On the topic of taste, I think of this quote in Susan Sontag's "Notes On 'Camp'" often:
“For taste governs every free — as opposed to rote — human response. Nothing is more decisive. There is taste in people, visual taste, taste in emotion — and there is taste in acts, taste in morality. Intelligence, as well, is really a kind of taste: taste in ideas. (One of the facts to be reckoned with is that taste tends to develop very unevenly. It's rare that the same person has good visual taste and good taste in people and taste in ideas.)
Taste has no system and no proofs. But there is something like a logic of taste: the consistent sensibility which underlies and gives rise to a certain taste. A sensibility is almost, but not quite, ineffable. Any sensibility which can be crammed into the mold of a system, or handled with the rough tools of proof, is no longer a sensibility at all. It has hardened into an idea..."
For those keeping track, this is now (at least) the third wholly or partially AI-generated article on the theme of "taste" as enduring source of human comparative advantage to hit the HN front page.
I have mixed feelings... I stumbled upon, read a post-mortem that looks even more AI-generated than the article.
"That says something about my writing. It says nothing about who wrote it." - Two contradicting sentences, author's writing skills are part of the author as-a-whole
"The short sentences, the reversals, the one-word lines" and then "I will write the next one plainer.". - even more?
And... I wonder if this is like some agent wrote article, and, in the loop, added port-mortem, after visiting hacker-news comments and manipulate that this is human written. Or maybe author is just trolling us with AI-like prose...
I don’t like the term taste, but the problem that I have is that LLMs don’t seem to work “good enough”. They seem to be able to solve the immediate problem, but stacking this on the scale of 3-4 devs over 6 months or so doesn’t seem to produce anything.
One thing that I’m particularly frustrated with is the writing quality of LLMs. Like this is the thing that they should be able to do, but I would say almost everything they write has almost no signal.
Over a mid sized AI generated codebase that means I’m reading like 500 words to figure out what a module is even doing.
We've been talking about taste for a year now as it relates to AI, so I'm going to counterpoint it: if competitors can reproduce your features, UX, and visual decisions within days, your taste is not really an advantage. How many apps implemented pull down to refresh once one person did it? How many people even remember (or care) whose taste spawned that UX pattern? [1]
AI is not inaugurating an age of taste. It is shortening the half-life of taste and commoditizing software. What happens when software features, UX, and visual fidelity aren't competitive advantages anymore? If you ask me, it looks like business did in the 90s. We'll probably see the return of department store software in the next few years.
This article resonated tremendously with me, speaking as someone who has been coding since the 1980s and had to develop my own taste the hard way, through lots and lots of mistakes. Whenever I see a software demo built by someone mostly using agents to do the work, I have to question whether it's any good on the inside - or whether any real intuition or judgment went into its design. On the flip side, maybe none of that matters anymore - as long as it works, who cares how it was built.
I've run this thought experiment a number of times with different sections of my (relatively) diverse friend circle. I ask them a version of this question on repeat: "what if AI automates X aspect of your work?", and the final answer (either because of frustration with the loopy nature of the question or a genuine stopping point) arrives somewhere along the lines of "my judgement," "my knack," "my embodied nature," or "my quality of decision making."
I think the core still leads back to human agency and the ability to consider aspects that wouldn't fit into an LLMs limited context window or be able to be vectorized into a DB, including ultra-long-term consequences (especially those with great thinking abilities).
I still happen to think that AI/LLMs will never be able to "fully" replace humans, because the evolutionary process that led to our cognitive abilities and the way we train LLMs are vastly different thanks to different pressures, but maybe that's just me defending the last bastions of our collective humanity as a human; I don't know what else to root for.
Hard disagree on "Good enough", most things are mediocre at best. Despite the massive AI boom I can at best count on one hand "fresh" products made with AI that are meaningful in day-to-day usage. Not to mention most of them seem to be targeting AI-space itself instead of bridging some actual business use case. Just look at all the products on ShowHN, you can "tell" a vibed product in seconds.
Generating the first version is almost free now. Understanding why that version is subtly wrong, debugging it in production, maintaining it six months later and deciding whether the abstraction was a mistake are all still expensive...
Taste was all that was left when product design peaked in like 2010-12 too and it was a great time.
But the advent of design systems and software like Sketch and Figma pretty much automated the job away. You didn’t need the raw Photoshop skills anymore - any CMU grad could be a top tier designer now.
I see the same thing basically happening here. Taste will matter for a while but eventually people are just choosing the most accepted/expected optimal choices and anything that falls outside the trend will be “not good” in the same way skeuomorphism in design is no longer good.
So we’re lucky we’re in a taste era - it’s a lot of fun. You (yes you!) can still change it, shape it, establish the status quo of tomorrow
> Suppose you have taste. Suppose you paid the full price and you can feel the sag in the sentence and the wrongness in the function.
>
> Congratulations! You now ship at exactly the same speed as the person who cannot.
Yes, previously the person who didn't understand shipped faster. It was a long standing problem. The people who crapped, got the bonus, and ran were much faster than the patient engineer of quality.
By the way, what were the essays that formulated Claude Style? if the author was indeed formed by them, I'd like to read them - or at least enough of them to understand.
This post is awful! Annoyingly written and zero insight. Most posts about this taste concept are bad (and there have been so many), but this is below average.
I keep seeing people trying to define judgement and taste and then they don’t actually like cite Kant’s Critique of Judgement where this is basically the entire concern.
Also just a shout out to the author - you're getting a lot of flak here, but your article hit me at the right time in the right way. I trust you're telling the truth about writing it yourself, and thank you.
There's a lot here that I enjoy muddling on. And I think taste is very vital. Taste, ambition, seeing possibility: these resonate to me, long resonate, as what I've wanted to see.
And honestly: I've been tired of an age of engineers that are beat down, that only want simple, that finding the most blunt approach is the only way. It has felt tasteless. Anti-ambitious. It often is still very sensible and practical and what you should shoot for! But there are also people wandering around now trying much harder! And I am excited for those frontiers! I think it will challenge and shake the foundations that we've accepted as true, as There Is No Alternative, in exciting bold and fun ways.
I am however pretty unconvinced by the article. There's a lot that doesn't quite work for me, that isn't building the case I'd like, whose takes are off from my read.
> The output is good enough—that is the problem. Good enough is a solvent. It dissolves the reason to do better. For as long as making things was expensive, the expense did quiet work on our behalf. It rationed output.
Agreed about the rationing of output. I do think that the article though continues to show a magical thinking. That we have these things now and they autonomously do the thing. That the LLM's have solved it all.
The article itself goes on at length about how nebulous and abstract taste is:
> Taste is that. It is the compressed, wordless verdict you reach faster than you can justify. It is the “no, again” you say to yourself with total conviction and no available argument
But this contrasts so distinctly to me against "the output is good enough". Is it? That depends. That depends on your taste. The proximate first results come quick. But the technics underneath? Those themselves, in my view, rely enormously on engineering taste to support and advance. I think we see a very similar sort of magical delivery thinking, very clearly on display here:
> When the factories came, they could suddenly make everything—cheaply, uniformly, by the thousand.
