Coding agents are not omniscient gods. Cognitive complexity of code is still a thing for LLMs. And every line of code that the LLM writes is still a line that needs maintenance, and has bugs.
Software engineering rules still apply. Coding agents move the needle somewhat on "what can be maintained by us vs. what someone else can maintain" but not _that_ far. And the same "Don't maintain it if you don't have to" approach still applies. More code to maintain means bigger context required, which means the agent burns more tokens to do a task. More context also means the agent makes more mistakes and adds more bugs.
Abstractions are still a good idea. Handing off maintenance through abstractions is still a good idea. Good software design is still a good idea.
It's ironic you talk about communism, because fundamentally this is a problem of "owning the means of production", i.e. who controls the AI and the robots.
If it's you and me, then hey, we can all live happily ever after pursuing hobbies and art, provided for by our mechanical friends, within the limits of resource availability.
If it's an AI trillionaire, things aren't so rosy. If they don't need workers, why would they even bother allocating outputs from the robots to keep us alive?
I don't doubt at all that some group or another will enjoy the fruits of artificial labour. The question is how we are going to get there, and whether that group is most people who are live today and their descendants, or just the families and friends of trillionaires.
My hope is that all wealth becomes concentrated in the hand of one person, thereby becoming worthless.
I guess one could make the argument that with enough control, a trillionaire could amass a private robot army. But I think regular humans still win (after an epic battle of course).
You're not looney at all. Frontier models do dumb things all the time, especially on mature codebases. Just yesterday Opus 5.5 butchered the OOP model in a codebase I work on - it duplicated a load of classes that should have just been subclasses. A junior checked it in very satisfied that it was perfect. The LLM review passed it, the tests were fine, and it implemented the feature successfully. It's just the code design had poor taste and poor long-term maintainability.
I keep seeing this kind of thing over and over, and honestly it's not got _that_ much better since the big breakthroughs about a year ago.
For sure I happily vibecode stuff without worrying about it when it's a greenfield project, and if the LLM has written it entirely from scratch then usually it's well structured and sane. But making changes in messy, mostly human-written mature codebases is still a minefield.
If it's just a toy, sure. But even frontier models reach a tipping point of complexity fairly early where they start introducing bugs quicker than they fix them.
A (good) framework abstracts away the complexity so you hit that tipping point much later in terms of codebase size, and reduces your token spend to boot.
I get that, seeing what a model can do with a Django codebase. Out of the major frontend frameworks, Svelte(kit) is actually the only one I would consider now as it seems to produce smaller packages.
People still buy into "research" from Gartner and Forrester despite it being an open secret for decades that they are pay-to-play.
These kinds of "research" companies are there to validate people's preconceived ideas and purchasing decisions rather than being genuinely unbiased. And they are very useful for that, both for consumers and for marketers.
it might be ancient, but it's also very simple. it's relatively easy to import a single header and write a working plug-in, which makes it attractive for enthusiasts.
"As effective as placebo" does not mean worthless. As you point out, placebo antidepressants are fairly effective.
It's a terrible bind. Patients want a pill to fix things, but if they know it's just a sugar pill, it doesn't work. It has to have active ingredients that might work. That's why there's little desire to change the status quo on antidepressants much. Anyone who reads the medical literature knows they're statistically underwhelming, but the experienced reality is that they help people a lot.
Talking therapies, CBT, exercise are all good alternatives but they take time and effort that a depressed person might not be able to manage. An antidepressant prescription they can get in 15 minutes.
> As you point out, placebo antidepressants are fairly effective.
This is a misunderstanding of concepts like regression to the mean, and also the active placebo elements involved.
> but the experienced reality is that they help people a lot
And the evidence is that the reality people think they are experiencing is wrong, i.e, they are improving and factually experiencing improvement, but misattributing the cause to the drug.
> "As effective as placebo" does not mean worthless
In one sense of worthless, perhaps, but since drugs have larger costs relative to placebo, well, we can argue they are worse than worseless in another.
Better instead to talk about cost-benefit tradeoffs, number-needed-to-treat vs number-needed-to-harm and etc though, and try to get better at prescribing more carefully to those they clearly benefit.
It’s unethical to prescribe a patient a placebo in a way that suggests that what they are receiving is scientifically proven.
In a clinical trial setting, you can prescribe a placebo, because the patient is fully aware and clearly consenting to the fact that they may receive a placebo. Lying to a patient, in a clinical setting, from a position of authority, is a completely different matter. The informed consent would be completely absent, the patient’s ability to make informed choices about their own healthcare would be undermined and withheld, and the provider would be deriving financial benefit from the patient and / or through insurance claims for what amounts to a scam.
The patient, who would be seeking a treatment from a trusted expert, would believe that they are receiving proven treatments in exchange for their time, patience, money, reduced quality of life due to side effects, and opportunity costs in terms of not going to a different provider or trying something else, but in reality their provider would be misleading them. Some patients would even die as a result of taking a particular placebo and depending on it—either because of side effects or due to the lack of effectiveness—when they tragically would have been better off trying a different approach or medication. How could a patient possibly give informed consent in such a scenario, for one thing? How could they meaningfully compare treatment options and make their own informed choices when the advice they receive includes lies?
Alas, some providers believe in placebo effects so much more strongly than their patients’ right to autonomy that when faced with complaints of side effects, they will just lie more and more to their patients in hopes that the side effects will go away, but that’s really just more gaslighting to people who are already in difficult situations.
That's where the fully automated weapons systems come in. Robotics and AI give the megalomaniacs of the world both a reason and the means to kill everyone with impunity.
The rest of the 21st century is going to be really messed up.
reply