💸 Sunk Cost Fallacy


The Sunk Cost Fallacy is the tendency to continue a project in which we have already invested time, effort or money, minimising downsides and new information in the process.

Let’s talk about the Sunk Cost Fallacy.

The Sunk Cost Fallacy is the tendency to continue a project in which we have already invested time, effort or money, minimising downsides and new information in the process.

🤔 Why you should care about it

The sunk cost fallacy is most dangerous when we have invested a lot of time, money, energy, or love in something. This investment becomes a reason to carry on, even if we are dealing with a lost cause.” - Rolf Dobelli, author of The Art of Thinking Clearly.

Engineering teams can encounter the Sunk Cost Fallacy at:

  • The feature level (carrying on coding a feature, even if there is no proof it will eventually work);
  • The code level (patching and maintaining legacy code, just because it took so long to write).

😫 Problem(s)

We fall for the Sunk Cost Fallacy due to the following cognitive biases:

Commitment bias —> Following a previous decision, disregarding new evidence contradicting the decision, just because we made it

Loss aversion —> Avoiding losing situations since losses feel much worse than gains.

Endowment effect —> Attributing more value to something we own (or make) than something we buy.

😃 Solution

To avoid falling for the Sunk Cost Fallacy:

1/ Always disregard previous costs —> Only make decisions based on current information, independently of past investments.

2/ Make bets instead of estimates —> Identify the maximum amount you’re willing to spend on a feature (based, for example, on projected benefits), and stop when you reach the threshold.

3/ Use committees—> People that are not emotionally attached to a project are less likely to be biased towards it.

4/ Organise regular “cleaning” sessions —> Ask developers to regularly inspect old code to improve, replace or delete it. The more we delete code, the less attached we are to it.

💡 Key Concepts

Cognitive Bias —> A systematic pattern of deviation from norm or rationality in judgment, sometimes leading to perceptual distortion, inaccurate judgment or illogical interpretation.

Bygones principle —> The action of ignoring past costs and only considering future costs and benefits when making decisions.

Iterative Decision Making —> Regularly revisit past decisions to update them with newly available data.

🤨 Typical Behaviours

“We’ve spent so much time trying to solve this problem, we can’t give up now!” —> If projected results are not there, it doesn’t matter how long you work on a project.

We just learned new information that makes us confident that, now, we’re gonna ship the project.” —> The only relevant information to consider is the one impacting future results, not past decisions.

If we had more resources, we could finally finish this project.” —> Adding more resources to a late software project makes it later (Brooks’s law)

📚 Top books

Thinking, Fast and Slow (English Edition) eBook : Kahneman, Daniel: Amazon.fr: Boutique Kindle
Achetez et téléchargez ebook Thinking, Fast and Slow (English Edition): Boutique Kindle - Personal Transformation : Amazon.fr
The Art of Thinking Clearly (English Edition) eBook : Dobelli, Rolf: Amazon.fr: Boutique Kindle
Achetez et téléchargez ebook The Art of Thinking Clearly (English Edition): Boutique Kindle - Neuropsychology : Amazon.fr

⚙️ Tools

How Susceptible Are You to the Sunk Cost Fallacy? - David Ronayne, Daniel Sgroi, and Anthony Tuckwell —> The research-based tool presented in this HBR article enables managers to measure their susceptibility to the Sunk Cost Fallacy.

📝 Content

📼 Why should I ignore sunk costs in a decision? (00:04:55) - Dan Ariely —> quick introduction video from Dan Ariely’s course on decision making.

📝 The Sunk Cost Fallacy, explained - The Decision Lab —> A comprehensive definition of the Sunk Costs Fallacy and how to avoid it, with multiple references to research work.

📝 The Sunk “Code” Fallacy - The Unicorn CTO —> My take on the Sunk Cost Fallacy for engineering teams.