As if there was some magical "good enough" transition where suddenly the aliens came and gave us this box that just does the thing. As if we discovered the right formulas and math and now: we had production. Again I think there's just an enormous amount of work and taste that is still actively required inside the factory box, that building the industrial processes is incredibly intense & difficult, even though we have reliable industry-line production and now robots doing the labor.
There's still so many gems, so much lovely material for thought. I love the provocations here, and I think there's a lot of great calibration.
> The friction was not an obstacle to developing taste. The friction was the curriculum.
Very nice post, I think it captures and expresses things that many people in the software and design worlds have thought about a lot in the last couple of years.
I don't fully agree with it — for one thing, I think you can develop taste in a field without actively participating, but you have to engage with it beyond passive consumption. Basically: I think it's possible to be a good movie critic without making movies yourself. Given the collapse of the friction/effort side of things, I think that is really worth unpacking.
I have to say I find the debates about the post's style and em dashes pretty disheartening. There have always been plenty of good jokes to be made about HN's culture, but wow. For a community that has been snarky about fools counting lines of code for 40 years to be out here counting em dashes is pretty damning.
These artsy fartsy takes on a subject that would be better studied as a science are fun but are they useful? I am not saying there is no such thing as taste and neither will I attempt to define taste except to say "tasteless" is probably the opposite. "Judgement" seems more valuable as some have mentioned, maybe judgement and taste together identify something we want?
Still, I like better when we actually do try to name what we're talking about. We've probably all read the motorcycle guys take on quality and things like that right? I guess the theme some are going with is that quality can't be objectively quantified and to define it is futile but I've never liked that take. If we think LLM generated code is low quality then it certainly makes sense to start our attempts to define or at least describe what makes the quality needle move, because language is how we encode lots of information and information is what feeds LLMs.
People also like to rag on uncle Bobs takes, clean code, etc but I like how at least he tried. I've been a fan of his for awhile but I'm not big on objects so don't care about his pushes on that sort of thing. The best take that offered hints that I would call quality is "A Philosophy of Software Design" by John Ousterhout, there is plenty of stuff in this short little book that was very satisfying to me, like for instance where Bob wants short/small functions, Mr Ousterhout was much more flexible on this, like for instance maybe some functions can be bigger if the need arises, things like that.
If you have taste in writing code, you spend less time on bugs and can implment features more quickly. This is still true with LLMs if you learn how to direct them instead of having them direct you.
Someone else without taste can re-implement what you have done with AI. But it will take them more effort, more time, and more money.
But that's always been the value in writing code. Any MBA can end up producing software to accomplish something by hiring other people. But their company might not get off the ground because of the expense of not having technical competency (or engineers with taste) at their core.
When resources are scarce, you are forced to use them in a meaningful way. You learn to do only what matters most: what is right, minimal, essential. When you need to choose wisely when resources are scarce -- this is how you learn the taste.
Now that creation is cheap, the scarcity has moved to adoption. Things built without taste are less likely to be adopted, to survive. At least that is what I'd like to believe.
I find this post disheartening. To my eye, it seems to be the author's first blog entry with obvious LLM contributions.
Given the recent "moving to a farm" ... 7 days later ... "leaving the farm" posts, it's hard not to read this as someone clearly struggling with the burnout which is all too common in our industry, and resultingly giving in to the false promise of a machine that can just handle the thinking for you.
Taste is something you use to build the guardrails of AI over time and many paper cuts endured. Once you have a big enough set of these (call them rules) recorded, they become your development substrate. You narrow the paths the agents can take to arrive at a coherent system. If you don't do that, superficially it may look similar, but it's not and users will be able to tell.
"I want to be careful here, because “taste” is doing a lot of work and it is easy to hear it as decoration."
Claude, is that you?
I haven't seen a human write like this, ever, and I've read a lot of books.
The 'want to be careful here', the 'doing a lot of work'... common.
If it had "I'd like to gently push back here" next to it, I wouldn't have been surprised.
Very nice article. I agree with it, call it taste, call it judgment, call it quality, I want things done the way I see as satisfactory, and it requires me to make LLMs do things differently several times. Like exploring a tree with a heuristic they don't have and need to rely on us humans to give.
Great read. Ironic that he cited Pirsig because I think that reading ZATAOMM was the last time that I experienced this type of writing that made me pause and think deeply about what I just read.
> I do not miss the wall, and I will not pretend to.
The wall is the point, if you want to truly know the truth of something. I still believe there is no royal road. Taste isn’t the answer to why you would choose a bump allocator over a slab when designing a system. It is experience, wisdom, and the long hard road of trying yourself and failing. Taste comes after.
There is much joy in suffering when the result is knowing.
Too many people have tried to declare victory over software engineering prematurely and it shows.
« Software works the same way. »
No, I don't think so.
Modern software has been around for about 50 years, the Internet & OOP for about 35 years.
People have been cooking steaks since antiquity!
If your opinion or understanding of how a program works or how it should work is your main asset for keeping your job, then I think you're the one who's going to get cooked.
You’d have to be rather disillusioned to wake up in the post-COVID era, with technology that triples its algorithmic efficiency every year, and conclude that a dehumanizing industry will still need our expertise in the future.
The key words here are “dehumanizing industry”.
You can always bake your bread the old-fashioned way or cook your steak in « your own small restaurant », no one will stop you. But small restaurants, at least where I live, aren’t doing very well.
But toil is not gone. We have still have code owners, lawyers, physical hardware, thousands of configurations that we can't test them all.
And notably: we will still be asked to write software we can maintain in the future.
Sure, there are short-lived non-critical software projects where LLMs will rule.
But for many big complex projects, you can't land code without reviews. You can argue its because taste is already being enforced in the name of maintainability.
If we define taste as the intuitive act of saying "no, again," then I fully disagree with this whole article.
Every AI-frustrated (but LLM-written, sigh) blog post about the loss of taste and craft and hard work in development sounds like we've given up the interaction with the machine. Like human work is sitting back in your chair and shitting on stuff.
That's obviously not true! That's not how any of this works!
It's hard and weird to develop with the LLMs because they just do stuff. Lots of it is good, some of it is OK, some of it's horrible. Unpacking what it's done is hard and weird because software isn't just lovely UX, it's also data structures that scale and performance and privacy and enterprise controls and SOC 2 and onboarding and accessibility.
If you want to build real software, all that stuff has to get done. Today you're working on the feature, tomorrow you're making it scale. It's long-term and iterative and complex and hard to pack into a prompt or a markdown spec.
The work is the work, done at and with the computer, and it's way more than just "taste."
"De gustibus non est disputandum"
("In matters of taste, there can be no disputes")
One thing I'm excited for is the expanded ability to modify and create my own interfaces and software so I don't have to be subjected to the bad taste of others. An example of this is something like Chrome reading mode that removes all the stupid decoration from what I'm trying to read.
It reminds me of I think Nietzsche, isn't it a quote of him? "In the end all that's left to us is taste"?
I can't seem to find it so quickly, so maybe my memory is failing me.
He definitely had a lot to say about taste.
