If I started a $1 million factory building project, and I had only $1,000 of work left to do on it when I found out the factory was sure to lose money, should I spend the money to finish it?

Instinct says that $1,000 is a lot less than $1 million and you hate to see that $999,000 go to waste. But the truth is, if I know the factory will lose money and I spend the $1,000 then that’s $1,000 more gone to waste. In this scenario, the $999,000 is called a sunk cost.

I did one of these irrational sunk cost deals last time I was in Dallas Fort-Worth airport. It’s a big place and you are supposed to take a little train to get around quickly. I rushed over and took the long, high escalator to what I thought was the train depot level, and realized I had accidentally taken the escalator to the pedestrian walking bridge instead. It was a long way to walk and I was short on time, but in my hurry and stubborn style I walked all the way across to the other terminal. It would have taken 30 seconds to simply ride an escalator back down and get back on the right track. But I didn’t want the reverse and the restart. Going forward felt better.

The purely rational way to evaluate these situations is to compare the future outcome of option A with the future outcome of option B. Whatever will produce the best future outcome, go with it. The past can’t be changed by the current decision, and must be accepted for the now-unavoidable loss that it is. Decisions should be based on how they will affect the future. Present decisions don’t change the past.

Sometimes the cost of indulging the straightforward, reversal-free choice is just a few minutes of time or some other price tag small enough to ignore. But when a lot is at stake, don’t let sunk costs cloud your judgment. It takes sober awareness of the numbers, courage, and humility. Sometimes you just have to admit a big mistake and turn the ship around.