I sympathize with this take but I think it’s conflating programming as a hobby vs code needed to run a business.
I always believed that programming was solving problems and building things. With AI, you can still solve problems - the better your questions/ prompts the faster you get to your answer. To ask those questions you need to probably have that meditative thinking to grasp the crux of what you’re trying to solve.
On the business side, paying a programmer for their meditative time would be the first line item to cut, when you can bark orders at an idiot savant instead.
When a business pays me to develop software, that's a lot more than coding. Sometimes it takes weeks to just translate the business needs into a plan, budget, and timeline (not that I ever meet it). Often that includes documentation, training, infra, and a maintenance contract (hopefully with retainer).
The coding really is such a small part of that process.
Absolutely. And honestly, I don't think AI can really replace that. It can execute a plan, but it can't really figure out the best plan from a sea of unknowns strewn within people's minds. There's a certain meditative aspect to this stream of work still, but I would hesitate to call it "programming".
Did they validate the result? Is it correct / accurate? If it is, then it passes the “good enough” test. For non-tech folks who just vibe code a fix, “technical debt” is not in their vocabulary. And if it works well, I don’t see the issue.
There was research showing how AI performance degrades as you go from green field into support. My guess ERP in question will break after some time as agent will add more bugs than new features. Plus recovering data will be extremely difficult.
The other day, someone tried to convince me to play Russian roulette. I was skeptical as it sounded dangerous. However, I validated the result by pointing the gun at my forehead and pulling the trigger several times in a row without anything bad happening. Now I'm been persuaded that it's good enough as a form of entertainment. Someone else told me it's still dangerous, but it works well so I don't see the issue.
Claude Code is definitely a harness. The term harness is pretty loosely defined, but it effectively means the software that wraps an LLM, asks the LLM what tools it should execute, then executes those tools.
The statement “These are child safety laws” premises that the laws would save children - so the way to refute is to ask for proof. Outside of vaccines I doubt there is anything concrete that will directly benefit children. There is no argument that holds up to scrutiny. They also don’t have any good way to measure safety, though Goodhart’s Law would have an impact.
I mean, I want to show an example where it is only a problem when someone else is not as convinced as yourself, one way or the other.
>so the way to refute is to ask for proof..
No use. The same proof will not be convincing for everyone. Let us take vaccine. The proof for a Vaccine for X saving Y lives is often the following.
"Before the vaccine there was Y deaths per year from X. But after vaccines there is zero deaths from X".
The implication is that vaccines saved X lives. This could convince most of the people. But someone can ask. Are those kids who are not dying of X (thanks to the vaccine), are dying of something else?
Now the picture is not so clear.
See, same data, different results depending on how deep you want to probe,
> Are those kids who are not dying of X (thanks to the vaccine), are dying of something else? Now the picture is not so clear.
Why is it not so clear? If they are not dying of X then the vaccine did the job and technically saved the kid’s life from dying of X. Notice that it doesn’t mean that they can’t die of other things.
Because just imagine that may be the underlying cause of death is malnutrition, or poor immune system. In this case the body won't be able to fight of any kind of infection. When you vaccinate such a person for X, that person is able to resist X. But since the immune system health is still weak or the person is still malnutrition, they could still succumb to an infection of Y.
But that is one less death from X, and one death less that the vaccine can claim credit to, but in the larger picture, the vaccine did not save a life.
So the point being 1000 deaths from X to 0 deaths from X does not necessary mean X lives saved.
> So the point being 1000 deaths from X to 0 deaths from X does not necessary mean X lives saved.
Yes, this statement is a non-sequitur. A vaccine preventing death from X does not prevent a death from other causes. That doesn’t mean X isn’t effective.
That’s fair pushback. I concluded that this was the underlying assumption because I couldn’t surmise anything else because you had said this:
> The implication is that vaccines saved X lives. This could convince most of the people. But someone can ask. Are those kids who are not dying of X (thanks to the vaccine), are dying of something else?
To me it assumes mutual exclusivity between not dying of X and dying of something else, but perhaps my bias is showing.
It did not assume mutual exclusivity. But if the underlying cause of succumbing to a disease is poor general health or a poor immune system, then preventing a single disease might not really save the person.
But it still shows up as a life saved by the vaccine (because it prevented a specific disease incidence). That is misleading.
This can be made very clear by taking a small example. Imagine there is an island with a population of 100. Among these 50 are very weak. And every year, these 50 die due to various illness. And among these 50 deaths, 30 are caused by X.
So in order to save these 30 deaths, the government started vaccinating against X. The next year there is zero deaths due to X. But there were still 50 deaths in total that year too. Only that the 30 lives that used to be claimed by X, is now claimed by A, B and C (they are so weak that they will succumb to any infection).
This is not an unrealistic scenario. If you have a weak immune system, any minor infection could end up fatal. The only unrealistic thing in this story is the occurance of constant 50 people who have weak immune system every year. But that does not void the point.
So while the vaccine prevented disease incidence, it did not prevent any deaths.
Sometimes I think it’s time to give shareholder supremacy a rest. If they don’t like what they own, they can sell it. But this is probably also why I’m not on a board of directors.
Interestingly enough, Apple at one point had that attitude even under Cook.
> “When we work on making our devices accessible by the blind, I don’t consider the bloody ROI,” Cook said. “If you want me to do things only for ROI reasons, you should get out of this stock.”
That was back in 2014.
But I do agree, it's time to normalize telling shareholders where they can shove it. If they don't like it, they can sell.
Ironically (and perversely) it's much easier to do this with exotic cap tables that keep voting rights and ownership effectively split. You needed a very valuable company to make this work and the most obvious example (Meta) is not exactly a beacon of corporate governance.
I don’t see how this is the case? The CEO can absolutely tell the board to shove it. The board can’t micromanage the CEO - they don’t have the levers to do that. So their options are few, and pretty much all nuclear along the lines of firing the CEO. Is the Apple board going to fire Tim Cook for making iPhones accessible? If they did, they’d find themselves fired by the shareholders.
I always believed that programming was solving problems and building things. With AI, you can still solve problems - the better your questions/ prompts the faster you get to your answer. To ask those questions you need to probably have that meditative thinking to grasp the crux of what you’re trying to solve.
On the business side, paying a programmer for their meditative time would be the first line item to cut, when you can bark orders at an idiot savant instead.
reply