Hacker News Reader

> when we learn an overview about our new topic our confidence (in understanding) greatly exceeds our competence, then we learn how much we don't know and our confidence crashes below our actual competence, and then eventually, when we reach mastery, they become balanced.

As a description of what Dunning and Kruger's actual research showed on the relationship between confidence and competence (which, as I've pointed out in another post in this thread, was not based on studying people over time, but on studying people with differing levels of competence at the same time), this is wrong for two out of the three skill levels. What D-K found was that people with low competence overestimate their skill, people with high competence underestimate their skill, and people with middle competence estimate their skill more or less accurately.

As a description of what actually learning a new subject is like, I also don't think you're correct--certainly what you describe does not at all match my experience, either when personally learning new subjects or when watching others do so. My experience regarding actually learning a new subject is that people with low competence (just starting out) generally don't think they have much skill (because they know they're just starting out), while people with middling competence might overestimate their skill (because they think they've learned enough, but they actually haven't).

Out of curiosity, if you google "Dunning Kruger Effect" and then look at the images, and see all those charts of confidence vs competence, do you disagree with them?

And as I responded to your other comment, just because the study took measurements of many people at one point in time didn't mean they weren't studying an over-time phenomenon. No one starts out as an expert.

> if you google "Dunning Kruger Effect" and then look at the images

Then I would be doing a very poor job of actually researching the topic.

If you are aware of any actual research papers that describe actually studying how the same people's competence vs. confidence evolves over time, by all means point me to them.

> just because the study took measurements of many people at one point in time didn't mean they weren't studying an over-time phenomenon.

Um, yes, it does. It's possible that the things the study saw were effects of an over-time phenomenon, but the study did not study that, and its results are not evidence either for or against any such hypothesis.

> Then I would be doing a very poor job of actually researching the topic.

I made the request to try and tease apart where we disagree exactly. Your interpretation and response to my request/offer comes off as unnecessarily disrespectful. (unkind, elitist and tramples curiosity)

It seems increasingly likely that I was mistaken in thinking we could explore this subject together. It seems you are intent on attacking my position and discrediting my points and not interested in building a shared understanding.

I find this disappointing.

> I made the request to try and tease apart where we disagree exactly.

I thought that was obvious: you're making a claim about how individual people's confidence vs. competence evolves over time, that as far as I know is not supported by any research--certainly it's not supported by Dunning and Kruger's research. That's why I asked you if you know of any research that does support it. "Images found in a Google search" does not qualify.

> It seems increasingly likely that I was mistaken in thinking we could explore this subject together.

That's the way I see it too, though evidently not for the same reason you do.

> It seems you are intent on attacking my position and discrediting my points

I'm interested in finding out what basis you have, if any, for the claim you made that I described above. "It's self-evident" is not what I was looking for. Nor is "do a Google search and look at the images".

> and not interested in building a shared understanding.

If you can point me at research I wasn't previously aware of that supports your claims, that would be helpful. If you want to try to get to "a shared understanding" based on claims about what you think is self-evident or by looking at images from a Google search, then no, I don't think we're going to get anywhere.

Yeah, I screwed up here.

I don't have a background in psychology or sociology (academically) but they are subjects that interest me.

I was introduced to DK by pop psychology, especially in business management contexts.

The reason I wanted you to comment on the viral graphs was because I was trying to triangulate our disagreement.

Turns out that there is a ton of misinformation about DK. Kind of ironic given the topic of this thread, but especially because it long predates LLMs. If anything it hints why LLMs are confidently incorrect, since the internet is full of it.

The reason why what I was saying is not self evident is because I was making a logic error, specifically ecological fallacy.

Also sort of ironic, I managed to get to the root of this using an LLM.

Part of me suspects psychology is your domain, and you may be an academic or at least have some graduate experience. The way you engaged with me is maybe the way you've been conditioned to act in these situations.

There's good reasons for why you responded the way you did. Part of me wants you to know that I would have really appreciated it if you had said something like "while I dont take issue with the claim that competence increases with time, it doesn't mean the DK relationship is time based, as there could other conflating factors." Also "I looked at those graphs and they don't match what the DK paper claims, that looks like pop psychogy and I can't confirm it has a scientific basis".

Or course you have no obligation or responsibility to help me improve my understanding, but the fact you responded to a bunch of my comments shows you care. So if you care about changing minds, especially the mind of someone who is curious and ignorant, than engaging this way would have made my interaction with you more pleasant.

And it's in the spirit of the community guidelines.

> Part of me suspects psychology is your domain, and you may be an academic

Neither of these are true.

