Published in the San Diego Union-Tribune, June 16, 2025
by Neil Senturia
Decision-making.
“Two roads diverged in a yellow wood / And sorry I could not travel both,” poet Robert Frost wrote.
There you have it.
There are some hidden traps in decision-making. The guru for this kind of understanding was the late Dan Kahneman, 2002 Nobel Prize winner for his work in behavioral economics. I am going to revisit one of the more famous ones.
Sunk Costs. This is the mother of all stupid biases. This is the rabbit hole. One of the primary reasons that people can’t free themselves from previous decisions is that they do not want to admit to a mistake. Ego.
When you have to admit to an initial poor decision, it causes you to feel a loss of self-worth, particularly in a public forum. In private, your mea culpas are easily given and with a good partner, they are quickly forgiven and forgotten. But in front of the employees or the investors, ah therein lies the rub.
And in the court of public opinion, it can have a much greater sting. That is exactly why CEOs don’t do it, unless cornered. The list of projects that fall prey to the sunk cost fallacy are legion. Dare I say bullet train. Or perhaps Ash Street.
The fascinating question is when you see it happening, like a train wreck in slow motion, why is it so hard to call it out?
I think it requires leadership, but not from the top. The top is dug in. It is the second and third level of executives that need to provide the intervention. Or a very loud and persuasive public. Far better to shine the light than to die in the darkness.
“When you find yourself in a hole, the best thing you can do is stop digging,” says Warren Buffett.
Seems obvious. But you have already bought the shovel, and the dump truck is in the parking lot.
It strikes me that in the startup world, this issue is less virulent. Startups pivot all the time. If it is not working, change something. Not only the product, but perhaps also a member of the team that in retrospect you should never have hired and now how long will it take to dismiss him/her.
One of the most difficult things to do is to change a sacred cow business model.
Consider the current venture financing model which has been in place for 50 years. Raise a fund, take your 2% management fee, then 20% of the profits above a threshold, then rinse and repeat and raise another bigger fund. Bigger is better if you multiply by 2%.
The current model for picking an investment hinges on the the idea that each bet needs to have the possibility of returning the entire investment of that fund. Thus, the hunt for unicorns.
What if that model is past its sell-by-date? What if the unicorn is a dying breed? They have been over-hunted, and maybe some of those unicorns are actually merely rabbits with a fake horn on their head.
I have spoken to some VCs. They still believe in that model. It is why one of the key questions asked of a founder is what is your TAM, your total addressable market, i.e. how big can your company become — in that closely structured vertical. Can you be the elephant in the jungle?
I would like to revisit the business model of a “conglomerate.”
Think Berkshire Hathaway, they own stakes in multiple companies. Consider Alphabet, the parent company of Google. They own pieces of multiple companies, many of them using their core technology. Think Procter & Gamble, they own pieces of multiples brands. They strategically “acquire and manage stakes in multiple unrelated businesses, often across different industries.”
Startup seed investing is treacherous. It is the “box of chocolates” problem. You don’t know for sure what you are going to get. But on the bright side, at least you know you bought chocolates, not bananas. And there is optionality, several different kinds of chocolate in that one box.
I agree that a company like that in its infancy might not be investable by a classic VC. They would be considered “all over the board.” But this lack of certainty gives optionality to both you and the founder. Don’t like caramels — try the fruit center.
Rule No. 794. Sunk costs. Note the word “sunk.”