Taste seems something that should be more related to a craft but when doing engineering we don't talk about taste, we talk about theory, systems, practices, skill, intuition, experimentation, teaching and learning. So the debate on whether taste matters or not might come down to whether creating software can be called craft or engineering. In my opinion, maybe programming was more of a craft before LLMs but it looks like it is being squished between the engineering aspects of producing and maintaining software going ahead.
I think whether "taste" was ever important comes down to whether software is a commodity to be bought and sold, or something more strategic.
The company that needs a webapp because everyone else has one is never going to value "taste".
To adapt to this AI world, we need to either ship more lower-quality software, or start seeking out higher value problems. I don't see any other way around it.
I agree with the conclusion of this, I think, but the consequences are more interresting. One of the biggest problems with this flood of apps and products is curation. For the things that "deserve to exist" how does anyone find them among the sea of slop? I don't think there's a good solution, at least one that exists today. Look at Steam, Netflix, Youtube, Amazon or really anything else. How do you find what's good? For me, most of the time, I just return to what I know. But that doesn't seem ... healthy?
I disagree that consuming content in a field is not enough to build taste. I'd argue it's actually better than creating, simply due to scale. 1:1 sure, creating something will give you more "taste" than consuming something. But it's not 1:1. Creating takes so long. And creating mixes design and implementation. When you're implementing, most of the time, you aren't building taste.
As far as creating quality things goes, and "the wall", doing it well becomes a practical problem. Writing code for example forces you to think deeply about what your building, what it does, and how it works. You can do that without writing code, you're just no longer forced to. It's not an intrinsic problem. It's just a nice side effect of coding that you lose, and need to figure out how to compensate for.
> Everyone can make anything. Almost no one can tell you what is worth making. That was always the harder skill. It is now the only one left.
But people can and do tell you what is worth making. Taste is largely set by class and power. The “skill” is having the right context (wealth, education, network) and then pattern matching on it.
Is taste really the one thing LLMs can't take away or do we talk about taste because judging has become the cheapest move available? I think focusing too much on taste is how we lose focus on the two questions that actually matter: what do I actually care about, and why do I make things at all.
I tend to like images in a split second, when they try to achieve a certain aesthetic at least. I can later inspect them more carefully and understand why they're good.
Reading text, it can either grab me or it doesn't. I've asked about that to an AI and there's some reason for that. Some text is more compatible with me, due to the rhythm or combination of chars/words used, how they actually look/feel, it's not just about the content. So I'm assuming some writers are going to be more compatible with me while others will make me dismiss them quickly or just skim.
I like to think I have the correct objective taste in music, but everybody thinks that about themselves.
People will always want something to gatekeep their chosen field. Code went away, so now people need to be able to say if the machine wrote good code. In a year or two people won't be able to use that because the code will be measurably better than something a typical dev would write, and in five years it'll be better than anything anyone could write.
Then we'll need to accept that coding is better when it's done by computers, and we'll start telling everyone it's actually the prompts and context that steered the machine that really matter.
This brilliantly articulates what is required for best practices in the AI era. I am putting this article on my required reading list for Software Engineering. On behalf of my students, thank you for writing this!
> I do write like this sometimes. The short sentences, the reversals, the one-word lines—all of it. It’s just the way it is. A LLM writes that way too, because it was trained on the same essays I grew up reading
I'm extremely irritated that the post is being accused of being AI slop. I have a visceral reaction to AI generated writing, and this post rang 100% human to me.
Makes me actually quite angry that one can no longer use short, pithy sentences without being accused of using AI. If AI causes everyone to reject actual decent writing because someone can't distinguish between it and that from a human, we have other problems on our hand.
The content of the post resonated with me 100% as well.
On a side note, since the author has included a detailed explanation that the text is not AI generated and this is her default writing style, I just find it very difficult to read this style of extra punchy writing. I don't know if it is just me or anyone else as well.
But this is basically reinventing the term "Aesthetics" which is a field that had several centuries of philosophy research about it. Not all of them will agree AI cannot have it.
Actually I think it is possible to reconcile Kant's third critique with the statement "LLM is possible to make aesthetic judgement" (with certain generous application of principle of charity, naturally).
> The market timed you both with the same stopwatch and it did not see the difference. It cannot see the difference. Taste does not show up in the diff.
Doesn't that mean guaranteed failure in the market if you use taste as your guiding star? Therefore, there's really no way to enter the market anymore.
I don't know how you then ship software or monetise it...? The market will just clone it.
So... Don't share stuff publicly? Only share with closed groups? Where does this leave small-time software development...?
In software development, good taste translates to readability, succinctness and maintainability of the code. It also translates to intuitiveness, reliability, security, performance, scalability and utility of the software.
All this doesn't necessarily translate to sales though. Because, to evaluate those things requires attention, trust, time and effort.
This is a significant barrier because a lot of software will appear to meet all of these superficially.
After 1 day of usage, a piece of software may appear to be intuitive, reliable, secure, performant and useful... But then after some time (sometimes a whole week or longer) you run into a critical scenario and discover that it cannot be solved with that software... Or performance drops off sharply after you created the 1000th record in the software... Or a hacker takes months to find that one endpoint which allows full remote code execution.
Even in an optimistic scenario, 1 week is a long time to evaluate a piece of software. I've encountered software which took 6 months and large teams of people to realize that it wasn't suitable. That's how long it took to hit the critical limits. It's a very long evaluation loop.
So you cannot judge new software efficiently by just looking at it. Even industry consensus is problematic if the software is very new and complex... Plenty of trends have fallen off a cliff in the past. You need to understand who is behind the software... And even that's not so easy; social proof can be misleading when it comes to deep technical ideas. The people who are good at social networking aren't necessarily good with tech.
I think AI slop code is going to be a much bigger problem than people anticipate. And the irony of it is that the solutions already exist... What doesn't exist is the mechanism to identify those solutions. But even the mindset needs to be corrected first.
It didn't require great character at all
our refusal disagreement and resistance
we had a shred of necessary courage
but fundamentally it was a matter of taste
Yes taste
in which there are fibers of soul the cartilage of conscience
Who knows if we had been better and more attractively tempted
sent rose-skinned women thin as a wafer
or fantastic creatures from the paintings of Hieronymus Bosch
but what kind of hell was there at this time
a wet pit the murderers' alley the barrack
called a palace of justice
a home-brewed Mephisto in a Lenin jacket
sent Aurora's grandchildren out into the field
boys with potato faces
very ugly girls with red hands
Verily their rhetoric was made of cheap sacking
(Marcus Tullius kept turning in his grave)
chains of tautologies a couple of concepts like flails
the dialectics of slaughterers no distinctions in reasoning
syntax deprived of the beauty of the subjunctive
So aesthetics can be helpful in life
one should not neglect the study of beauty
Before we declare our consent we must carefully examine
the shape of the architecture the rhythm of the drums and pipes
official colours the despicable ritual of funerals
Our eyes and ears refused obedience
the princes of our senses proudly chose exile
It did not require great character at all
we had a shred of necessary courage
but fundamentally it was a matter of taste
Yes taste
that commands us to get out to make a wry face draw out a sneer
even if for this the precious capital of the body the head
must fall
I've been circling this insight for a bit previously, leading me to a convo with AI about how we are now approaching more of a spec - no, prompt - no, context - no, [insert newest paradigm] driven design approach
The point being more that the development does weigh more heavily on having fun ideas to ask AI to build
I think there's probably also a place for people to make guides for AI-assisted programming or even personal essays on their views of "taste" or what they want code to look like?
