of investment and risk; as a startup founder, however, it scares me: There’s no way I would even consider working in the dark for 18 months, and investing several million pounds into an important project.

So… How can you de-risk a project that is so large, and so core to front-line policing? What would a start-up do to get early product validation and a tight feedback loop to ensure that if you are running at full speed, at least you’re running in roughly the right direction?

Thinking Minimum Viable Product (MVP)

As a police officer, you sometimes have to check people’s details — to see whether they are wanted by police, for example. At present, the procedure is as follows: You use your radio, go on the support channel and hail an operator. You’re then put in a queue, which can sometimes be pretty long. Then you use the phonetic alphabet to read the person’s details (“Last Name Sierra Mike India Tango Hotel. First name Mike India Charlie Hotel Alpha Echo Lima Alpha, Female, date of birth 20 June 1988”). They look it up in a computer, and tell you the results.

In summary: You use a radio to talk to someone who types something into a computer, and then reads out what’s on the screen in front of them. In my mind, that’s like calling your sister, who is a nurse in a busy emergency medicine unit at a hospital, asking her to take time out of her day to Google something for you.

So, my MVP — the smallest possible slice of the Mobile Data Terminal project — is to replace the most common requests done over the radio with an app.

The basic idea is to find an answer to the following statement: “Yes, we do want to replace everything about the MDT. Simultaneously, we want to give you something that can be completed as quickly as possible. If I were to give you an app that does X, would you use it?”. If the answer is ‘yes’, then you go build it. If the answer is ‘no’, then you scrap the idea. If the answer is ‘yes, but how about you do Y and Z instead? That would make more sense to me as a front-line officer’, then you are really onto something: You will be building the right thing, but better than you would have without proper feedback.

My ‘app’ is a type of MVP — a prototype. The idea is to create a highly realistic mock-up of how an app would work, and talk to a series of police officers. I would talk them through a very visual and hands-on mock-up of how something could work.

Can I see the mockup?

I thought you’d never ask!

Launch the app, and admire the fantastic splash screen. :

Image for post
Image for post
Oh yes.
Image for post
Image for post
You’re then taken to the home screen, which enables you to choose the mode you want to use. In the demo, only the top 4 menu options do anything.

Vehicle checks

Let’s take a closer look at ‘vehicle check’…

Image for post
Image for post
You’d be taken to the ‘what to scan’ screen. ANPR mode would be a running mode for doing continuous . Instead, we’re going to take a look at a single vehicle, so click that button.
Image for post
Image for post
Obviously, this is a mock-up — in the ‘real’ version, this’d be a camera app, much like the one you’d find in your phone. Instead, however, you’d also have OCR — Optical Character Recognition — built in.

Now, after the photo has been taken, the app has gone to the server, and it has discovered that there’s something fishy about the car you just scanned. Instead of showing you any other details, it goes straight into ‘officer safety’ mode, and flags up the relevant details right away. In this case, it flags that this vehicle has certain warning codes associated with it, and should only be stopped with Trojan assistance (i.e. armed police) as a result.

Image for post
Image for post
On the other hand, the last incident is quite a long time ago, so the green dot signifies that this information might be out of date. As such, it’s down to the officers to do a dynamic risk assessment, and decide how to handle this particular case.
Image for post
Image for post
The user can now click Full Results to get the full results of the details. It shows the picture of the car…
Image for post
Image for post
And prompts the user to tick which databases you want to query.
Image for post
Image for post
In this case, let’s check all of them…
Image for post
Image for post
Press ‘go’, and the details come back right away. In this case, the registered keeper is Steven Smith. The vehicle is insured and taxed, and has three named drivers.

One of the drivers have a red dot next to their name, signifying that there’s something odd going on with that person. Perhaps they have a driving ban, or maybe there is something else you need to know about them before stopping the vehicle. In this case, since the red dot is next to the only female on the list; if the driver is female, that should increase the risk assessment.

Image for post
Image for post
Click on the person’s name to check them. In this case, it is shown that this person is known to the Police National Computer, with several warning flashes.

Person checks

Go back to the main menu and click ‘Person Check’. You’re now taken to a menu where you can choose what you want to scan — a passport, driving licence, or just manually enter details

Image for post
Image for post
Image for post
Image for post
Click Driving Licence, and take a photo of the D/L. The app then returns the fully OCR’ed data
Image for post
Image for post
When you scroll down, you can correct any details that are incorrect, before sending the request off to the relevant databases…
Image for post
Image for post
In this case, let’s check PNC, DVLA, and Electoral roll data:

The details came back… And should look familiar, because this is the same screen as you got via the other route.

Nothing about this prototype is rocket science, and obviously all of it is simulated. Building the connections to the Police National Computer etc is the tricky part, from a security and policy point of view — and, given that the PNC is an ancient piece of software running on scarily old servers — it is non-trivial to implement from a technical aspect, too.

What’s the benefit?

I demoed this simple prototype to a load of officers throughout the week-end, and was able to make a large number of adjustments as I was iterating through feedback from the officers.

One example: One of them said “Why can’t I check as well?” The beauty of using rapid prototyping is exactly moments like that. In my mock-up, I am able to add 5 lines of code, and I’ll have added a slider for Crimint.

Compare that with the risk of finding out at the end of a 18-month project, that an officer would like to be able to check Crimint. Adding it into a ‘finished’ product is a much larger task, with huge costs associated. The flipside is also true: What if you included an interface for checking a particular system, that no front-line officer would ever use? In that case, you’ll have spent time and money developing something nobody will ever use. In start-up world, that would be an unacceptable level of waste — and in police world, the same rules should apply.

Ultimately, the point is that building prototypes and putting it in front of end users as early as possible in the process helps refine the product development cycle.

Haje is a founder coach, working with a small, select number of startup founders to build exciting, robust organizations that can stand the test of time. Find out more at . You can also find Haje on and .

Written by

CEO of Konf, pitch coach for startups, enthusiastic dabbler in photography.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store