Series: Cognitive Biases

1. The Danger of Cognitive Bias (this post)
2. Action Bias
3. Sunk Cost Bias

I used to think the hardest part of building games was technical. The engine, the platform, the tooling, the bugs that only appear on one device in one country at 2am. Those problems are real, but they’re rarely the ones that threaten the whole business.

The moments that nearly took Kempt down were not technical at all. They were decision moments. The kind where everyone in the room thinks they’re being rational, and later you look back and realise you were running on instinct, ego, and momentum.

Two patterns did the most damage: sunk cost bias and action bias. We kept pushing because we had already invested too much to stop. We kept moving because moving felt like leadership. Both felt responsible in the moment. Both were expensive.

There’s only one actual problem with building games and software: it’s done by humans. Humans get tired, stressed, defensive, territorial, ambitious, frightened. They want to feel competent. They want to feel consistent. They want the last six months to mean something.

Cognitive biases sit in the middle of all that. They are not a moral failing. They are a feature. They help us simplify messy realities. They also make the wrong decision feel sensible, especially when stakes are high and time is short.

The biases that matter most in teams

There are a lot of cognitive biases. You can lose an afternoon to the Wikipedia list. The trick is not memorising them. The trick is recognising the behaviours they produce when a team is under pressure.

In my experience, the most dangerous biases in games and software tend to show up in leadership and decision making.

Egocentric bias

The tendency to rely too heavily on your own perspective. It shows up as “I know what players want”, “I’ve seen this before”, or “trust me”. Sometimes that confidence is earned. Sometimes it is just comfortable.

Self-serving bias

Attributing success to your own ability and failure to external factors. It’s psychologically protective, which is precisely why it’s dangerous. It can block learning. It can prevent apology. It can turn a team into a place where nobody is ever wrong, they are merely unlucky.

Those two are often the gateway. They make a leader’s internal narrative resilient, which is great for confidence until it starts insulating decisions from evidence.

But if I had to pick the two biases that have caused the most direct, measurable harm in creative teams, it would be these.

Sunk cost bias

This is the tendency to value something relative to what has already been invested in it, rather than its objective value now.

It’s why teams keep polishing the wrong thing. It’s why a plan becomes identity. It’s why “just a little longer” becomes a habit.

It’s also one of the easiest to rationalise. You can always make a plausible argument for continuing because creative work rarely has a clean finish line. Sunk cost bias turns that uncertainty into a one-way ratchet.

If you want the deeper dive, it’s here: Sunk Cost Bias (post 3).

Action bias

Action bias is the tendency to favour action over inaction, even when action is not supported by evidence.

In deadline-driven creative environments, action is celebrated. Initiative is rewarded. People are praised for being decisive and fast. The trouble is that sometimes “do something” is not a strategy. It’s a stress response.

Action bias produces movement that looks like progress. It fills the day. It creates the sensation of control. It can also burn teams out and make a product worse, faster.

If you want the detailed version of how it shows up in briefs and leadership behaviour, it’s here: Action Bias (post 2).

Two quieter ones worth watching

Availability bias
We make decisions based on the most easily available information rather than the best information. In product terms, this is how teams steer off a small set of familiar KPIs because they are easy to pull and easy to debate, while richer feedback gets ignored because it takes effort.

Confirmation bias
We look for evidence that supports the position we already hold and downplay evidence that challenges it. Confirmation bias becomes particularly potent once a team has publicly committed to a direction.

Both get sharper as the team gets more stressed, more political, or more invested.

Reducing the effect of bias

If biases are hard-wired and sometimes psychologically useful, the goal is not becoming bias-free. The goal is designing decision making so the wrong story doesn’t win by default.

A few things that actually help:

1. Be aware of them, but focus on the ones that cost you money – Read the long list, then pick a short list. For most teams, sunk cost and action bias are a good place to start.
2. Write down the biases you personally fall into – Not as a self-flagellation exercise. As a checklist. If you know you have a tendency to push on, or to react quickly, name it.
3. Put a small pause into the system – Give yourself permission to think. Sleep on big decisions when you can. Re-read briefs the next morning. A lot of bad decisions are simply rushed decisions.
4. Make goals clear, measurable, and revisited – Bias thrives in ambiguity. If you can’t tell whether something is working, sunk cost bias becomes almost impossible to counter.
5. Remember the leader’s job is to pull thinking out of the team – Leaders don’t need all the answers. They need the best questions. Ask questions relentlessly and listen more than you speak.
6. When the room feels stuck, get an external viewpoint – A third-party opinion can help you pick through the narrative and see the situation with fewer emotional commitments attached.

And if you’d like an external perspective on a product, roadmap, or team decision, do get in touch.

Further reading

• https://en.wikipedia.org/wiki/List_of_cognitive_biases
• https://en.wikipedia.org/wiki/Egocentric_bias
• https://en.wikipedia.org/wiki/Self-serving_bias
• https://en.wikipedia.org/wiki/Depressive_realism
• https://en.wikipedia.org/wiki/Sunk_cost#Fallacy_effect
• https://en.wikipedia.org/wiki/Action_bias
• https://en.wikipedia.org/wiki/Availability_heuristic
• https://en.wikipedia.org/wiki/Confirmation_bias