Don’t vibe over what you don’t understand

I was in my first year of college studying programming. Not to brag, but it wasn’t very challenging because I started writing code at 13 and it was breezy. I was sitting with a dorm buddy of mine who was studying mechanical engineering and we were having a conversation that I will never forget. He said something to me which I have heard so many times prior and after that evening. “You programmers are not real engineers. You just sit and type on a keyboard all day. Everyone can do that.”

Fast forward one year and this same guy had to take some coding classes himself because it was part of this program. I met him on the street and … “Dude, this is so difficult. I don’t understand anything. You must help me!”, he says to me. I didn’t bring up our conversation from a year ago, but I realized something fascinating. In defence of their ignorance, some people like to belittle what they don’t understand.

We live in a world with so many purposefully obscured complexities, just so we can enjoy everyday products and services without having to worry about the underlying mechanisms that make them possible. I don’t know much about jet engines and aerodynamics, but I don’t have to worry about that when I fly on a plane because someone made the plane for me to use. I have a phone with a battery in it, and because I’m not a chemist, as a customer I have delegated the job of designing the battery to someone who is competent in doing just that. The world is very complex, but we don’t need to worry about it all. We only need to worry about the domains of our personal competence. It is a beautiful and important thing to be curious. Some people are very curious and have extremely broad spectrum of knowledge, this is admirable, but nobody is an expert at everything.

It is important to realize this and to appreciate the people who make this world tick because we stand on the shoulders of a lot of very smart people. You would be grateful to the doctor who saved your life, but would you be grateful to the engineer who designed the engine of the ambulance car that took you to the hospital? How about the transmission? Also, a very important part of your rescue but taken for granted. Actually, not even considered. We seem to be highly ignorant when it comes to acknowledging our interdependence.

The same principle of obscured complexity applies in full force when it comes to software. Almost everyone uses software daily, some perhaps even too much. I don’t expect regular users to understand how software works because they don’t have to. Just be aware that software is complex, sometimes very complex and the pieces that matter the most require a ton of deep knowledge and understanding to develop. I’m not asking for a medal or a pat on the back but like Aretha Franklin, “all I'm askin' is for a little respect”. It seems however, that software engineers aren’t getting much love lately. Mass layoffs, shrinking job market, threats of being replaced by machines. It appears to me that the media, the public and even some executives of the industry take the news about the downfall of the programmer with a generous dose of schadenfreude.  Why is that?

The image of the typical programmer in the public eye has changed dramatically in the past couple of decades. I don’t think any other stereotype has gone through such a drastic metamorphosis. From the awkward nerdy geek of the last century to the organic soy latte drinking vegan hipster of today. The latter type indeed exists in real life and was allowed to establish its performative presence in the industry around the same time the industry itself realized that making profit is outdated, unnecessary effort and all you need to do is convince some investor that your app-controlled teapot is basically the new iPhone.

Companies, big and small, wanted more free money and the best way to ask for money was to say that you need to “scale up”. In other words, to hire more people. Investors loved “scaling up” because then they could sell the company for more money later, or at least they thought so. This growth bubble created a lot of demand for IT personnel, and the lucrative salaries attracted people from all walks of life to join the booming IT industry.

The apparent ease with which people got hired at tech companies created scepticism around the validity and the value of these jobs among the general society, and rightfully so. I personally know people who got hired after a 3-month programming course to make more money than a young doctor who studied for 6 years.

But wait a minute! Are capitalists really that generous? They sound like champions of equality. Spreading the wealth around to the masses. A true socialist utopia! Unfortunately, no. Investors are still there for profit, and they started to realize that “scaling up” by hiring hordes of IT workers recently retrained from other industries rarely pays off in real tangible performance and quality increases, therefore, it doesn’t increase the value of their investment. As a result, after more than a decade of forced artificial growth, many companies weren’t allowed to continue hiring, had to become profitable and even had to let some people go. Mostly recent hires, juniors and redundant middle management. Also, no more free soy milk in the office fridge.

This first wave of layoffs, which happened just after the pandemic and was pioneered by Elon Musk firing most of Twitter, showed both investors and executives that you can indeed cut headcount and miraculously, nothing much happens. The code keeps on running and the servers keep on serving. This seemingly paradoxical result which aligned with the public’s opinion that software people are overpaid, overhyped, useless keyboard monkeys, is caused by the fact that in most companies, 20% of the people do 80% of the work. In companies previously overinflated to the moon and back with full-time scrum masters and HTML developers, this ratio might be even steeper.