I see pop up frequently objections to vibe coding (different from mere AI assistance in coding - vibe coding being creating things without looking at the code at all) that it doesn't account for architectural descisions, cybersecurity decisions, edge cases, and other such things. So to me it looks like a next step might be in identifying these things "vibe coding" doesn't do well, that people can learn more manually, so they can do at least AI assisted coding well.
I really don't get why people here are so obsessed with "taste". I also find the premise of the article to be completely wrong. Taste, not only in software, is only the only thing that's left for humans, if you ignore the complex social environments in which our technological systems are embedded. AI puts the emphasis in the socio-technological system on the social. If you were never social to begin with, or if you deliberately ignore the social embeddedness of our systems b/c you are terminally valley pilled, you might come under the impression though that indeed, taste is the only thing that's left.
> This post reads off as AI slop. You said it, I see it.
I completely disagree. I think this post has emdashes in the first paragraphs and that most of the people saying that it was generated didn't go any further than that.
Where is the repetition? The sense of deja vu? Where is the artificial sweetener, the overuse of adjectives? Where is the lack of meaning?
Suppose there was a group of highly motivated individuals that saw this article and felt called out by it -- prompt engineers, if you will.
Because saying that "taste is all that's left" and that you can't develop taste without doing the work?
That's saying that there's something still out of their reach.
Something that can't be copied and reproduced ad nauseum.
This article gets at the heart of the issue and calls out -- though not explicitly -- these practitioners.
It does this in a calm and reasoned way, in a way that they cannot defend themselves from.
I'm not making the claim that the article is an attack on anyone -- my claim is that a class of people will see it that way and react to that like Gamers seeing a 7/10 review for the game they're hyped for.
I won't try to posit whether this post was AI-generated or not and I do not think it matters. That focus is a defense mechanism for handling an uncomfortable idea.
The idea is this: the commodification of intelligence requires the codification of domain expertise to function reliably, and proper governance to scale.
Taste, in my opinion, is a fill-in for absent function and lack of utility. Systems function; they don't require taste. Taste is why PG put Sam Altman where he did — not a system of merit, not evidence. Taste is why DJT can do no wrong in the eyes of his base. It is unprovable and has no true measure, and that is exactly why it gets leaned on so heavily: it lets you claim superiority no matter what slop your system produces.
You say the box moved taste up a level. Out of the making and into decision layer. That relocation is a smell. Your unfalsifiable thing conveniently gets moved to the one place it still can't be checked, at precisely the moment the making became checkable. Just like: Iran/Strait of Hormus, AI curing diseases and all of the other bold promises made by people who claim to be "taste makers" but in the end its all the same. Bullshit! Because thats what taste is a euphemism for in this context. Convincing others that you are correct in some way with no actual proof to back it up.
The vector here should not be "taste"; it's tolerances: written down, measured, enforced. The judgment gets encoded into a spec, and the spec is what survives the person who had it. That is codification; not "taste".
If taste that's all that's left you've kinda really got nothing. Taste is an anti-moat, it's sort of a measure of how well you, as a literal receptor, agree with a pile of people out there on something. Like sure its something you can sort of develop but it costs you nothing to use, and if you're not creating something off-manifold you're just smearing stuff around that's already on it. Horrendous kind of take on "taste" that's taken hold of everyone. Kinda like if a music artist stopped making music to make spotify playlists. Sure they could maybe make _good_ playlists, but if they are sitting there crowing "taste is all thats left!" it's.. what is that? It just sounds like either a deluded giving up or just a very direct admission of giving up.
LLM-based code generation requires disciplined operation to lead to the best outcomes. Shops who only ever had taste as a proxy for discipline were missing this before, and they're missing it now. The fact that it's missing is why so many organizations are tolerating ever-growing mountains of slop. (Not new, by the way; just dramatically accelerated.)
Discipline can be linked to system design standards and your review process, and it can be translated to measurable outcomes at scale and over time, regardless of how agents are involved: unplanned work, time to recovery, employee retention, etc. When outcomes like these become your priority, you see that you can't rely on taste alone.
People in this thread are using the phrase "AI slop" as if everything made by AI is automatically slop. If you believe this, fair enough, but then you must support the abolition of AI.
If you believe that AI is capable of making good things, then it doesn't matter whether this article, or anything, was AI generated. Whatever it is -- software, prose, images -- if you have taste, you don't have to figure out if it was made by AI or not. All you have to do is apply taste and determine how good it is.
It's an interesting article. To me it reads like someone who is not confident in their writing, but is very interested in getting their idea across, iterated with an LLM to produce it.
It set off all my LLM writing triggers and I noped out after about 3 sentences, because I'm designing/coding with AI all day and I'm 100% sick of its writing. I don't really care if it was written by an LLM or a person who writes in LLM style.
HOWEVER, the surprising fix for me was to ask a number of LLMs to summarize it. I omitted the "is this AI slop?" parts when I pasted it in. Like I said, I don't care about who wrote it, but I hated the writing style.
The AI summaries were quite readable.
They converged at a high level on the summary, but with a good bit of variance in nuance. I had a good follow up discussion with one of the LLMs, so I'm glad I tried it.
I agree with the idea that the taste, for lack of a better word, was the only thing that ever mattered. What I disagree with is the idea that slop that is “good enough” is indistinguishable from something that is crafted is “indistinguishable by the market”. I think it is indistinguishable in the short term, but very distinguishable on a long enough time span. The same way that hacked together slop has always been “good enough” to close a ticket and even get someone a promotion, but later costs the company thousands of engineering hours to rewrite once it becomes completely unreliable, slow and impossible to modify. We’re just speed running that now. It’s the same battle engineers with “taste” have been fighting with sales teams and product managers and lazy teammates before LLMs, just in fast forward.
I found this post a bit too verbose and incoherent. It does read like AI slop to me, but the bigger issue is that its core argument itself seems flawed.
The author seems to be advocating for choosing to "carry the burden" of one's taste (or aesthetics, judgement, standards, etc) while bemoaning that the market may not sufficiently reward for doing so.
IMHO, the core flaw is in mixing these two things 1) the market rewarding what it chooses to reward; and 2) individuals choosing to pursue some external/internal rewards.
I broadly agree with the premise that Gen AI has lowered the cost of software development and, as a result, the quality of the idea and its execution have become more salient differentiators than just the ability to bring an idea to reality. I also agree that pursuing one's own sense of taste/judgment or "objective function" is often intrinsically rewarding and worth it. But ultimately, an individual has to decide how much they want to (or need to) play for the external market rewards vs their own intrinsic rewards/motivations.
But I don't think it's coherent to argue that the market "ought to" reward one for their oh-so-special but mysterious, amorphous, and unmeasurable way of doing things, or that pursuing one's own taste/judgement is somehow automatically an exercise in greatness.
