ArriveCAN hot takes miss the point

This kinda blew up over on Twitter, so I thought I’d share it here with a bit more detail.

Yes, ArriveCAN cost a lot of money. Way too much.

(It was a pandemic.)

Going way over budget happens all the time in enterprise. It’s often the developers’ fault. It even happens outside global public health emergencies, believe it or not.

To prove a point, a couple of development shops are spending the weekend demonstrating that they can build the app faster, better, and cheaper than government can.

These dev shops are building an app in idealized conditions to prove a point. The app they release—unclear to whom, or on what platforms—will not be rolled out to dozens of airports and thousands of employees and millions of travellers over multiple updates. It will be a Potemkin app, a facade that oversimplifies and stigmatizes some really important issues we should be talking about a lot more.

I avoid political rants because as the co-founder of FWD50 , I try hard to be nonpartisan (the FWD part means "neither left nor right, but forward.) But I couldn’t stay silent on this one because it is literally why we started the conference. And unlike many topics, I feel like I’m qualified to have an opinion on this one. I have a lot of experience with startups and enterprise software, and for the last 7 years I’ve had the great fortune to speak with over 600 digital government folks from 43 countries.

What it takes to build an app in government

There’s a lot involved in building apps in government. Off the top of my head (and I originally typed this list at 2AM on my phone, so I’m sure it could be much longer) this includes:

1. A TON of governance to be done when building any app, particularly one offering a service that literally launched a cross-country truck protest and shut down the capital for a week. Remember how mad everyone was that the government was trying to steal their data? That leads to laws, and regulations. That takes time and money and committees. Public consultations.
2. Training. Of the people who will run it; and the people who will use it. Including airline staff and border agents. There’s also documentation on both sides. Actual humans in busy places like security lines and check-in counters.
3. Interfacing with all the back-office stuff, like a password lookup, or passport details, or airline flight timetables. You know, across borders and jurisdictions. With different rules and laws. About personal medical, travel, and identity data. Easy, right?
4. Accessibility. For-profit companies target the lucrative middle of the market; governments are for all, and have to make it accessible to everyone. People using screen magnifiers need to be able to read the app. Private developer shops charge extra for this stuff, and only do it if it’s in the spec. Governments don’t get to choose that (nor should they be expected to.)
5. Language. Not just translation, but fine-checking the wording to make sure it complies with regulations. Because one embarrassing mis-use of a pronoun somewhere will lead to some politician getting yelled at and some public servant getting thrown under the bus.
6. Procurement, and licensing, with all kinds of legal terms and Most-Favoured-Nation clauses and escrow and intellectual property assignment.
7. Deployment, DevOps, testing the scripts, not just for production, but for a parallel development platform on which to test the next version (because ArriveCAN was being updated and modified for 18 months.)
8. Backups. Source code repositories. Oh, and having a physical alternative if the digital one goes down. With recovery processes. You know, a backup plan. For, you know, all cross-border travel.
9. It has to work in a lot of degraded conditions. The app will be used, then immediately disconnected from the Internet (it’s called Airplane Mode for a reason.) And when the passenger gets to their destination, they may not have a data plan. Cache and synchronize everything.
10. User testing. On a wide range of phones and system language settings. That includes testing camera's ability to scan, and the QR code you scan on arrival, and the long pauses while someone finds their flight number while in line, and the database their personal data is being stored in.
11. These developer shops can just copy every interface and screen from the existing app, or at the very least the wording, navigation, fields of every drop-down. The ArriveCAN app had to be tailored. The folks cloning the app get to skip the discovery phase. They stole the plans.

I promise you, there are so many damn things going on behind the scenes in government software development—obstacles and rules and legacy systems, en deux langues. Old coding environments. Control over what code you can include and what frameworks you can use.

I am not saying ArriveCAN did all the things I listed above, or did them perfectly.

To make matters worse, there aren’t a lot of good tech managers in the public sector. Many people in the public sector simply do not get “digital.” The Peter Principle often applies. Moreover, they needed to find subcontractors, right when everyone else was hiring developers for better pay. It’s hard to retain talent when all you can promise is uncompetitive salaries, outdated tech stacks, and a wave of public vitriol surrounding a hot-button project.

Yes, government is slow

Government is often ponderously slow. And while a little predictability comes at the cost of innovation, sometimes slow and boring is a feature, not a bug. But I absolutely think we should speed up government a bit, invest in retraining, hiring talent, and changing how goals are set and employees are incentivized. I believe this so much I launched a conference on it and devote much of my life to bringing the people who have done these things to a stage where government employees can learn from them.

People are slagging on ArriveCAN. Was it a beautiful app? That’s a matter of taste. It worked, which is more than I can say for the Federal payroll system.

