Friday morning. Someone has been working on a project for six months. It's hard, progress has stalled, and last week someone forwarded them an article about the sunk cost fallacy. They're thinking about quitting.

The sunk cost fallacy is real. The lesson most people have drawn from it — "don't let past investment stop you from cutting your losses" — is causing more harm than the original fallacy ever did.

Here's the problem: the sunk cost fallacy describes the error of continuing something only because you've already invested in it. This error exists. It's documented in the literature. But the research on when and how often this actually drives individual behavior is thinner than the concept's popularity suggests.

What's not thin: the research on quitting too early.

The Less-Discussed Direction

Angela Duckworth's research on grit — the tendency to persist through difficulty toward long-term goals — shows that grit predicts meaningful outcomes better than intelligence, talent, or prior achievement across a wide range of domains: graduation rates, performance in competitive settings, creative achievement. The mechanism isn't mysterious. Almost all meaningful skills and projects involve extended periods of nonlinear progress — plateaus, apparent regressions, stretches of invisible compounding that only look inevitable in hindsight.

The J-curve is real. Writers who become good spend years writing badly. Mathematicians who achieve breakthroughs work for months producing nothing publishable. Businesses that eventually succeed often go through extended periods of flat or declining metrics. The person who quits during the flat part — because "the data shows it isn't working" or "I shouldn't let sunk costs influence me" — never sees the inflection point.

This pattern is consistent across domains because the learning curve for most complex skills isn't linear. Early gains come fast because the low-hanging fruit is plentiful. Then you hit the part where progress requires changing fundamental patterns, and nothing seems to work for a while. This is not evidence that you should quit. It's structurally predictable, and the people who understand it persist through it.

What the Sunk Cost Research Actually Shows

The classic sunk cost fallacy research involves lab scenarios with clear numerical values: you've already spent $100 on a nonrefundable ticket, should you attend the concert even though you'd rather do something else? In these contrived scenarios with explicit numbers, yes, people over-weight past investment.

Real-world examples of the sunk cost fallacy in action are harder to find, and the clearest cases are usually organizational rather than individual. The Concorde fallacy — Britain and France continuing to fund the supersonic jet after it became economically inviable — is the canonical case. But it was driven largely by political accountability: politicians couldn't admit the waste without also admitting the original decision was wrong. The famous cases of throwing good money after bad tend to involve institutions with diffuse accountability, not individual decision-making.

For individuals making private choices about projects and commitments, the research on grit, deliberate practice, and expertise development paints a different picture: most meaningful failures involve stopping too soon, not too late.

Why "Sunk Costs Don't Matter" Is Dangerous Advice

There are three ways the overcorrection damages you:

It misidentifies the problem. When your project isn't working, the most common cause is not that you're persisting past a rational exit point — it's that something about your approach is wrong. The right response is to change your approach, not abandon the project. "Cut your losses" as a default response to difficulty is a thought-terminating move that skips the actual diagnostic work.

It becomes indistinguishable from avoiding hard work. Genuine sunk cost reasoning says: "I should keep doing this despite mounting evidence it won't work because I've already spent so much on it." That's a specific, identifiable pattern. What more commonly happens is: "This is getting harder and I'm losing confidence and there's a concept I learned that seems to apply here." The concept provides intellectual cover for a much simpler impulse: stopping because it's difficult.

It disables one of your few reliable sources of feedback. Being genuinely invested in something — emotionally, temporally — creates pressure that generates information. When you're truly committed, you notice what's not working more clearly than you would as a detached observer. Pre-emptively depressurizing that commitment (by reminding yourself that sunk costs don't matter) removes a feedback mechanism that works.

How to Know When to Actually Quit

The genuine sunk cost fallacy requires a specific condition: you must be continuing primarily because of past investment, not because of future prospects. This is worth testing directly.

Ask: If I were starting fresh today, with no prior investment, would I start this project? If yes, keep going — adjust the approach, but keep going. If no, ask yourself why. Is the underlying logic of the project broken? Has the context changed fundamentally? Or are you just tired and it's hard?

Being tired and finding it hard is not a reason to quit. It's the price of almost everything worth doing.

There's also a version of quitting that deserves more credit: quitting this approach, not this project. Most projects that succeed do so after significant pivots. The question isn't always "continue or stop" but "continue doing what?" If you've genuinely tested your current approach and it's producing no signal, change the approach. Persistence doesn't mean doing the same thing repeatedly. It means staying committed to the outcome while remaining flexible about method.

Takeaway

The sunk cost fallacy is a tool for a specific situation: you have new evidence a path won't work, and past investment is the only reason you're staying on it. It applies to much less of your life than you think.

For everything else — the project in a plateau, the skill that isn't clicking yet, the work getting harder because you're in deeper — the default assumption should be persistence with adaptation, not exit. The moment something gets hard is usually the moment you're at the threshold of the learning that would make it worth doing.

Sunk costs don't matter. That doesn't mean stopping when things are hard is free.