Five principles to guide any government IT effort

Posted August 23, 2015 | View revision history

Three years ago to the day, the first class of Presidential Innovation Fellows were sworn in. Being part of the inaugural class was a lot like founding a startup: we had to define our problem space, figure out basic mechanics like where to work or how to get laptops and servers, and most importantly, had the opportunity to create our own culture around how we approached problems.

Well before we wrote our first line of code, we defined a set of guiding principles. The principles helped us lay the blueprint for how to deliver on the Project MyGov charter:

To reimagine how citizens interact with government through an experience designed around their needs rather than a confusing and fragmented bureaucracy.

Looking back three years later, many of those principles still remain valid guideposts for any reform effort in government:

  1. Change the culture. The PIF program, 18F, USDS… half of what they deliver is code. Only half. The other half of their product, the value they provide to government, is leaving their agency partners with a better understanding of and appreciation for what modern software development looks like and to disrupt bureaucracies and challenge assumptions in the process.

  2. Serve the people first, and the government second. Reform efforts in government often confuse stakeholders with customers. The customer, the end user of the product or service being reengineered, is always going to be the American people. It may be developers, it may be veterans, it may be retirees. Despite the intermediary steps, the customer is rarely, if ever, a government bureaucrat. Optimize accordingly.

  3. Engineer simplicity. Citizen-centric design means reducing the cognitive burden required to utilize government services. It means leveraging technology to absorb the complexity of bureaucracy on behalf of the public. A user shouldn’t need to study the org chart to know where to submit a form and they definitely shouldn’t have to tell the government the same information twice, especially information it should already know.

  4. Solve for actual needs, not perceived ones. Between interest groups, the media, and Congress it’s all too easy to be a perpetual slave to the squeakiest wheel. Data-driven decision-making can often solve for this request whack-a-mole. At the FCC, for example, we used analytics to show that lawyers and lobbyists were actually a tiny subset of visitors to, and the use cases the site was optimized for shifted accordingly.

  5. Build a platform. We’re well past government as a platform. Government is a platform. Heck, most tech. companies today are platforms (see, e.g. Google, Apple, and Microsoft). If you’re not building a platform first and a website, app, or workflow second, you’re engineering from the outside in. Be lean, be modular, be decentralized. Expose APIs. Encourage reuse and collaboration through interoperability. Grow an ecosystem. Eat our own dog food.

An organization’s principles are more than mere words. They’re the underlying assumptions that constrain all decisions. In government, it’s all too easy to focus on the code, on the concrete deliverable. After all, RFPs, rarely, if ever, mention culture or customers or citizen-centricity. As the MyGov principles ask, “are [you] solving for the right thing”?

H/T @californiakara, the author of the original post, and whose Tweet let me to rediscover it.


Ben Balter is a Senior Manager of Product Management at GitHub, the world’s largest software development network, where he oversees the platform’s Community and Safety efforts. Named one of the top 25 most influential people in government and technology, Fed50’s Disruptor of the Year, and winner of the Open Source People’s Choice Award, Ben previously served as GitHub’s Government Evangelist, leading the efforts to encourage government at all levels to adopt open source philosophies for code, data, and policy development. More about the author →

This page is open source. Please help improve it.