Economic Arguments Against Vibe Coding \ stacker news ~econ

pull down to refresh

Economic Arguments Against Vibe Coding www.bitcoinandmarkets.com/economic-arguments-against-vibe-coding/

1812 sats \ 16 comments \ @BlokchainB 6 Oct econ

This is as a thought provoking piece by Ansel who I am a big fan of.

Every hour spent vibe coding is an hour not spent on the work that compounds your edge. That tradeoff is the essence of opportunity cost. Time is the scarce resource. If your highest-value activity is selling, writing, managing clients, researching, or leading a team, then time diverted into building an app is time taken from the activity that moves your scoreboard.

Comparative advantage says you are most valuable when you spend your time where your relative strength is greatest. AI can help you write code, but it does not erase your unique advantages in domain knowledge, relationships, judgment, and distribution. A real estate agent who wins listings, a nurse manager who improves staffing outcomes, or a researcher who generates trusted insights creates more value by deepening those strengths and partnering with a builder, not by becoming a part-time software engineer.

Made me rethink my approach to leaning to code. Maybe I am best not wasting my time and money on a Replit app.

view all related items

298 sats \ 7 replies \ @SimpleStacker 6 Oct

I don't think know if this is accurate. I am not primarily a coder, but as a researcher I cannot afford to pay someone to code for me. Thus, I need to do the code myself, and AI can be pretty helpful with that (though I haven't found 100% vibe coding to be useful.)

I think there are lots of people in a similar boat, where their specialization is outside coding, but they have a vision that requires some coding and they can't afford to hire a professional coder for that purpose. AI fills that gap fairly nicely.

42 sats \ 6 replies \ @BlokchainB OP 6 Oct

But the article is emphasizing if the market needs or even cares about your vibe coded app

281 sats \ 4 replies \ @optimism 6 Oct

Don't code apps. Code tools.

100 sats \ 2 replies \ @BlokchainB OP 6 Oct

This is hard to do. Plus are you coding tools for self use or for the general market?

123 sats \ 0 replies \ @optimism 7 Oct

I'll work on a bigger post about this. I started a reply but it grew rapidly so I need to take a step back and do the human inference thing to make it coherent.

185 sats \ 0 replies \ @optimism 7 Oct

The answer you probably didn't want: #1250807

100 sats \ 0 replies \ @justin_shocknet 6 Oct

More people need to understand this framing.

0 sats \ 0 replies \ @SimpleStacker 6 Oct

That's hard to say, right? I mean, the creator of Stardew Valley basically did everything himself. Someone might have told him, "Don't do the artwork yourself, hire a professional artist," or "Don't write the dialogue yourself, hire a professional writer." But in the end, he made his vision and it turned into one of the most successful games of all time.

286 sats \ 0 replies \ @optimism 6 Oct

If your specialization is in coding, then it is unlikely that you should (regardless of whether you actually do it) be vibe coding for production work. There are much better patterns for professional use, like using it as a sounding board, prototyper, or even as a tester (if you are actually going to read all the slop it produces.)

If you're not a pro coder, then, as long as you don't be vibe coding for production work that you cannot build yourself, you're fine. I use it a lot in R&D when I otherwise can't be bothered to spend the time coding something up, because I want to emerge in the subject and not spend hours building tooling to enhance the research. For this, it is awesome: tool building.

But you have to be able to make decisions on what the LLM should produce for you, and this is the hard part. LLMs aren't really trained for this role so it's not like you can ask it how it can help. Look at a coding LLM as a junior data scientist or staff coder that you have to carefully instruct, and only when needed, otherwise they will waste your time. If you're rich, you can hire a non-LLM (i.e. human) to instruct the LLM for you: this is a real skill and it is not lazy work.

Honestly, the only way to develop the skills to instruct LLMs coding tools for you is by spending the time, and by making mistakes. Should everyone do it? No. Should you just ask your LLM what it could do to help? No. But you can definitely vibe code a need up, for something that you're doing manually right now, or that you're postponing because it takes so much manual labor. And if it sucks then you learn.

70 sats \ 2 replies \ @fourrules 6 Oct

This is bullshit. There is a huge amount of inefficiency is alignment issues at the beginning of ventures that fundamentally don't need the complexity of multiple people with even slightly different visions or understandings or priorities, even in the cases where both visions, if committed to, would lead to success.

Often finding product market fit requires basic knowledge of a domain not in your wheelhouse because the opportunity exists at an intersection. By vibe coding a solution, which is like no-code on steroids, a CTO can be brought on for far less equity or even as a first hire, resolving the alignment problem entirely.

0 sats \ 1 reply \ @BlokchainB OP 6 Oct

I beg to differ. Applications that don’t have strong technical support are often bad. This is one of my main criticisms of Bitcoin applications the technical support is absolutely horrible. Teams relying on telegram for technical support is all you need to know.

So even if your vibe coded app gains traction you better have capital to hire a team and technical support for it to survive in the market!

0 sats \ 0 replies \ @fourrules 7 Oct

If you're finding product market fit it's not about the quality of the finish. I've worked with many developers and the vast majority of them are terrible founders used to getting paid and overly conscientious, they don't understand (as you don't seem to) the process of getting from zero to one.

A baby isn't a scaled down adult. Its it's own thing, optimised only for coming into existence.

So even if your vibe coded app gains traction you better have capital to hire a team and technical support for it to survive in the market!

The money and equity you save by agentic development and no-code tooling is far better spent after product market fit than before.

21 sats \ 0 replies \ @itsrealfake 6 Oct

Made me rethink my approach to leaning to code. Maybe I am best not wasting my time and money on a Replit app.

It sounds like you're learning to "vibe code" instead of learning to "code". This could be an important distinction in your decision matrix.

Arguments around comparative advantage are fine, but remember the advantage of having a good time also.

86 sats \ 0 replies \ @BitcoinIsTheFuture 7 Oct

Be lean and agile. 168 hours in a week. Gotta maximize it!

190 sats \ 1 reply \ @justin_shocknet 6 Oct

I like Ansel too but this is a terrible take

Coding, vibe or otherwise, is just a means to an end. Code is a manifestation of creativity, creativity that is informed by your unique domain knowdge/relationships/judgement/distribution

Vibe coding makes applying your uniqueness faster or more accessible, which means more people can try things in the market more often.

Time is indeed the scarcest resource, vibe coding is leverage on that time. It reduces the cost of experimentation which insures actually good ideas can be proven faster.

10 sats \ 0 replies \ @BlokchainB OP 6 Oct

But Ansel is making the case that it doesn’t increase productivity when people are vibe coding as a side hustle.

I don’t disagree agree with your point. I was vibing for a bit to make my dream application but it wasn’t going to increase my productivity or bring value to me in any monetary way.