Hacker News Reader

Comment by ulrikrasmussen | Hacker News Reader

An important feature of taste is also that it is highly variable between people, and because there is no system or proof of good taste, we cannot really say that one person's taste is objectively better than another's. This is what causes one person to build a large software system in a classical OOP language while another builds almost the same system in a typed functional language, each following their own internal logic of what is good taste. In the end we learn something from it, and maybe someone who worked on both projects refine their taste and starts preferring something that lies in between the two paradigms and comes up with a new interesting programming language.

One thing I fear is going to happen when more and more people use LLM's for coding is that they will default to the "average taste" of the LLM, and we will stop getting new interesting ideas about how to build things.

View on HN

> One thing I fear is going to happen when more and more people use LLM's for coding is that they will default to the "average taste" of the LLM, and we will stop getting new interesting ideas about how to build things.

I don't think it will have any impact at all on creativity in coding.

Creativity is rewarded in humans, just not by their employment where both training examples and monetary rewards can be found, rather the creativity is rewarded by challenges such as code golf, demoscene, and so on.

Go to a job interview and I have to know the shibboleths of Clean Code, SOLID, an arbitrary one of five frameworks, three different ways to handle asynchronous code, and four design patterns, and the magic word "agile". Invent something new at the workplace? If you're in a small team, or high enough in the hierarchy, fine. Big team? People will reject it with reference to those same shibboleths.

I wrestle with this also. A colleague pointed out that LLMs reduce language choice to an implementation detail… don’t have to know assembly to write python, and now you don’t have to know python to write a functional app. So why stick with python when rust might be better? Leave that to the LLM and let your engineering effort go toward designing the app instead of operating a programming language.

In which case yeah, projects are likely to converge on whatever pattern LLMs broadly fall into. But maybe that part doesn’t really matter. Not every project needs the engineer to care about that. If I’m not writing hardware level code, I don’t care what instructions my python compiles to, either.

Meanwhile nothing prevents someone from caring about these details if wanted or needed. Someone who knows and cares might absolutely do a better job than an LLM. Those people won’t necessarily go away.

But there -will- be a whole new category of software contributors who have ideas they don’t know how to express in a software language. Those ideas might well be very tasteful. So it could be argued that software development as a field will gain access to a whole new range of tastes.

I don’t know the answers but it all seems plausible

To be a bit more precise, SE is the practice to organize large codebase in a meaningful and human friendly way. AI will attract more hackers doing fun projects without careing about implementation details under a much higher level of abstraction, but they will not introduce more variation in SE.

It is a possibility. But I also think we're underestimating the human need to create cool, unique things. Like yes most of the projects in this situation will have "average taste" but most of the projects out there today already do. Very few people have true taste and I don't think we're just going to see them disappear or be satisfied with LLMs feeding them "average taste." They will either likely stop using LLMs if that is the case or find ways around it to make it do things the way they want to, hence preserving their taste.

But I think it is a bit of an overblown fear that LLMs will just kill creativity and innovation in coding. Hackers will be hackin

I also don't think it will kill creativity in the space, although my phrasing might have suggested that, just that it might diminish it and make implementation styles more conformative.

On the other hand, we might trade that for creativity in other more high level designs where things like coding style is an unimportant detail for the experiment. For example, I have started playing around with microcontrollers because I can have the LLM generate the software while I focus on learning how to solder and think about how to modularize my hobby automation system. That's creativity and taste at a higher level which I didn't have access to before because I didn't want to invest my precious free time in learning the software ecosystem.

I feel two ways about it. I've written some algorithms and architectures for motion control that were good, after many years of chewing on the problem through iterations. I don't think I'll replicate that effort or those results, now that I use LLM's so much.

On the other hand, I did that in pursuit of a goal, and LLM's take me further, faster, toward that goal.

>But I think it is a bit of an overblown fear that LLMs will just kill creativity and innovation in coding. Hackers will be hackin

Hacking and a taste for hacking is a muscle you develop. LLMs will make most people not train it at all. Same way a huge number of people are obese and out of shape. For LLMs to be bad news in this regard, it doesn't matter if some people will still find a way to do something creative. As long as far more people (than without LLMs) don't, that's already bad news.

It is exactly like that.

