Don’t Prematurely Optimize Mobile Health

This piece grew out of an mHealth working group discussion in March 2013. Its argument is simple: in mobile health, as in software engineering, optimizing the wrong thing too early is a costly mistake.

Professor Donald Knuth, a legend in the field of computer science, wrote a famous passage about premature optimization that will resonate with any technologist:

“There is no doubt that the grail of efficiency leads to abuse. Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%.”

Optimization is not just about speed

This passage is usually cited in reference to performance and speed, but the same logic applies to scale and sustainability. The number of mHealth technologies is increasing, as is the maturity of certain platforms. Yet mHealth technology is mostly just an amplifier of human intentions. Deciding which “parts of a program are really critical” to scale and sustainability must be well understood before we can identify where to invest our time and energy.

Technology amplifies human intent

In our work with community health workers (CHWs), we often struggle to determine how the performance improvement we are trying to achieve with systems like CommCare will actually be realized. This goes well beyond whether the technology is working, extending to human resources, ownership, and accountability. The motivation or management capacity may be lacking, and this is likely to be exacerbated at scale.

Mobile technology cannot solve that on its own. If we are trying to better enable supervision and management, the cost of fixing the underlying HR or training issue may dwarf the entire cost of the mHealth technology. So if the return on investment of deploying the technology hinges on functioning supervision and management for the CHW system, then any optimization within the economics of the technology should not be addressed before the supervision and management issues are.

But design for scale early

That said, designing for scale and sustainability early on is absolutely critical. If you do not proactively design pathways for optimization in certain areas, you may find them impossible or prohibitively expensive to improve later. Two design pathways worth leaving open:

1. Assume technology will change, and evolve with it. The pace of innovation is faster than we can imagine. Do not build only around today’s technology, and do not be afraid to jump to new technologies when needed.
2. Build a positive feedback loop between the technology and the system. There is an old management saying that “culture beats strategy,” and culture can absolutely destroy technology. Even scalable technology will be a poor use of resources if it is introduced into a system that is not designed to use it. When good technology meets the right system dynamics, it improves the system, which generates more demand on the technology, and both get better together.