The Illusion of Speed: A Critique of My Own Heuristics

I have made a lot of gut-based decisions when I should not have. Even knowing this, part of my mind still believes that my instincts are unusually good, that my first impression is fast and correct.

Recently, I decided to challenge that belief. I wanted to put serious effort into becoming more rational and to develop real critical-thinking skill rather than relying on intuition alone. In that process, I discovered the LessWrong community and Eliezer Yudkowsky’s Rationality: From AI to Zombies.

I had encountered terms like “sunk cost fallacy” before, but reading about cognitive biases at a mechanical level was different. Instead of treating bad decisions as character flaws, the book frames them as predictable failure modes of the human mind.

The idea that stood out most was cognitive heuristics. These are mental shortcuts, compressed rules of thumb, that evolved to help us make fast decisions in environments where speed often mattered more than precision. The problem is that the modern world, especially online markets, is optimized to exploit those shortcuts rather than reward them.

I can see this clearly in one of my own decisions. I once purchased an expensive online course created by popular YouTubers who had the exact career path I admired. Their branding was polished, the marketing was persuasive, and the course had thousands of five-star reviews. They looked like successful people who knew what they were doing, so I assumed their product would deliver similar results. I bought the course almost immediately.

It turned out to be the most money I had ever spent on an online course, and the least value I received. The material was shallow, disorganized, and largely recycled from free content. When I reached out for a refund, the same creators who appeared helpful and approachable on camera never responded.

Looking back, the error is obvious. I relied on a cluster of heuristics rather than evidence. I fell for social proof, assuming that many positive reviews implied high quality. I was influenced by the halo effect, letting the creators’ perceived success spill over into my judgment of their product. Most importantly, I neglected base rates. I never asked the statistical question: How often do highly rated online courses actually deliver meaningful value, especially in markets driven by hype, early reviews, and influencer authority?

The ratings told a compelling story, and I accepted that story without interrogating how those ratings were produced or what incentives shaped them. My brain chose the fast explanation over the accurate one.

What unsettles me most is that living through this mistake did not automatically make me better. I now have “experience” buying a bad course, but experience alone did not reprogram the heuristic that caused the mistake in the first place. I have repeated similar errors in other domains.

This is where Yudkowsky’s distinction between experience and expertise matters. Experience is passive; it is simply what happens to you. Expertise is active. It involves understanding why your reasoning fails and deliberately installing safeguards against those failures.

I no longer want to rely on the passive hope that I will become wiser with age. I want to build rationality expertise. That does not mean analyzing every trivial choice, that would be paralyzing. But for decisions that are expensive, irreversible, or reputation-defining, I can no longer rely on gut instinct alone.

My goal is to treat my mind like a codebase that needs debugging. When the stakes are high, I now try to slow down and force a few explicit checks: What is the base rate for success in situations like this? What evidence would disconfirm my current belief? Am I being persuaded by appearance, authority, or popularity rather than substance?

My intuition is fast, and it is sometimes useful, but speed is not the same as accuracy. If I want better outcomes, I need to stop mistaking confidence for correctness and start intervening where my ancient mental shortcuts predictably fail.