Personally, I kinda liked it. I’ve used it a bunch. It never crashed, I was never confused about what to enter, and it worked with things like document uploads. It did things like remember your state when you moved back a page, despite having to satisfy a nightmarish list of stakeholders, often with conflicting demands. By comparison, when I click back on Air Canada's website, I get an error message saying I pressed back too many times and need to restart my search.

To recap:

• A highly visible app that pissed off a small, but incredibly vocal, minority of Canadians built during a pandemic went way over budget, partly because of contractor hell.
• Some developer shops with none of government’s baggage are saying they could build it in a weekend.
• It’s not clear what they’re trying to prove other that it’s easy to clone a UI when someone’s done all the hard parts already.

It’s naive to compare apples to orchards. An app alone is not a solution. This is a PR stunt that serves two interests: Development shops wanting visibility (and perhaps rightly so, since procurement processes in the public sector do not favour small, agile developer teams); and political adversaries looking for punching bags. Reporters I trust and respect are publishing irresponsible hot takes for clicks, designed to make us mad.

The thing we should be mad about

The thing is, we should be mad: Why does it cost so much more for government to build an app? Why is it so hard, and expensive, and slow? Why do we see disasters like healthcare.gov and Phoenix and other trainwrecks so bad they have their own Wikipedia pages?

The short answer is that government work has changed dramatically, and government has not.

Governments are, at their core, information systems. While the edges of government are physical things like building roads and testing crops and staffing service desks and protecting coastlines, the squishy inside of government is pure information. Governments are the OG of Information Technology. It was governments who used the punched card for the census and gave us the mainframe.

We humans invented government in the first place, as a way to organize ourselves. We drafted a constitution, and swore allegiance, and paid taxes, and held elections. We have government; government does not have us. And with the advent of widespread digital platforms, we have a chance to rethink it entirely so it can take advantage of all the incredible things it might now do for us.

Digital government is way better on so many fronts

Digital is cheaper, faster, better, and more popular in almost every way. Be honest: Is there any part of your relationship to your government you’d prefer to do in person, over a fax, or on the phone instead of on an app?

We should be demanding a digital-first government.

• The government has more information on us than anyone, from our taxes to our medical history to our travel records. Why can’t we have single-sign-on to a dashboard—not multiple dashboards organized by department, but one, simple dashboard—that shows us everything government collects and measures and knows about us, in context? Nobody should know more about you than you do, including your government.
• Every interaction with government should have a support ticket, so we can easily see dates, times, and status.
• We should be able to use a trusted single-sign-on account to log into other sites. We’re fine using Google, Facebook, Twitter, and others to sign into things, and none of those organizations are legally accountable to us.
• We elect representatives based on little more than a face on a poster painted with our chosen party’s background colour. But we can do so much better. We should be demanding easy, intuitive records of how our representatives voted. Heck, our system of government was invented around the time of the printing press. Maybe we should update that now that everyone has a screen in their pocket.
• Tech has made so many new things possible, but we aren’t thinking radical enough because government is terrified of experimenting. For example, we should be clamouring for quarterly live-streamed debates among party leaders, rather than waiting for an election cycle to hear them challenge one another. If Twitch and YouTube work, why aren’t we making them work for us?

We simply aren’t thinking big enough.

Why doesn't the government have an app dev shop?

Some countries’ citizens can do their taxes in ten minutes. Why not us? Why isn’t there a team of software developers, inside government, able to release an app just as quickly as an app development shop? Why doesn’t the government have an app development shop?

WTF, Canada?

Why isn’t the government full of the best damned technologists in the world, using technology to make society better? It’s more than just the incredible imbalance in compensation versus private sector jobs. It’s cultural.

Many of the newspapers covering the story treat it like a political hot potato, another sign of government excess, or the inherent superiority of the private sector. But this story is not about app development shops being better. It is about government being better.

I believe that’s something we should invest in. I don’t want my government to be a hollowed-out shell of policymakers and bureaucrats, completely dependent on the public sector for operations. There’s plenty of room for public/private collaboration around the utility parts of technology, such as cloud computing. But government needs to have root. Someone needs to know how the system works.

Government should code.

Some sidenotes

Some other stuff you may find useful if you made it this far:

• Have you read Canada's Digital Ambition? You should. There's a plan to fix this, and smart people are working on it, but they need the public to care, and get involved, rather than just throwing up our collective hands and saying “just outsource it to the private sector forever.”
• If you want to come and you’re in regional government, you can claim a pass. And if you work for SSC, EDSC, HC/PHC, the CRA, and some other departments, you have a departmental pass waiting for you to attend the whole event online (details on this page.)
• This thread from Dan Hon dives into salary expectations and other factors, in a previous tempest about the cost of building (well, cloning) vaccine finder websites.
• I know many really, really smart developers who, despite being underpaid and constantly attacked by armchair pundits, have devoted their lives to government and civic tech causes. None of this is meant to say that the public or the private sector is superior. What I am saying, in case you missed it, is that the changes we’re aspiring to in the public sector need to be more ambitious.