Agree on how to decide before arguing about what to build. In automotive feature reviews, familiar arguments come up:
“Competitors already have it.”
“Customers will love it.”
“It’s a safety feature.”
“The driver is still responsible.”
Each may deserve consideration. None should replace an agreed decision process.
When we choose the criteria during the debate, we can keep changing them until our preferred feature wins.
My approach: agree on a decision framework before discussing individual features.
Let’s walk through the attached framework, question by question.
1. Automatic gear shifting
Does the function appear when the driver wants it to?
Yes. The driver selects the transmission’s operating mode through PRND.
Will the function automate some routine tasks?
Yes. The driver no longer needs to change gears manually.
Does the function allow the driver to usefully focus on something else?
Yes. Their right hand is available for other actions, such as changing the wiper setting or taking a sip of water where lawful and safe.
Does a malfunction in the function’s indication affect decision-making?
For this example, assume no: the transmission fault also becomes apparent through the vehicle’s behaviour. The car may fail to move or behave abnormally, giving the driver another reason to recognize a problem.
This assumption needs checking against specific failure modes; some faults may not be obvious.
Decision: provide an indication of the anomaly. The driver noticing abnormal behaviour does not explain what has gone wrong.
2. Highlighting detected pedestrians on the HUD
Does the function appear when the driver wants it to?
No. The indication appears when the system detects a pedestrian. The system decides whether to show it.
Will the function increase risks for road users?
Let’s initially assume no: the intended benefit is to reduce collision risk.
Does a malfunction in the function’s indication affect decision-making?
Yes. A driver may develop the habit: “I see a highlight, so there is a pedestrian, so I brake.”
If the highlight fails to appear, they may miss the person — or even stop actively looking for pedestrians.
That challenges our initial assumption about risk.
What will the driver do if they don’t see any indication?
A driver who has become accustomed to reliable highlighting may fail to notice the malfunction.
Imagine a person in dark clothing at night. Without a highlight, the driver might never realize someone was there. They may recognize neither the pedestrian nor the missed detection.
Framework branch: Fail to notice the malfunction.
Will moral and/or legal responsibility be transferred to the car’s systems in the event of an error?
Yes in the sense of practical reliance: the driver may have delegated part of pedestrian detection to the system. Legal responsibility requires a separate assessment.
The possibility of striking a pedestrian is a serious reason to avoid creating that dependency. Drivers must retain the habit of scanning for people themselves.
That is why I oppose pedestrian highlighting on the HUD: I do not want the absence of a highlight to become reassurance that nobody is there.
The value of agreeing on this framework beforehand is that neither “it improves safety” nor “the driver remains responsible” can end the discussion.
We still have to answer every question, expose our assumptions, and identify the evidence needed.
Agree on the decision rules before anyone starts defending a feature.
Author
Alexey Dmitriev