The Disappearance of the Public Bench

I don't believe it's a deliberate decision by (most) policymakers. I think it's a structural failure across several axes, including failure to make enough housing for it to be affordable, and attacks on every front by people who treat all social programs and public assistance as evil. Most places have one or the other problem, if not both.

I think the reason most people do not sympathize with this argument is that most HN readers are programmers, and many of them are still in a relatively secure class position.

Traditionally, programming has had a high barrier to entry, but it has also been a profession where compensation has remained relatively strong. As societies become harsher under pressure from high housing costs and economic displacement, they tend to become more aggressive and violent. But many people do not sympathize with this issue because they are not personally in that situation. They mostly experience the visible disorder: aesthetic damage, drug use, and the social harms produced by deeper structural failures.

But if we compare this to HN debates about LLMs, an irony appears. In labor-market terms, LLMs are similar to hostile design.

LLMs are not installing benches for programmers. They are closer to removing the benches.

In the past, there were lower-level tasks where junior developers, non-traditional developers, non-native English speakers, and small open-source contributors could remain inside the profession. CRUD work, documentation fixes, test writing, small bug fixes, simple UI, repetitive glue code — these were not glamorous tasks, and they were often inefficient. But they functioned like public benches inside the profession. They gave people a place to sit long enough to learn.

LLMs attack exactly that layer.

From a company’s point of view, this is rational. Code that might take a junior developer several days can now be drafted by a model in minutes. Documentation, tests, boilerplate, simple screens, and repetitive API wiring no longer seem worth preserving as training grounds for humans.

As a result, the market may look more efficient. But that efficiency resembles the history of removing benches. It is not only the “problematic” people who disappear. Elderly people, children, travelers, disabled people, and ordinary people who simply needed a place to sit are pushed out as well.

Software has a similar problem. If we remove low-level work, low-quality work may appear to decrease. But at the same time, we also remove the space where beginners can fail, receive correction, observe others, and slowly acquire the instincts of the profession.

So LLMs are not merely productivity tools. They can also function as a force that removes public seating inside the software profession.

That is why I find it difficult to reconcile the logic of people who argue that public benches should be removed, while also arguing that LLMs should not be accepted.

They are already sitting inside the profession. They already have experience, English, networks, code review experience, and existing project history. For them, LLMs look like a faster tool. But for people trying to enter from the edge of the profession, LLMs are not just a tool. They are a change in the structure of entry itself.

The lower seats where people could once sit and learn are disappearing. Newcomers are expected to start from a higher level of abstraction and with stronger verification skills from the beginning.

In cities, the logic for removing benches is usually expressed in the language of order, safety, aesthetics, and maintenance cost. In software, the logic for adopting LLMs is expressed in the language of productivity, efficiency, cost reduction, and quality control.

But behind that language, what disappears is the buffer zone through which a community receives people.

A city without benches may look cleaner, but it does not become more public. Likewise, a software market without entry-level work may look more productive, but it is hard to say that it has become a healthier ecosystem.

When I read HN, I often see this kind of irony. And perhaps we all live inside such ironies. That may also be part of what makes communities interesting.

People do not seem to have a consistent attitude toward publicness itself.

Instead, they show completely different moral intuitions depending on where they are positioned within that public space.

I always find that interesting to watch.

> US HUD says it's ~16% of homeless people, other sources give different numbers, but it's certainly not a majority

"Homeless people" is a broad category that includes people temporarily living in vehicles, bouncing between family members, or sleeping on a friend's couch. It also includes people who are about to lose their home, young people living alone.

But when everyday people use the term, they usually mean, specifically, visible homeless people - i.e. people who are homeless long-term, sleeping rough on the streets or in parks, etc.

The two groups are pretty different to each other. I would be very surprised if the rate of drug addiction in the second group was the same as the rate of drug addiction in the first group

[flagged]

> I would be very surprised if the rate of drug addiction in the second group was the same as the rate of drug addiction in the first group

But that's a far far weaker claim than the one above.

If the rate is 90% or higher in the second group, then we get close to the claim being true. (Though still a subset rather than the circles being the same; lots of people with drug problems have homes.)