Usefully Wrong Models
Thinking Inside and Outside the Box
Early in my life, I rebelled against putting most anything into boxes, especially people. Categories and models felt like they predicted not just possibilities but limits. Worse, they risked enforcing those limits.
My first career in music seemed to embrace that freedom: creativity! No boundaries! But it turned out music is filled with systems and models, both intuitive and explicit, that help us create and interpret. My second career, first in programming and then in leadership, showed me that models can also support efficiency and fairness.
I learned to appreciate models not as prisons but as tools: simplifying complexity to help us act faster and better. George Box said it best: “All models are wrong, but some are useful.”
But I’ve learned—hat tip to
—that some are also attractively wrong: misleading in ways you might not notice at first. Here’s how I categorize models today.Entirely wrong. Just bad. Avoid these. My examples include fixed mindsets and managing innovation from fear.
Attractively wrong. Maybe fun at first, but dangerous over time, like a playground in a construction zone. These models demand vigilance. My examples include “10x engineers,” and ORMs.
Usefully wrong. Models that simplify reality in ways that let you work faster, think smarter, or deliver better outcomes. These are the keepers. My examples include Wardley maps and Lean.
I hope this model is usefully wrong for you. Let’s explore how to make it work.
Three Lessons for Evaluating Models
Here are three principles I use to navigate the world of wrong models.
Understand the Trade-offs. No model is perfect. Ask: “What risks does this introduce? What doesn’t this account for?” For example, Wardley maps excel at strategic context, but they don’t tell you when the next change will happen, or what you should do next. If someone’s selling you a model without caveats, they’re probably selling something else, too.
Focus on Outcomes. Models are tools. Their job is to help you achieve better outcomes. If a model isn’t working—like when an ORM slows down your system as complexity grows—revise it or let it go.
Evolve and Adapt. No model lasts forever. Stay curious, keep learning, and be ready to tweak or toss even your favorite frameworks as the world changes.
Closing Thoughts
The more I’ve worked with models, the more I’ve learned to appreciate their imperfections. Models simplify complexity, but the world is always bigger than the boxes we draw. The trick is to use them thoughtfully, stay aware of their limits, and know when to step outside the lines.
What’s the most usefully wrong model you’ve encountered? How did it help you succeed despite its flaws? And have you ever been burned by an attractively wrong model that seemed perfect at first?
