📖 Business
Experimenting in Public
Mitchell Weiss tackles the central paradox of public-sector innovation: government must serve all citizens reliably, yet it must also experiment to improve — and experiments, by definition, can fail. His solution is not to avoid experimentation but to do it smartly: running low-cost, time-bounded tests in public settings that generate real evidence before committing large budgets. Weiss shows through cases like Singapore's early COVID-19 contact tracing experiments and various city pilot programs that disciplined public experimentation is not only possible but produces dramatically better outcomes than the alternative of rolling out untested programs at full scale and hoping they work.
2
Minutes
2
Concepts
+45
XP
1
How It Works
- The Experimentation Paradox — Government has a democratic obligation to serve everyone, which creates pressure to never fail. But refusing to experiment means committing to programs that may be ineffective, wasting far more public money than any failed pilot. Weiss resolves the paradox by reframing experiments as responsible stewardship rather than reckless gambling.
- Low-Cost Probes Before Large Commitments — The methodology centers on testing the riskiest assumption first with the smallest possible investment. Rather than building a $50M program and discovering it does not work, run a $50K pilot that tests whether the core mechanism actually changes behavior. This is the public-sector equivalent of the lean startup's minimum viable product.
- Designing Experiments for Learning, Not Just Success — Good public experiments are structured to generate useful information regardless of whether they "succeed." This means defining hypotheses, comparison groups, and success metrics in advance. A failed experiment that teaches you something valuable is better than a vague pilot that produces ambiguous results.
- Managing the Politics of Failure — Weiss acknowledges that failed experiments create political vulnerability. His prescription: frame experiments as learning before you start, set expectations that some will fail, and build bipartisan coalitions around the methodology rather than specific outcomes. Transparency about what you are testing and why creates political cover.
- From Pilot to Scale — The final piece is having a clear pathway from successful experiment to full implementation. Many government pilots succeed in isolation but never scale because no one planned for the transition. Weiss advocates building scale criteria into the experiment design from the start.