I think the "taste" described here actually can be formulated as a measurable quantity, just one that can't be accurately measured in practice. According to James Shore's 2006 essay "Quality With a Name" [0], "good design" amounts to nothing more or less than minimising the expected engineering time needed to create a good enough product over the space of changes that are likely to be needed during its lifetime (I'm paraphrasing slightly). Realising this was a revelation to me. I would argue that OP's "taste" is Shore's "good design".
This conclusion may be unwelcome here, since it implies that AI can acquire "taste".
This was a great and resonant post. Don't worry about the green names.
I think there is almost a split between people who have read ZAOMM and people who have not.
I recently made a post in my circles about how "meaning" is what human quality is. If you look at classical art a lot of it could look objectively nonsense and have technical mistakes, but just because the machine can make a picture does not mean that it attributes any meaning or communicate any intention. The difference between human-made classical artwork and machine-made classical artwork is the meaning and symbology behind it
Unconscious intuition you can't explain in words, thus can't reason about, as the main guiding principle in the ever more cognitively demanding world, yeah, good luck.
Before God and as an honest man I tell you that your son is the greatest composer known to me either in person or by name; he has taste, and, furthermore, the most profound knowledge of composition.
Speaking of taste the author basically tells me not to read the article immediately as soon as I hit the page because the commenters on this site (HN) have made them feel so bad about it
I’m pretty sure it was not technology that killed the web it was the kind of toxic crap like that, that we’re seeing from communities and that’s nothing new
If I ran a dev team, I’d spend a day a week on deliberately hand coding. Even if the clankers write most of the code, IMO we need deliberate practice to develop taste. There’s a side of taste that’s Rick Rubin, how the customer experiences something. There’s also what a classically trained cellist hears in a more tactile, layered sense in how another cellist plays a piece.
I'm really not convinced that the word "taste", as used in these type of AI discussions, refers to any meaningful concept. The AI that wrote this blog post (look at footnotes 2 and 3!), like many others, is great at defining what it is not and how people psychologically relate to it, but I've never seen a clear definition of what it is.
My best guess is that it's a pointer not to any actual concept but to latent semantic space between concepts. Either a very novel or very Buddhist phenomenon, depending on how you look at it.
Honestly, a lot of people are denying this, but the latest models ARE at the point where their systems design generally surpasses mine because they have a better grasp of how the domain of any problem is typically solved.
The first is an XKCD 605 (Extrapolating). The author makes bold predictions for the future as though they're inevitable, but gives zero evidence, because there is none, we're still at T=0. IBM is still in business. Wake me up when it's been taken out by someone vibe-designing a mainframe in kicad. Or even, let's see a dropbox clone get off the ground. Should be easy, right? (It's an HN tradition, after all.)
The second mistake is mistaking a large multiplier for infinity. The cost of code has not "collapsed to almost nothing" as the author claims. Talk to your CTO/CFO if you want the real story. They're probably freaking out over AI spend. But the point is, yes, AI has made things cheaper, but they didn't become monotonically zero.
AI had also raised the bar. You can't ship "good enough" anymore, it has to be great. And great still takes time. Time in QA, time obsessing over every workflow, time talking with customers and planning features.
Taste hasn't even entered the picture, and the analysis is already wrong.
Ironically, this essay is an example of at least somewhat "questionable taste". It has provoked dialogue and thus is valuable, but I think it makes a number of cognitive errors in its argument.
"Taste" is just knowledge. Advanced knowledge. Knowledge that is not easily put into words.
This is another restatement that knowledge, in a human head, is the determining factor of production, over land, labor and capital. This a post-capitalist argument that's been made for over 30 years (Peter Drucker in his 1993 book is my favorite), and it's been at least somewhat true even back to the 70s, just increasingly so as AI takes off.
The author's sidebar of AI or even cloud computing "taking ownership of the means of production away from workers" is entirely misunderstanding how the modern economy operates and shows how badly we need a new Marx (and Keynes, Schumpeter, or Drucker) since his analysis doesn't seem to provide the same analytical value in late capitalism. Renting vs. owning the means of production is now entirely a function of knowledge of the entrepreneurs (and their team).
Capital is ubiquitous now and is desperate for return - it remains powerful but is not all-powerful. We've had years of near-low interest rates and even after COVID inflation, they're still historically low. It is far more accessible to entrepreneurs than it has ever been in history, between grants, loans, subsidies, private equity, angels, crowdfunding, donations, etc. It's practically a cliche that billions in capital can be wasted on an initiative because the people involved had no idea what they were doing. I don't think AI changes that.
We don't need to rent our tokens, we can run open models on our local rigs (which still requires capital). We don't need to rent our servers, we can buy, rack and stack them on our own. Which we choose is fentirely dependent on the knowledge we have. If we don't know how to run an AI rig, we must rent it. If we don't know how to run a server, we must rent it. etc.
There are several walls driven by taste
- Design
- Testing
- Production
- Security
- Scalability
In particular, the author seems confused on what "production" is. Production is still the wall. There was a joke on X the other day:
"I vibe coded my dream app! Try it out here: http://localhost:8080/coolapp"
Software running in production is not tokens. I'm sure properly prompted AI can deploy and run simple software in production but this requires a lot of knowledge to prompt to constrain it to run in a secure, scalable, adaptable way as complexity, scale, and capability grows.
Taste Is All That's Left
(notashelf.dev)663 points by tsak 6 August 2026 | 521 comments
Comments
“For taste governs every free — as opposed to rote — human response. Nothing is more decisive. There is taste in people, visual taste, taste in emotion — and there is taste in acts, taste in morality. Intelligence, as well, is really a kind of taste: taste in ideas. (One of the facts to be reckoned with is that taste tends to develop very unevenly. It's rare that the same person has good visual taste and good taste in people and taste in ideas.)
Taste has no system and no proofs. But there is something like a logic of taste: the consistent sensibility which underlies and gives rise to a certain taste. A sensibility is almost, but not quite, ineffable. Any sensibility which can be crammed into the mold of a system, or handled with the rough tools of proof, is no longer a sensibility at all. It has hardened into an idea..."
Previously:
Taste in the age of AI and LLMs - https://news.ycombinator.com/item?id=47677241
You Had No Taste Before AI - https://news.ycombinator.com/item?id=45288551
"That says something about my writing. It says nothing about who wrote it." - Two contradicting sentences, author's writing skills are part of the author as-a-whole
"The short sentences, the reversals, the one-word lines" and then "I will write the next one plainer.". - even more?
And... I wonder if this is like some agent wrote article, and, in the loop, added port-mortem, after visiting hacker-news comments and manipulate that this is human written. Or maybe author is just trolling us with AI-like prose...
One thing that I’m particularly frustrated with is the writing quality of LLMs. Like this is the thing that they should be able to do, but I would say almost everything they write has almost no signal.
Over a mid sized AI generated codebase that means I’m reading like 500 words to figure out what a module is even doing.
AI is not inaugurating an age of taste. It is shortening the half-life of taste and commoditizing software. What happens when software features, UX, and visual fidelity aren't competitive advantages anymore? If you ask me, it looks like business did in the 90s. We'll probably see the return of department store software in the next few years.
