Before I started working at Icons8 I was a system administrator. I was repairing PC’s, doing backups and restoring deleted emails that had suddenly become very important. But sometimes I got really weird requests. Like fixing a light bulb in women’s bathroom (I hope that wasn’t an euphemism) or checking on a microwave. One day I was even asked if I’m good with blenders.

That’s the way people tend to see tech people. Yes, we know stuff: our stuff (in most cases). Yet not all tech-related stuff is our stuff. Ironically, even we often think we know stuff we’ve never had any experience with. Because we too see ourselves as “tech” people.

Thus formed a myth: tech people are tech savvy. Which easily extrapolates into tech people are savvy in everything.

Well, if that’s true, then tech people can easily repair, say… a bicycle.

Hold on. Bicycles? Yes, bicycles. And before you get scared that I’m jumping like spiderman from IT to bicycles, let me explain myself.

If you want to repair a bicycle, you have to know how it works. And if you know how it works, you can easily draw a bicycle. Most people are confident they know how a bicycle works. And yet, if you ask them to draw a bicycle, their drawings may look like this:

Scientists call this an illusion of knowledge. Our brain more readily convinces us that we know something, than it lets us admit that we don’t. National Geographic made a whole episode about this phenomena where 90% of participants drew bicycles unrealistically.

In his famous project Velocipedia, Gianluca Gimini pushed it even further. he asked different people to draw a bicycle and then created 3D-models, based on their sketches:

So, 90% of people don’t know how a bicycle operates. And yet…

In both aforementioned examples (National Georhaphic and Velocipedia project) participants were people with general knowledge (many were students). No specific criteria. This is where I got lucky.

Recently we’ve been training a neural net, whose sole purpose was to recognize icons, sketched by people. We sent an email to our user database asking our customers to draw all kinds of things: cars, houses, trash boxes… and bicycles.

And our audience consists of:

Developers. Half of the people, who we asked to draw a bicycle, were developers. Now the question stands: will 90% of bicycles still be unrealistic, or will there be any improvement given the fact that half of the audience are developers?


I asked my friend, an engineer and avid biker (told you I’m lucky) to help me analyze 200 drawing of bicycles that we then put into 4 different categories:

Not rideable: it’s usually a very primitive drawing of two wheels and a frame, attached the way that prevents wheels from rolling at all.
Ride by rolling:these bicycles can roll, but can’t turn. Or, sometimes, be sat on. So these bikes are for very straightforward people, who have no time to sit.
Rideable (more or less): these have small issues like no pedals/chain or redundant structures of frame.
Totally rideable: people really knew what they were drawing there.

Overall 76% drawings are unrealistic, not 90%.

Developers draw bicycles that are actually rideable slightly more often.

Before we jump to conclusions, however, there are a few important factors that could affect this number.

Constructed Drawing vs. Prepared

Before asking my engineer friend for help with analyzing the drawings, I asked him to draw one. Unprepared and receiving no hints. Here’s how it went:

Take a look at how he doesn’t use some prepared mental shortcuts for elements, but constructs the bicycle. It is a constructed drawing.

The lines may be irregular and proportions may be messed up, but that’s not important. You’ll see that these bicycles were also constructed:

On the other side, there are quite a few designers among our audience (~30%). Their drawings are very accurate.

These are prepared drawings. Fortunately, only a very small portion of all drawings seemed like prepared ones. So I could conclude two things:

  • It does not affect overall numbers that much
  • 30% of our audience is designers, yet only a small portion of drawings were prepared. I suppose not many designers are able to draw a bicycle from memory. By the way, judging by this experiment, even if you give them a reference, it may not be enough.

Other factors

There are few more factors at play here:

Uncontrolled experiment

Drawings were not restricted by time and there were no observers – people could google for a reference. However, judging by the number of unusual bicycles, not many of them did that. Even when given a chance, people still prefer to invent:

“A single designer could not invent this many new bike designs in 100 lifetimes. And this is why I look at this collection in such awe.”Gianluca Gimini, Velocipedia


In some countries bikes are more popular than in others. However, only one of our top 5 traffic countries (Japan) is on the list. I’d love to see how good people are at drawing bicycles in the Netherlands, where there are more bicycles per soul than horses in the Mongol army.


With all the factors mentioned, plus statistical faults etc. the number of people, who draw bicycles unrealistically is 5 – 10% lower among developers than in general. But that is a big stretch, for there are so many factors in play: general popularity growth of bikes during last years, gender specifics (92% of developers are men) etc. Yes, I’d like to think techs are savvy, however,the numbers are not that dramatic to be really sure that’s because of their job.

So I’ll end this article just as I started it. With my own example.

I’ve repaired hundreds of PC’s, configured many network devices. I’ve never repaired a single bike in my life.

So, after watching hundreds of drawings of bicycles, I’ve decided to draw one myself. Not to copy it, but to construct it in my head, from scratch. Not using references and mental shortcuts:

Though I’ve had a half dozen of them, I’ve never had a chance to repair one and never had the motivation. I messed up the wheeling frame. My bicycle can’t turn.

I’m not saying tech people can’t fix things. I do believe tech people are savvy, and repairing a bike is easier than migrating from one framework to another, while covering every bit of code with tests, in a train, using a phone… You got it. Everything is possible, if you have the motivation.

I’m saying that it shouldn’t be expected of them to fix anything. So if you really want me to repair your bike, try giving me some motivation. A cupcake is a good start.

About the author
Andrew started at Icons8 as a usability specialist, conducting interviews and usability surveys. He desperately wanted to share his findings with our professional community and started writing insightful and funny (sometimes both) stories for our blog.