18F Blog

If you’re building a rules-based system, don’t assume that you need a separate business rules engine product. Rules can be implemented more easily and with less overhead by cross-functional teams working to describe the rules and policy directly in code using a general purpose programming language like Python, Ruby, etc.

As agencies try to adopt agile development practices and modular contracting methods, there are several anti-patterns that we have noticed through the course of our work. We address how these can hinder success and alternatives to consider.

At many government agencies, a central IT team manages DNS directly. Other teams must request changes using help desk tickets, which can have inconsistent turnaround times, and are susceptible to human error. Having DNS records as code and doing changes through pull requests brought turnaround time down from multiple days to under ten minutes.

Providing government data to the public almost always requires building a data processing pipeline between its place of origin and the systems that will serve it. Data must be copied, transferred between digital storage formats, reshaped to meet the needs of reporting systems, groomed for readability, and cleansed for accuracy.

As part of our work building login.gov, a single sign on service for government, we’ve been looking at ways to effectively verify people’s identity online. Not only did we need to find a technology solution to meet this need, we need to find a solution in a stack of brand new possibilities.