1: for the record https://en.wikipedia.org/wiki/Pull-to-refresh
I think the core still leads back to human agency and the ability to consider aspects that wouldn't fit into an LLMs limited context window or be able to be vectorized into a DB, including ultra-long-term consequences (especially those with great thinking abilities).
I still happen to think that AI/LLMs will never be able to "fully" replace humans, because the evolutionary process that led to our cognitive abilities and the way we train LLMs are vastly different thanks to different pressures, but maybe that's just me defending the last bastions of our collective humanity as a human; I don't know what else to root for.
But the advent of design systems and software like Sketch and Figma pretty much automated the job away. You didn’t need the raw Photoshop skills anymore - any CMU grad could be a top tier designer now.
I see the same thing basically happening here. Taste will matter for a while but eventually people are just choosing the most accepted/expected optimal choices and anything that falls outside the trend will be “not good” in the same way skeuomorphism in design is no longer good.
So we’re lucky we’re in a taste era - it’s a lot of fun. You (yes you!) can still change it, shape it, establish the status quo of tomorrow
Yes, previously the person who didn't understand shipped faster. It was a long standing problem. The people who crapped, got the bonus, and ran were much faster than the patient engineer of quality.
By the way, what were the essays that formulated Claude Style? if the author was indeed formed by them, I'd like to read them - or at least enough of them to understand.
And honestly: I've been tired of an age of engineers that are beat down, that only want simple, that finding the most blunt approach is the only way. It has felt tasteless. Anti-ambitious. It often is still very sensible and practical and what you should shoot for! But there are also people wandering around now trying much harder! And I am excited for those frontiers! I think it will challenge and shake the foundations that we've accepted as true, as There Is No Alternative, in exciting bold and fun ways.
I am however pretty unconvinced by the article. There's a lot that doesn't quite work for me, that isn't building the case I'd like, whose takes are off from my read.
> The output is good enough—that is the problem. Good enough is a solvent. It dissolves the reason to do better. For as long as making things was expensive, the expense did quiet work on our behalf. It rationed output.
Agreed about the rationing of output. I do think that the article though continues to show a magical thinking. That we have these things now and they autonomously do the thing. That the LLM's have solved it all.
The article itself goes on at length about how nebulous and abstract taste is:
> Taste is that. It is the compressed, wordless verdict you reach faster than you can justify. It is the “no, again” you say to yourself with total conviction and no available argument
But this contrasts so distinctly to me against "the output is good enough". Is it? That depends. That depends on your taste. The proximate first results come quick. But the technics underneath? Those themselves, in my view, rely enormously on engineering taste to support and advance. I think we see a very similar sort of magical delivery thinking, very clearly on display here:
> When the factories came, they could suddenly make everything—cheaply, uniformly, by the thousand.
As if there was some magical "good enough" transition where suddenly the aliens came and gave us this box that just does the thing. As if we discovered the right formulas and math and now: we had production. Again I think there's just an enormous amount of work and taste that is still actively required inside the factory box, that building the industrial processes is incredibly intense & difficult, even though we have reliable industry-line production and now robots doing the labor.
There's still so many gems, so much lovely material for thought. I love the provocations here, and I think there's a lot of great calibration.
> The friction was not an obstacle to developing taste. The friction was the curriculum.
Makes me think of yesterdays @apenwarr banger,
> Every slow prototype started out as a fast prototype, I think that’s how it goes https://bsky.app/profile/apenwarr.ca/post/3msemlo4rds2h
I don't fully agree with it — for one thing, I think you can develop taste in a field without actively participating, but you have to engage with it beyond passive consumption. Basically: I think it's possible to be a good movie critic without making movies yourself. Given the collapse of the friction/effort side of things, I think that is really worth unpacking.
I have to say I find the debates about the post's style and em dashes pretty disheartening. There have always been plenty of good jokes to be made about HN's culture, but wow. For a community that has been snarky about fools counting lines of code for 40 years to be out here counting em dashes is pretty damning.
How convenient it is that we’re all so awesome according to the one metric that can’t be measured
Still, I like better when we actually do try to name what we're talking about. We've probably all read the motorcycle guys take on quality and things like that right? I guess the theme some are going with is that quality can't be objectively quantified and to define it is futile but I've never liked that take. If we think LLM generated code is low quality then it certainly makes sense to start our attempts to define or at least describe what makes the quality needle move, because language is how we encode lots of information and information is what feeds LLMs.
People also like to rag on uncle Bobs takes, clean code, etc but I like how at least he tried. I've been a fan of his for awhile but I'm not big on objects so don't care about his pushes on that sort of thing. The best take that offered hints that I would call quality is "A Philosophy of Software Design" by John Ousterhout, there is plenty of stuff in this short little book that was very satisfying to me, like for instance where Bob wants short/small functions, Mr Ousterhout was much more flexible on this, like for instance maybe some functions can be bigger if the need arises, things like that.
If you have taste in writing code, you spend less time on bugs and can implment features more quickly. This is still true with LLMs if you learn how to direct them instead of having them direct you.
Someone else without taste can re-implement what you have done with AI. But it will take them more effort, more time, and more money.
But that's always been the value in writing code. Any MBA can end up producing software to accomplish something by hiring other people. But their company might not get off the ground because of the expense of not having technical competency (or engineers with taste) at their core.
Now that creation is cheap, the scarcity has moved to adoption. Things built without taste are less likely to be adopted, to survive. At least that is what I'd like to believe.
Given the recent "moving to a farm" ... 7 days later ... "leaving the farm" posts, it's hard not to read this as someone clearly struggling with the burnout which is all too common in our industry, and resultingly giving in to the false promise of a machine that can just handle the thinking for you.
Claude, is that you?
I haven't seen a human write like this, ever, and I've read a lot of books. The 'want to be careful here', the 'doing a lot of work'... common. If it had "I'd like to gently push back here" next to it, I wouldn't have been surprised.
There will be no mass job layoffs for antying but relatively automatable stuff.
The wall is the point, if you want to truly know the truth of something. I still believe there is no royal road. Taste isn’t the answer to why you would choose a bump allocator over a slab when designing a system. It is experience, wisdom, and the long hard road of trying yourself and failing. Taste comes after.
There is much joy in suffering when the result is knowing.
Too many people have tried to declare victory over software engineering prematurely and it shows.
If your opinion or understanding of how a program works or how it should work is your main asset for keeping your job, then I think you're the one who's going to get cooked. You’d have to be rather disillusioned to wake up in the post-COVID era, with technology that triples its algorithmic efficiency every year, and conclude that a dehumanizing industry will still need our expertise in the future.
The key words here are “dehumanizing industry”. You can always bake your bread the old-fashioned way or cook your steak in « your own small restaurant », no one will stop you. But small restaurants, at least where I live, aren’t doing very well.
But toil is not gone. We have still have code owners, lawyers, physical hardware, thousands of configurations that we can't test them all.
And notably: we will still be asked to write software we can maintain in the future.
Sure, there are short-lived non-critical software projects where LLMs will rule.
But for many big complex projects, you can't land code without reviews. You can argue its because taste is already being enforced in the name of maintainability.
Every AI-frustrated (but LLM-written, sigh) blog post about the loss of taste and craft and hard work in development sounds like we've given up the interaction with the machine. Like human work is sitting back in your chair and shitting on stuff.
