Increasing the awareness of the Dunning-Kruger effect?

Question

I recently came across the Dunning-Kruger effect and I think it is well represented here and we should be aware of it.

Basically, it states (citing wikipedia):

• Unskilled individuals suffer from illusory superiority, mistakenly rating their ability much higher than is accurate.
• Those persons to whom a skill or skills come easily may find themselves with weak self-confidence

Which is depicted in this plot found on the web:

This sounds like we are observing this concerning our treatment of unskilled newcomers by hi-reps skilled peoples.

For instance (and I deliberately forge an extreme example here, for illustration purpose)

An unskilled amateur EE comes here and ask about his new fancy project:

Do a 5.1 audio real time processing board

He is able to blink LEDs on his new Arduino. He can play some beeps using a PWM and a buzzer and he knows that some shield exist somewhere that have mic inputs.

He stands on the left or my illustration picture. He is not aware that there are things he doesn't know. For him, this looks not so challenging and he is quite confident that he can make it without too much effort.

Then comes the hi-reps skilled EE. He stands on the middle-right of the picture. He is aware of the complexity of things and he knows that he would never be able to really master anything. Things are too complex. Models always have limits. etc.. He is quite pessimistic, and probably over pessimistic. He also know that the OP is far less skilled than himself.

The result is often:

• This is too broad!
• One would require a entire book just to understand the basic principle of audio processing.
• We should close this question. Answering is a waste of time. the answer would go over your head.

For skilled people, who like to do things the right way, the devil is in the detail. A simple LED blink project often ends up asking:

• Did you use proper decoupling? What is the length of the leads of your capacitors? Unwanted inductance from capacitor leads would decrease the decoupling performances of your power supply.
• Did you compute the thermal dissipation of your linear regulator? Did you take a margin with the ambient temp?
• Are you sure that your linear regulator has enough dropout? Here is the graph of the required dropout according to the current drawn from the regulator.
• I recommend that you use this type of capacitor because the MTBF is far better.
• Using thermal balancing on your footprint would increase the yield of your soldering process by some ppm.
• I recommend using a continuous GND plane, this would be far better for EMC/EMI performance. And I suggest that you refer to the Ott book and read it at least twice before trying to layout your first board.

Well, this guy wanted to blink a LED using a PIC on a breadboard...

This is humorous on purpose. But what is the best answer to the unskilled amateur guy:

• Be confident and try to make its LED blink by getting the right level of answer? That way he could have some results. Learn from his mistakes. And then go to something more complex, learn again. etc..

• Getting overwhelmed by important, but less significant to him at the moment, information? The guy would be afraid and won't even try something. This is too difficult, it's for the pro

According the the Dunning-Kruger effect, we, skilled EE, will almost always over-estimate the difficulty of a task. And it may not be a good thing to share this pessimistic view with the newcomer. Or at least we should be aware of this to adapt our answers.

It's like during you first day at your EE school your professor said: Do you see all the books on the bookshelf behind me? 240 books (E24 series). I read all of them and I have been teaching electronics for 15 years. I am still not mastering everything...

Here you go back and think: Wow this looks difficult! I think I'm gonna choose theoretical physics instead...

Answer 1

Thank you for posting the theory, graph and explanation. I recognise that behaviour in myself, and others sometimes, in my 'day job', but didn't have a name for it.

Beyond my 'day job', I 'teach' Arduino in workshops and classes to hundreds of people aged 11 to 80+ for 7+ years. Edit: in this context the Dunning-Kruger is helpful; success encourages people to tackle projects which are so complex I might be tempted to divert them to something easier.

My recent favourite was a 6yo who happily wrote a small Arduino-style C++ program from scratch, and got it working with almost no help. Initially, I quietly suggested to his dad, his son could start with a ready-to-use, working example that he could build on. His parents smiled and said, "no, he really likes to type".

I try to work on the basis that people achieve amazing things, with a bit of support and encouragement, when they really want to. I try to cut people some slack, and not let them fail miserably.

Edit: I try to help them succeed in their own terms, and not impose my standards.

I have met very few (but there are one or two) who are so out-of-touch with their ability that I can't help them learn or make progress. A friend calls those 'fantasists' 'self-swatting flies', they actively work to undermine themselves, and don't know it. I don't see my role as weeding those 'fantasists' out of the gene pool, so I try to be honest and clear about what they need to do if they want to use some more of my time.

I was lucky to work with a 13/14yo, who inspired me. He has taught me stuff, and more importantly has shown most things can be achieved. He won one of the UK micromouse competitions 7 times in a row, and explores all sorts of fun science and technology stuff in his spare time, well beyond what the school could offer. He is now at a very good UK university, doing EE, about to start his second year.

One 'trick' has been to encourage the 'student' to break things down into do-able pieces. I try to help them plan next steps, prioritise value, or rate difficulty. They figure out how to do the project. By making mistakes yet having support, I believe they learn faster than if I 'tell them the answer' or try to 'pre-think' on their behalf.

IMHO, humans are extremely good at figuring out how to do things, and learning from mistakes, when the goal is important to them. After a few mistakes, students learn to 'triage' possible options, figure out when to seek advice, and how to use that advice. Advice offered which creates complications, ahead of time, is rarely welcomed or acted upon. Some people need to learn how to be flexible if they are going to progress quickly; failure is a way to cultivate flexibility!

I worked with a product development guy many years ago, and he said "fail fast, fail cheap".

I think it applies to learning (and life :-). Failure is not bad. Failure tells us something. It is the manner of failure, and what happens next that is important. We don't need to ensure learners always avoid failure. However, learners might need help to consider the impact before hand! So I'll put 'failure' in quotes because it is slightly different from the day-to-day meaning.

I try to encourage folks to understand what 'failure' might be telling them. I try to use my experience to try to alert them to the possibility and scale of 'failure', and ask them what they might learn from that possible 'failure', or whether there is a way to reduce the cost of 'failure'. I am hoping to turn steps-into-the-unknown into an experiment which teaches us something quickly and cheaply, and hence they can afford to do it.

However, my experience should not be the important thing. They need to learn to do things without my help. So we have to also focus on a process which moves them forward. It needs to be a process which builds confidence, and which they can use and improve.

I try to use iterative, agile, development as a model; small, cheap, understandable, steps which produce results and evidence, rapid feedback, review with others, learn and improve.

IMHO, a very light-weight, process-centric approach is a way to handle the problems in that Dunning-Kruger model, harness expertise and enthusiasm, and reach realistic levels of confidence while making rapid progress.

Full disclosure: I am no magician, and I don't always succeed.

Answer 2

Note that the authors of the chart have picked the over-confidence level of a beginner as 100%, while the expert's level of confidence is less-than-100%. Perhaps, there is a point in that.

An alternative way to scale this would be to pick the expert's level as 100%, then beginner's over-confidence would be more-than-100%. That would be as unrealistic a number as over-confidence itself.

May be, the best way to scale this would be to pick 100% somewhere between expert's level and beginners over-confidence. That would illustrate both: that an expert still has a healthy amount of doubt, and that over-confidence is unrealistic.

Shoshin (beginner's mind)
On a different note, Zen Buddhism has a concept of Shoshin meaning "beginner's mind".
It refers to having an attitude of openness, eagerness, and lack of preconceptions when studying a subject, even when studying at an advanced level, just as a beginner in that subject would.
It would seem that they were aware of the peak of eagerness and confidence (genuine, albeit under-informed), and wished to transmit it to the more experienced practitioners.