Vibe coding is just another layer of abstraction

There is a lot of noise about how vibe coding will play out. What it means for software, for jobs, for people learning to code right now. I have mostly been reading rather than posting, and my opinion has moved around a few times. Right now I would call it wait and watch. Both extremes seem wrong to me, and it is worth spelling out why.

On one end you have the skeptics. It is all hype, the models are parrots, this ends the way crypto ended. On the other end you have people announcing the end of all knowledge work by next quarter. I don't buy either story. The honest middle is more interesting.

So here is where I actually stand.

1.Vibe coding is real, and it's a force multiplier.

AI assisted programming is a real force multiplier. I use these tools every day and they make me faster. Things that took weeks a few years ago take days now. No argument there.

But the useful question is not whether this is transformative. The useful question is what kind of thing it is in the long history of programming. Ask it that way and the answer gets a lot less dramatic.

2.Programming has always been a stack of abstractions.

The story of programming is one abstraction stacked on another. First the logic lived in the hardware itself. Then came assembly. Anyone who has written assembly knows how painful it is to just add two numbers. So we built high level languages on top. COBOL, FORTRAN, and then C, which opened the floodgates.

From C we got C++, then Java, then Python and the rest. And when you write Python there is still a C layer underneath. The stack never went away. It just got hidden.

Vibe coding is the next layer up. You write English and a model turns it into working code. That is really all it is. A new abstraction on top of the old ones, the same way C sat on top of assembly. Still impressive, just not unprecedented.

3.The skill that matters is peeling back the layers.

Every working engineer learns this eventually. The abstraction you are standing on will fail you at some point. You pick up a library, learn the API, ship things. Then one day something breaks in a way the docs don't explain. You can wait for someone on GitHub to answer your issue, or you can open the library and read the code. Sometimes the bug is one layer deeper. Sometimes two.

That is the skill. Being willing to go down a layer when the problem demands it. Vibe coding doesn't change this. It adds one more layer you might have to dig through. When the model writes something that almost works, you still have to read what it wrote and understand why it is wrong.

4.Top-down and bottoms-up are both valid. The problem decides.

People argue about how to learn programming. Start from the fundamentals and build up, or start from the API and dig down when you have to. I don't think either is right on its own. The problem in front of you decides.

Building a small internal dashboard nobody else will use? Top down is fine. Vibe code it and move on. But even internal tools live longer than you expect. Software is never really done. A dependency goes stale, an integration breaks, one day you need backups. The maintenance tail is the part the vibe coding crowd underestimates.

5.Public products are a different game.

Anything public facing, anything that is actually a business, is a different beast. You cannot type "build me Facebook" into a chat box and get something that holds up. It might work for ten users. Past that the cracks show. Security, scale, edge cases, payments, identity, regulation.

A quick prototype? Sure, vibe code it. A system that survives real users? Not yet. Not without people thinking carefully at several layers at once. Taste and judgment still matter, and they will keep mattering for a while.

6.Follow the money on the hype.

If you want to understand the loudest takes, follow the money. LLM providers benefit from saying AGI is around the corner, because that justifies the valuations. Course sellers benefit from saying your job is disappearing, because fear sells courses. I am not saying these people are lying. Many of them believe it. But the incentives point toward the dramatic version of the story, and you should discount for that.

Will knowledge work change? Sure. It always does. On the timeline being shouted about in podcast clips? I doubt it.

7.What about students?

Should people still study computer science? I think yes. Maybe more than before, because the people who understand the layers underneath will be the ones who can fix things when the shiny layer breaks.

Here is an analogy I keep coming back to. I was at a store recently and saw cheap clones of well known products. On the shelf they looked fine. In the hand you could tell instantly. The weight, the materials, the way the plastic flexed. The difference was invisible at a glance and unmistakable up close.

Software is heading the same way. There will be a flood of generic vibe coded apps and some will do fine. But there is also a market, a growing one I think, for software that was clearly made by someone with taste and intent. The MacBook versus generic laptop gap never went away, even when the specs matched. People who build software for a living can usually spot the generic app within thirty seconds of opening it.

8.On gatekeeping.

None of this is gatekeeping. Someone with no formal background can build genuinely good things, and the lower bar to entry is mostly good news. But the people who succeed will be the ones willing to dig through the layers when the problem demands it. Not always, just often enough that the willingness becomes the difference.

You don't need a degree. You need the persistence to keep going when your current layer doesn't have the answer, and the curiosity to look one layer down. Generalist problem solving. The thing that has always mattered.

So, wait and watch. Not fence sitting. The honest answer to how this plays out is that nobody knows yet. The skeptics are wrong about the size of the change. The maximalists are wrong about the speed. The truth is in the boring middle. A real shift, slower than the hype, with the same fundamentals underneath.

If you are building right now, use the tools, ship faster than you used to, and stay willing to go down the stack when something breaks. That part has not changed. I don't think it will.