> or at least have some graduate experience.

I do, but not in psychology. My degrees are in engineering.

> I would have really appreciated it if you had said something like "while I dont take issue with the claim that competence increases with time

But I do take issue with that claim, at least if it's made in the broad way you are making it here. I responded to that in another subthread with you.

> "I looked at those graphs and they don't match what the DK paper claims

The DK paper was already linked to by someone (I don't think it was you) elsewhere in this thread, and in any case it's easy to find online. I was taking for granted that you would be able to look for yourself to see what is in the paper, graphs or otherwise.

> Dunning-Kruger effect is more about how confidence and competence evolve over time.

I don't think there is anything about this in the actual research underlying the Dunning-Kruger effect. They didn't study people over time as they learned about new subjects. They studied people at one time, different people with differing levels of competence at that time.

It's the relationship of confidence and competence.

Competence is gained over time.

The "over time" is almost self-evident since no one is born a master of a skill or subject. And if it's not self-evident enough for you, there is lots of research into what it takes to develop competency and mastery in any subject.

So while the original paper was a snapshot taken at one point in time, it was a snapshot of many people at different stages of a learning journey...

And journeys take place over time.

I responded to this in another post in response to you upthread. The short version is that I don't agree with you about what is and is not "self-evident" here, nor will I accept your unsupported word that whatever research does exist supports your claims.

Interesting.

So just to be clear...

You are claiming that it is not self-evident that people learn things over time?

That it's not self-evident that experts start out as beginners?

Please respond in the other subthread where I made my points in more detail. As you will see from that more detailed post, I never made either of the claims you are asking about here.

You could chart the same curve by measuring the confidence of people at different competence levels.

"Could" based on what research?

Yes and no. Dunning-Kruger also explains this evolution of skill estimation, but the original paper frames the effect specifically as an overestimation of skill in the lowest-performing quantile. This is clearly even cited in the article.

Okay I will agree with the "Yes and No". I initially clicked the source in the article, which is a broken link to wikipedia and rolled my eyes at it.

After reading your comment I navigated to it directly and found the first two sentences:

The Dunning–Kruger effect is a cognitive bias that describes the systematic tendency of people with low ability in a specific area to give overly positive assessments of this ability. The term may also describe the tendency of high performers to underestimate their skills.

Unsatisfied that this was the authority, I dug up the original paper here:

https://pubmed.ncbi.nlm.nih.gov/10626367/

And sure enough, the emphasis in the abstract is exactly as you say.

So while I stand by that the principle behind the "effect" is how confidence and competence evolve over time as someone discovers and masters a domain, I will concede that the original authors, and most people, assign the name for it to primarily the first phase.

Here I was thinking the OP Author was DKing about DK, but in reality I was DKing about them DKing about DK.

> the real "effect" is how confidence and competence evolve over time

What research is this based on?

Yeah that was poorly worded. I edited it even before I read your comment.

However the over-time aspect is kind of self-evident. No one is born a master of any skill or subject.

So while the original research was based on a snapshot of many people along a developmental journey, just because the data collection wasn't done over time doesn't mean that the principle behind the effect isn't time/effort based.

The graph is really competence vs confidence, but competence can only increase over time. If this isn't self-evident enough, there is lots of research on how mastery is gained. I don't have time to throw a bunch of it at you, but I suspect you won't need me to in order to see my point.

> the over-time aspect is kind of self-evident

It's self-evident that, in order to become a master at anything, you have to pass through increasing levels of competence to get there.

It is not at all self-evident, at least not to me, that while a particular person is going through that process, their estimate of their skill vs. their actual competence will be similar to what Dunning and Kruger found when they compared different people's confidence vs. competence at a single time.

> just because the data collection wasn't done over time doesn't mean that the principle behind the effect isn't time/effort based.

It could be, but a study that only collects data at one point in time can't possibly provide evidence either for or against any such hypothesis.

> competence can only increase over time

This is simply false. Skills and knowledge that aren't used can deteriorate. People can forget things they used to know.

In a much narrower sense, we would hope that if a person is studying a subject with the intent of mastering it, that their competence will only increase over time. But even that's not always true.

> there is lots of research on how mastery is gained. I don't have time to throw a bunch of it at you, but I suspect you won't need me to in order to see my point.

Sorry, but since, as you can see from the above, I don't agree with several of your key points, you're going to have to back them up with research if you want me to consider them further. If you don't have time for that (which is fine--we're all posting here in our spare time), then we'll just have to disagree.

Gell-Mann amnesia

Bingo