Hacker News Reader

Comment by RobMurray | Hacker News Reader

Dunning-Kruger has been thoroughly debunked. https://www.psychologytoday.com/us/blog/how-do-you-know/2020... https://scitechdaily.com/math-professor-debunks-the-dunning-...

View on HN

> The majority of people can actually gauge their competence and knowledge accurately.

I know plenty of C-suite folks who must be in the minority then.

It is a skill to genuinely or convincingly look past your own lack of competence.

Have you considered that you might not be able to gauge peoples competence and knowledge accurately? Statistically, if you are observing so manly outliers that is likely to be the case. Hope this helps!

Ahhh this must be it.

The CEO babbling about how AI "works" must be knowledgeable and competent, and I, an engineer actually working on our AI product, am the idiot.

Thanks for the help!

I wonder if that's a case where folks in leadership feel like their job is to show confidence, provide direction, and they still maybe know they don't know enough (I mean who does...) and they still make the call (right or wrong).

Perhaps more of a poor leadership thing we observe.

I think that's more about leadership. For whatever reason "we" seem geared to follow people who pretend to know their way forward even when they're completely lost, perhaps even more lost than average.

Like Teddy Roosevelt said - 'Speak loudly, aggressively flail about, and you'll go far.' Or something like that.

Common usage doesn't typically meet that specific definition:

“If you are really, really stupid, then it’s impossible for you to know you are really, really stupid.”

Typically it is used to describe when people who don't have complete knowledge about the nuances of a problem or have opinions based on bad information or bad assumptions confidently or aggressively state positions and that confidence is confused with competency.

LLMs are a good example of the colloquial use. They are confident, superficially competent but almost always wrong in specific domains.

It is a problem when people underestimate their blind spots.

Building culture that allows all coworkers to safely raise consent and learning to depersonalize feedback helps with that.

It may not have as popular of a label, but it is well documented in fields like human factors.

I think a better model is unconscious incompetence > conscious incompetence > conscious competence > unconscious competence

In HN, to create a new line, you will need to leave two lines in the editor instead of one. And thanks for the links

Okay, but that doesn’t mean overconfidence isn’t real.

I don’t want to suggest this person themselves is wrong, but it seems that this is just part of the back and forth among scientists and not in and of itself the correct answer.

Meta: what if this researcher themselves is exhibiting Dunning-Kruger and is overconfident.

Reply on news.ycombinator.com