In life, everything that goes too far ultimately triggers overcorrection. The layoffs allowed for some sweet money savings which could be redirected into executive pay checks and shareholder dividends. It seems like we don’t need those lazy, overpaid programmers after all. The businessman who doesn’t know how many bits are in a byte already managed to replace the nerd in running the industry the nerd created in the first place. Now is time for the next step. Wouldn’t it be a dream if we could fire them all and keep all profit to ourselves? Such a beautiful dream, but two problems had to be solved first. How do we run a software industry without software engineers and how do we justify all the layoffs?

The perfect solution to both problems came in the perfect time like manna from the heavens and mixed up with the already shaky state of the industry to create a perfect storm. The introduction of modern Large Language Models (LLMs). Marketed as “AI”. Those are very large statistical models created by absorbing most of the information on the internet, almost exclusively without permission. Since the internet contains a lot of source code, these language models could “learn” the patters of the different programming languages. Natural language comments left in the source code by the original human programmers to help other programmers understand it, and websites like Stack Overflow helped the LLMs connect code with description of utility. This allowed the process to be reversed and now LLMs can output trivial, derivative code in response to natural language prompts.

Oh, joy of all joys, finally we can get rid of them. No more programmers. No more of those leeches who dare ask for money in exchange for their expertise. Everyone can be a software developer now.

LLMs are the weapon of the ignorant and the mediocre. The one who doesn’t understand the intricate complexities of the world around them and was always envious of the talent of others. Delusional and sponsored individuals show up on TV to tell us how it’s a waste of time to study law or engineering now because it will all be replaced by “AI” in five years. Incapable to comprehend that, unlike them, some people study because they have curiosity and it’s their calling and their passion. LLMs give confidence to the incompetent to stand up to the expert and say, “you are now as useless as I am” or “I am now as capable as you are”. Suddenly, we are being told that all the time we invested in studying, reading, practicing, and creating was a wasted time and now we are no better than anyone else down the street. Knowledge was a virtue which is now being reclassified as foolishness.

I hate to break your enthusiasm, Mr. Vibe Coder, but couple of things. When an actual programmer writes code, this code is the description of a unique logical machine that they created in their mind with the purpose of solving a problem. There is this abstract device that the programmer imagined, and the code is merely the way to describe the device to the actual physical computer, so the computer can materialize it and let it run. The hard part of programming is not typing the code but coming up with the best device to solve the problem. Just like the hard part of writing music is not writing the notes but coming up with the melody. What the LLM does for you is steal other people's mental devices per your request and serve them to you. That’s all there is to it.

Without understanding the device, there’s nothing you can do to repair it, change it or improve it. You are not a creator; you are not in control. You will also never invent anything because in this situation you are just a user of someone else’s work. Just like I’m a user of my car, not a mechanic, because I don’t understand the intricacies of the engine well enough to repair it without destroying it.

I know how much you want your name in lights, just like a real engineer, but the LLM will also never invent anything unique for you to claim, simply because it can’t, it can only steal the inventions of others. And even if it could, then it can also invent it for everyone else, not just for you. So don’t expect to be filing any patent applications anytime soon.

I’m sorry to disappoint, but the shameless arrogance of the “AI” companies and their devoted early adopters might prove to be a bit premature. Thanks for revealing your true intentions, all is noted, but I think we will be talking again in 2 to 3 years when every single software project has been turned into an unmanageable mess by Bob, the previously unemployed cousin of the doorman, now a vibe coder.

I’m not quite sure who the fool is here. Is it me for spending more than 20 years to get good at this or is it the vibe “coder”, who thinks their little “AI” generated React app is the pinnacle of computer science? Let me clarify something for you. The existence of your app is only possible because there is a web browser to interpret the assorted scraps of JavaScript you were given. This browser itself runs within an operating system which consists of hundreds of user space tools, APIs, frameworks and a kernel. The kernel employs drivers to talk to devices which run various types of firmware, all of it written by programmers. And we are not even discussing the hardware here. All of this working together, millions of lines of code, thousands of function calls, billions of instructions executed every second, just so you can see your slow, inefficient note taking app or game of “Snake” on the screen.

It took thousands of talented people decades of hard work and innovation to build the giant and deep iceberg, that is the computer software infrastructure of today. I know that you are ignorant of the complexity of what you can’t comprehend, but if you think a fancy autocomplete machine has the slightest chance of maintaining this monumental human achievement, expand on it and keep it running for the years to come, then you might be in for an unpleasant surprise. Just please, don’t vibe over what you don’t understand because then other people will have to fix it after you.