That's obviously not true! That's not how any of this works!
It's hard and weird to develop with the LLMs because they just do stuff. Lots of it is good, some of it is OK, some of it's horrible. Unpacking what it's done is hard and weird because software isn't just lovely UX, it's also data structures that scale and performance and privacy and enterprise controls and SOC 2 and onboarding and accessibility.
If you want to build real software, all that stuff has to get done. Today you're working on the feature, tomorrow you're making it scale. It's long-term and iterative and complex and hard to pack into a prompt or a markdown spec.
The work is the work, done at and with the computer, and it's way more than just "taste."
One thing I'm excited for is the expanded ability to modify and create my own interfaces and software so I don't have to be subjected to the bad taste of others. An example of this is something like Chrome reading mode that removes all the stupid decoration from what I'm trying to read.
Prescient.
I really have no idea how future software engineers will know how to write software in a world wherein software is written for them wholesale.
The company that needs a webapp because everyone else has one is never going to value "taste".
To adapt to this AI world, we need to either ship more lower-quality software, or start seeking out higher value problems. I don't see any other way around it.
I disagree that consuming content in a field is not enough to build taste. I'd argue it's actually better than creating, simply due to scale. 1:1 sure, creating something will give you more "taste" than consuming something. But it's not 1:1. Creating takes so long. And creating mixes design and implementation. When you're implementing, most of the time, you aren't building taste.
As far as creating quality things goes, and "the wall", doing it well becomes a practical problem. Writing code for example forces you to think deeply about what your building, what it does, and how it works. You can do that without writing code, you're just no longer forced to. It's not an intrinsic problem. It's just a nice side effect of coding that you lose, and need to figure out how to compensate for.
But people can and do tell you what is worth making. Taste is largely set by class and power. The “skill” is having the right context (wealth, education, network) and then pattern matching on it.
Reading text, it can either grab me or it doesn't. I've asked about that to an AI and there's some reason for that. Some text is more compatible with me, due to the rhythm or combination of chars/words used, how they actually look/feel, it's not just about the content. So I'm assuming some writers are going to be more compatible with me while others will make me dismiss them quickly or just skim.
I like to think I have the correct objective taste in music, but everybody thinks that about themselves.
Then we'll need to accept that coding is better when it's done by computers, and we'll start telling everyone it's actually the prompts and context that steered the machine that really matter.
Just reading the title and I was already thinking about that book. It's a great book though it gets progressively more difficult
I'm extremely irritated that the post is being accused of being AI slop. I have a visceral reaction to AI generated writing, and this post rang 100% human to me.
Makes me actually quite angry that one can no longer use short, pithy sentences without being accused of using AI. If AI causes everyone to reject actual decent writing because someone can't distinguish between it and that from a human, we have other problems on our hand.
The content of the post resonated with me 100% as well.
- _why the lucky stiff
Actually I think it is possible to reconcile Kant's third critique with the statement "LLM is possible to make aesthetic judgement" (with certain generous application of principle of charity, naturally).
Doesn't that mean guaranteed failure in the market if you use taste as your guiding star? Therefore, there's really no way to enter the market anymore.
I don't know how you then ship software or monetise it...? The market will just clone it.
So... Don't share stuff publicly? Only share with closed groups? Where does this leave small-time software development...?
All this doesn't necessarily translate to sales though. Because, to evaluate those things requires attention, trust, time and effort.
This is a significant barrier because a lot of software will appear to meet all of these superficially.
After 1 day of usage, a piece of software may appear to be intuitive, reliable, secure, performant and useful... But then after some time (sometimes a whole week or longer) you run into a critical scenario and discover that it cannot be solved with that software... Or performance drops off sharply after you created the 1000th record in the software... Or a hacker takes months to find that one endpoint which allows full remote code execution.
Even in an optimistic scenario, 1 week is a long time to evaluate a piece of software. I've encountered software which took 6 months and large teams of people to realize that it wasn't suitable. That's how long it took to hit the critical limits. It's a very long evaluation loop.
So you cannot judge new software efficiently by just looking at it. Even industry consensus is problematic if the software is very new and complex... Plenty of trends have fallen off a cliff in the past. You need to understand who is behind the software... And even that's not so easy; social proof can be misleading when it comes to deep technical ideas. The people who are good at social networking aren't necessarily good with tech.
I think AI slop code is going to be a much bigger problem than people anticipate. And the irony of it is that the solutions already exist... What doesn't exist is the mechanism to identify those solutions. But even the mindset needs to be corrected first.
TASTE
For Professor Izydora Dambska
It didn't require great character at all our refusal disagreement and resistance we had a shred of necessary courage but fundamentally it was a matter of taste Yes taste in which there are fibers of soul the cartilage of conscience Who knows if we had been better and more attractively tempted sent rose-skinned women thin as a wafer or fantastic creatures from the paintings of Hieronymus Bosch but what kind of hell was there at this time a wet pit the murderers' alley the barrack called a palace of justice a home-brewed Mephisto in a Lenin jacket sent Aurora's grandchildren out into the field boys with potato faces very ugly girls with red hands
Verily their rhetoric was made of cheap sacking (Marcus Tullius kept turning in his grave) chains of tautologies a couple of concepts like flails the dialectics of slaughterers no distinctions in reasoning syntax deprived of the beauty of the subjunctive
So aesthetics can be helpful in life one should not neglect the study of beauty
Before we declare our consent we must carefully examine the shape of the architecture the rhythm of the drums and pipes official colours the despicable ritual of funerals
Our eyes and ears refused obedience the princes of our senses proudly chose exile
It did not require great character at all we had a shred of necessary courage but fundamentally it was a matter of taste Yes taste that commands us to get out to make a wry face draw out a sneer even if for this the precious capital of the body the head must fall
> I do not care about your taste.
Is the point of this essay that there's nothing left?
The point being more that the development does weigh more heavily on having fun ideas to ask AI to build
I think there's probably also a place for people to make guides for AI-assisted programming or even personal essays on their views of "taste" or what they want code to look like?
I see pop up frequently objections to vibe coding (different from mere AI assistance in coding - vibe coding being creating things without looking at the code at all) that it doesn't account for architectural descisions, cybersecurity decisions, edge cases, and other such things. So to me it looks like a next step might be in identifying these things "vibe coding" doesn't do well, that people can learn more manually, so they can do at least AI assisted coding well.
I completely disagree. I think this post has emdashes in the first paragraphs and that most of the people saying that it was generated didn't go any further than that.
Where is the repetition? The sense of deja vu? Where is the artificial sweetener, the overuse of adjectives? Where is the lack of meaning?
Suppose there was a group of highly motivated individuals that saw this article and felt called out by it -- prompt engineers, if you will. Because saying that "taste is all that's left" and that you can't develop taste without doing the work? That's saying that there's something still out of their reach. Something that can't be copied and reproduced ad nauseum.
This article gets at the heart of the issue and calls out -- though not explicitly -- these practitioners. It does this in a calm and reasoned way, in a way that they cannot defend themselves from.
I'm not making the claim that the article is an attack on anyone -- my claim is that a class of people will see it that way and react to that like Gamers seeing a 7/10 review for the game they're hyped for.