Yes, there are exercise nuts who are exercising like crazy, but the average population is getting weak and obese.

The same with mental muscles.

How many people can do mental calculations in their head nowadays?

What about remembering several phones numbers?

What about simply navigating with map but without GPS? (Or, heaven forbid, just by common sense and observation?)

Properly thinking about hard problems is exhausting. Majority will be glad to offload the hard thinking to an AI 'assistant'. The same way we offloaded computing to calculators, and memory and navigation to phones.

What remains is the easily manipulated lizard brain...

Heck, how many people in this room remember their own phone number?

Understandable but why would you assume people will stop getting a taste for hacking just because LLMs exist? Why will they stop just because of that? Technology will make them lazy just because they can prompt an AI now? I think the human drive for excellence and tinkering cannot just be vanished just because LLMs exist.

And I don't under the fat and out of shape part of the comment. Being in shape and being good at coding are not equal things in many aspects:

1. Most people learn code for employment. Most people get fit for personal reasons.

2. It takes considerable physical effort to get in shape. Getting good at code requires considerable mental effort but it is a very different form of effort.

You can claim LLMs will make people dumber and lazy. But that doesn't mean working out and being fit = being really good at computers in real life.

Lastly, you seem to have a cynical view of the world and believe in a merit-based structures but most people don't generally become super fit or hackers. Its always a small percentage of the population in general and most likely will always be. It is debatable if the availability of an LLM can reduce them to 0.

In my experience, using LLM's for generating code has definitely dulled my capacity to think about function hierarchies, composition and having a "picture" in my head of what will go where in the repository.

Raw "leetcode" algorithm intuition and database modelling/SQL are still largely intact for me because I never trust LLMs for those and want to actually understand what I'm building. But it's scary to think how easy it is to have your skills eroded and how quickly you become dependent on these tools.

From your comment, it seems like the only skill that has eroded in you apparently is the fact that you don't understand the structure of a repo that was written by someone else? Was that a skill you had pre LLM that you could just touch a repo you didn't work on and magically knew where everything went?

Furthermore, if you can see it so transparently and have put a stop gap, why do you think the rest of the world or programmers will not? Why will they fall into the temptation that you can resist? Especially if we are talking about that specific subset of people that are already really good at coding and understand the pitfalls pretty well.

Maybe you can make the argument that someone who purely vibecodes never gets the skills of raw algorithm intuition/specific knowledge but in those cases, no skills get eroded because there were none to begin with.

>Understandable but why would you assume people will stop getting a taste for hacking just because LLMs exist?

It's not a prediction, it's already the case. Like students coming into computer science degrees and letting the LLM do all the work, learning nothing. Interns phoning it in too.

>Technology will make them lazy just because they can prompt an AI now?

Yes, that's how it works. The path of least resistance is what's followed. Same way we walk less because we can just get in a car, or do or order things online now.

>I think the human drive for excellence and tinkering cannot just be vanished just because LLMs exist.

There's no trancedent "human drive for excellence and tinkering". There's either an actual force shaping or there's a path which is easier and most will turn into slobs and take.

Same way much much much fewer people tinker with their electric devices or cars than they did in the 1920s to 1970s. They just throw them out and buy a new one, even when the device is simple and repairable. And countless other examples where "excellence" or even "good" is traded for covenience.

Have you ever used enterprise software?

Seriously. Whatever slop is going to be an improvement for 99% of enterprise products. 99.9% of internal tools at large companies.

something something SAP something

>because there is no system or proof of good taste, we cannot really say that one person's taste is objectively better than another's

We can't say it in times of a general culture of confusion, like now. In times when there was an established cultural and artistic hierarchy, they could.

Was it objectively correct or was it just the king's taste?

literally: sumptuary laws restricting what you can wear, eat, or own

https://en.wikipedia.org/wiki/Sumptuary_law

not kings per se, maybe sultan, or the royal court, roman senate, etc.

Casey Muratori says they're both bad language choices as long as the software is slow.

https://www.youtube.com/watch?v=1h0CHrQE794

there's a tension between innate preferences we’re born with, versus cultivated tastes that require effort

The problem may not be that LLM taste is bad, but that it is incredibly easy to converge on

Reply on news.ycombinator.com