Sunk Cost Is the Reason You Finish the Book You Hate

The most expensive book I own is one I didn’t finish, and that is precisely why I kept reading it for another month. It sat on the shelf, and every time I looked at it I thought about the money already gone, so I forced myself through chapters I wasn’t enjoying to make the purchase feel justified. The logic was backward in a way I couldn’t see while I was inside it: the money was spent whether I read the book or not, so the only thing reading it could change was how much of my remaining time I wasted. That backward logic has a name, and knowing the name is the start of noticing it before it costs me an afternoon — or a year.

The sunk-cost effect is the tendency to let what you’ve already invested — money, time, effort, pride — pull you further into something, even when the future costs exceed the future benefits. Behavioural researchers have been documenting it for decades; the classic demonstration by Arkes and Blumer back in 1985 showed people paying more and persisting longer once they’d sunk money into a course of action, even when the sunk amount logically shouldn’t change the decision. The pattern shows up far beyond economics: the gym membership you don’t use, the project you won’t kill, the relationship you stay in past its natural end, the queue you won’t leave after twenty minutes.

The reason it’s so hard to shake is that it borrows the force of a deeper, well-established bias: loss aversion. We feel losses roughly twice as strongly as equivalent gains, so walking away from something we’ve paid into feels like losing the money a second time, even though the money is already gone either way. The sunk-cost effect is loss aversion pointed at the past instead of the future — we keep paying attention to a cost that a rational accountant would have written off the ledger the moment it left our hands. This is the same engine behind why we defend purchases we already regret, and it runs below the level of conscious reasoning.

Where it gets its sharpest teeth is when the thing we’d be abandoning is tied to our identity rather than just our wallet. A person who’s told everyone they’re “writing a book” will keep writing it long past the point of diminishing returns, because quitting now means admitting the sunk investment wasn’t just money but a version of themselves. A person who paid for a year of a course will attend it resentfully, because stopping would mean the past decision was a mistake. The bigger the sunk cost is measured in self-image rather than dollars, the harder it is to let go, and no spreadsheet fixes that.

The interesting part is that we apply this logic to others’ decisions easily and our own almost never. Watching a friend refuse to leave a movie they hate because they “paid for the ticket” is transparently silly to us; we can see the ticket is gone. But the identical structure in our own lives — the subscription, the degree, the half-renovated room — feels entirely different, because in our own case the sunk cost is entangled with the story we tell about who we are. That asymmetry is the giveaway that what’s operating isn’t a calculation; it’s a narrative we’ve built to defend a conclusion we already reached.

There’s a reliable countermeasure, and it’s a single question asked at the right moment: “If I were deciding today, with nothing already invested, would I start this?” The question works because it strips out the history — it forces you to evaluate only the future costs and benefits, which are the only ones your choices can still influence. If the honest answer is “no, I wouldn’t start this today,” then the only reason you’re still in it is the sunk cost, and that reason was never a real reason. The difficulty is that the question feels disloyal, and that feeling is the bias talking.

A second tool is to make the decision before the sunk cost exists. This is the pre-commitment move: decide in advance what evidence would make you quit — “if I’m not using it twice a week by March, I cancel” — and write it down while you’re still objective. Then when the sunk-cost pull arrives, you’re not making a fresh emotional decision; you’re following an instruction your earlier, clear-headed self already left you. It’s the same principle as the checks I described for estimating before the planning fallacy sets in: decide the rule before the feeling shows up.

The subtler form of the countermeasure is to reframe the sunk cost as tuition rather than loss. Money spent on a book you didn’t finish, a course you dropped, a tool you stopped using — reframed as the price of learning what you don’t actually want, it stops being a reason to keep paying and becomes a reason the purchase was, in its way, still informative. This is not self-deception; it’s a deliberate re-labelling that neutralises loss aversion by converting a “loss” into a completed transaction with a lesson attached. You can only get that lesson, though, if you actually stop, which is the thing the bias is trying to prevent.

There’s a limit to this reframing, and it’s worth stating, because the bias can also be recruited as an excuse in the other direction. “Sunk cost” is sometimes what people say to justify quitting the moment something gets hard, when the real problem is ordinary discomfort rather than a genuinely closed decision. The test that separates the two is the same question, run honestly: “Would I start this today?” If the answer is yes, the thing is hard and you should push through; if it’s no, you’re staying for the wrong reason. The bias isn’t a licence to abandon things — it’s a way to tell the difference between persistence and self-punishment.

The reason this one bias earns a whole essay is that it’s a decision most of us make this week, more than once, without noticing. Whether to cancel a subscription you’ve “already paid for” this year, whether to finish a book or show you’re not enjoying, whether to stay in a class or project that’s gone cold — all of them are the same shape, and all of them are resolved by the same move. Knowing the name doesn’t protect you from it, as I’ve argued before about naming biases without changing behaviour — what protects you is running the question at the moment the pull arrives.

So the next time you catch yourself saying “I’ve already come this far” or “I’ve already spent so much on it,” treat the sentence as an alarm rather than an argument. The past cost is the one input your next decision should ignore, and the very fact that it’s the one you’re reaching for is the sign it’s time to ask the question. You already paid. That’s exactly why you shouldn’t keep going.