The idea is this: the commodification of intelligence requires the codification of domain expertise to function reliably, and proper governance to scale.
Taste, in my opinion, is a fill-in for absent function and lack of utility. Systems function; they don't require taste. Taste is why PG put Sam Altman where he did — not a system of merit, not evidence. Taste is why DJT can do no wrong in the eyes of his base. It is unprovable and has no true measure, and that is exactly why it gets leaned on so heavily: it lets you claim superiority no matter what slop your system produces.
You say the box moved taste up a level. Out of the making and into decision layer. That relocation is a smell. Your unfalsifiable thing conveniently gets moved to the one place it still can't be checked, at precisely the moment the making became checkable. Just like: Iran/Strait of Hormus, AI curing diseases and all of the other bold promises made by people who claim to be "taste makers" but in the end its all the same. Bullshit! Because thats what taste is a euphemism for in this context. Convincing others that you are correct in some way with no actual proof to back it up.
The vector here should not be "taste"; it's tolerances: written down, measured, enforced. The judgment gets encoded into a spec, and the spec is what survives the person who had it. That is codification; not "taste".
Discipline can be linked to system design standards and your review process, and it can be translated to measurable outcomes at scale and over time, regardless of how agents are involved: unplanned work, time to recovery, employee retention, etc. When outcomes like these become your priority, you see that you can't rely on taste alone.
If you believe that AI is capable of making good things, then it doesn't matter whether this article, or anything, was AI generated. Whatever it is -- software, prose, images -- if you have taste, you don't have to figure out if it was made by AI or not. All you have to do is apply taste and determine how good it is.
It set off all my LLM writing triggers and I noped out after about 3 sentences, because I'm designing/coding with AI all day and I'm 100% sick of its writing. I don't really care if it was written by an LLM or a person who writes in LLM style.
HOWEVER, the surprising fix for me was to ask a number of LLMs to summarize it. I omitted the "is this AI slop?" parts when I pasted it in. Like I said, I don't care about who wrote it, but I hated the writing style.
The AI summaries were quite readable.
They converged at a high level on the summary, but with a good bit of variance in nuance. I had a good follow up discussion with one of the LLMs, so I'm glad I tried it.
We still have a metric for good taste, and it is $. Solve a problem that people care about? They give you $. Solve problem no one care about? No $.
It is not perfect, but I challenge you to find a better metric that is as effective and simple.
Great artists ship.
The author seems to be advocating for choosing to "carry the burden" of one's taste (or aesthetics, judgement, standards, etc) while bemoaning that the market may not sufficiently reward for doing so.
IMHO, the core flaw is in mixing these two things 1) the market rewarding what it chooses to reward; and 2) individuals choosing to pursue some external/internal rewards.
I broadly agree with the premise that Gen AI has lowered the cost of software development and, as a result, the quality of the idea and its execution have become more salient differentiators than just the ability to bring an idea to reality. I also agree that pursuing one's own sense of taste/judgment or "objective function" is often intrinsically rewarding and worth it. But ultimately, an individual has to decide how much they want to (or need to) play for the external market rewards vs their own intrinsic rewards/motivations.
But I don't think it's coherent to argue that the market "ought to" reward one for their oh-so-special but mysterious, amorphous, and unmeasurable way of doing things, or that pursuing one's own taste/judgement is somehow automatically an exercise in greatness.
This conclusion may be unwelcome here, since it implies that AI can acquire "taste".
[0]: https://www.jamesshore.com/Articles/Quality-With-a-Name.html
That so many developers believe this, is surely a sign of the fields impending death.
I think there is almost a split between people who have read ZAOMM and people who have not.
I recently made a post in my circles about how "meaning" is what human quality is. If you look at classical art a lot of it could look objectively nonsense and have technical mistakes, but just because the machine can make a picture does not mean that it attributes any meaning or communicate any intention. The difference between human-made classical artwork and machine-made classical artwork is the meaning and symbology behind it
Haydn to Leopold, 1785
https://en.wikipedia.org/wiki/Haydn_and_Mozart#The_%22Haydn%...
I’m pretty sure it was not technology that killed the web it was the kind of toxic crap like that, that we’re seeing from communities and that’s nothing new
so emacs is also left :)
My best guess is that it's a pointer not to any actual concept but to latent semantic space between concepts. Either a very novel or very Buddhist phenomenon, depending on how you look at it.
The first is an XKCD 605 (Extrapolating). The author makes bold predictions for the future as though they're inevitable, but gives zero evidence, because there is none, we're still at T=0. IBM is still in business. Wake me up when it's been taken out by someone vibe-designing a mainframe in kicad. Or even, let's see a dropbox clone get off the ground. Should be easy, right? (It's an HN tradition, after all.)
The second mistake is mistaking a large multiplier for infinity. The cost of code has not "collapsed to almost nothing" as the author claims. Talk to your CTO/CFO if you want the real story. They're probably freaking out over AI spend. But the point is, yes, AI has made things cheaper, but they didn't become monotonically zero.
AI had also raised the bar. You can't ship "good enough" anymore, it has to be great. And great still takes time. Time in QA, time obsessing over every workflow, time talking with customers and planning features.
Taste hasn't even entered the picture, and the analysis is already wrong.
"Taste" is just knowledge. Advanced knowledge. Knowledge that is not easily put into words.
This is another restatement that knowledge, in a human head, is the determining factor of production, over land, labor and capital. This a post-capitalist argument that's been made for over 30 years (Peter Drucker in his 1993 book is my favorite), and it's been at least somewhat true even back to the 70s, just increasingly so as AI takes off.
The author's sidebar of AI or even cloud computing "taking ownership of the means of production away from workers" is entirely misunderstanding how the modern economy operates and shows how badly we need a new Marx (and Keynes, Schumpeter, or Drucker) since his analysis doesn't seem to provide the same analytical value in late capitalism. Renting vs. owning the means of production is now entirely a function of knowledge of the entrepreneurs (and their team).
Capital is ubiquitous now and is desperate for return - it remains powerful but is not all-powerful. We've had years of near-low interest rates and even after COVID inflation, they're still historically low. It is far more accessible to entrepreneurs than it has ever been in history, between grants, loans, subsidies, private equity, angels, crowdfunding, donations, etc. It's practically a cliche that billions in capital can be wasted on an initiative because the people involved had no idea what they were doing. I don't think AI changes that.
We don't need to rent our tokens, we can run open models on our local rigs (which still requires capital). We don't need to rent our servers, we can buy, rack and stack them on our own. Which we choose is fentirely dependent on the knowledge we have. If we don't know how to run an AI rig, we must rent it. If we don't know how to run a server, we must rent it. etc.
There are several walls driven by taste - Design - Testing - Production - Security - Scalability
In particular, the author seems confused on what "production" is. Production is still the wall. There was a joke on X the other day:
"I vibe coded my dream app! Try it out here: http://localhost:8080/coolapp"
Software running in production is not tokens. I'm sure properly prompted AI can deploy and run simple software in production but this requires a lot of knowledge to prompt to constrain it to run in a secure, scalable, adaptable way as complexity, scale, and capability grows.