Since the November/December Opus and Claude Code, I found I don't need to read the code any more. Architecture overview sure, and testing yes, but not reading the code directly any more.
Me (and my friends similarly) inspect code indirectly now - telling agents to write reports about certain aspects of the code and architecture etc.
I do regularly read the code that Claude outputs. And about 25% of the time the tests it writes will reimplement the code under test in the test.
Another 25% of the time the tests are wrong in some other way. Usually mocking something in a way that doesn't match reality.
And maybe 5% of the time Claude does some testing that requires a database, it will find some other database lying around and try to use that instead of what it's supposed to be doing.
And even if Claude writes a correct test, it will general have it skip the test if a dependency isn't there--no matter how fervently I tell it not to.
If you're not looking the code at all, you're building a house of cards. If you not reading the tests you're not even building you're just covering the floor in a big sloppy pile of runny shit.
> I do regularly read the code that Claude outputs
You probably could have s/Claude/Human/ in your rant and been just as accurate. I don't know how many times I've flagged these issues in code reviews. And that's only assuming the human even bothered to write tests...
What I find is that when I ask AI to write tests it writes too many, and I agree with you that a lot of them are useless. But then I just tell it that, and it agrees with me and cleans it up. Much faster feedback loop and much better final result.
I feel like people that look at a poor result and stop there and conclude it's useless have made up their mind and don't want to see the better results that are right in front of them if they just spend an extra 5 seconds trying.
How do you know whether the tests it’s spits out are bad if you don’t read the tests.
We’re not dealing AGI here. Tests aren’t strictly necessary for humans. They are for AI. AI requires guardrails to keep from spinning out. That’s essentially the entire premise of the agentic workflow.
I’m pretty sure they just meant they do testing not that they read the tests and that’s what everyone else who responded interpreted that as well.
You can get Claude to write good tests but based on what I’m seeing at work that’s not what’s happening. They always look plausible even when they’re wrong, so people either don’t read them, skim them very quickly, or read the first few assume the rest work and commit.
I think Claude is great for testing because setting test data and infrastructure is such a boring slog. But it almost always takes a lot of back and forth and careful handholding to get it right.
I read the tests, it also is really really good to have Claude verify that removing the changes in question break the tests. This brings the quality way way up for me.
I'd understand not reading the code of the system under test, but you don't even read the tests? I'd do that if my architecture and design were very precise, but at this point I'd have spent too much time designing rather than implementing (and possibly uncovering unknown unknowns in the process).
> Me (and my friends similarly) inspect code indirectly now - telling agents to write reports about certain aspects of the code and architecture etc.
Doesn't this take longer than reading the code?
I can see how some of this is part of the future (I remember this article talking about python modules having a big docstring at the top fully describing the public functions, and the author describing how they just update this doc, then regenerate the code fully, never reading it, and I find this quite convincing), but in the end I just want the most concise language for what I'm trying to express. If I need an edge case covered, I'd rather have a very simple test making that explicit than more verbose forms. Until we have formal specifications everywhere I guess.
But maybe I'm just not picturing what you mean exactly by "reports".
I've seen the code these models produce without a human programmer going over the results with care. It's still slop. Better slop than in the past, but slop none the less. If you aren't at minimum reading the code yourself and you're shipping a significant amount of it, you're either effectively the first person to figure out the magic prompt to get the models to produce better code, or you're shipping slop. Personally, I wouldn't bet on the former.
Yeah, these models have definitely become more useful in the last months, but statements like "I don't need to read the code any more" still say more about the person writing that than about agents.
Eh, I don't think Swift would ever have dethroned Python. What pain point would it practically solve? I don't use Python often but I don't hear folks complaining about it much.
I do, though, think Swift had/has(?) a chance to dethrone Rust in the non-garbage collected space. Rust is incredibly powerful but sometimes you don't really need that complexity, you just need something that can compile cross-platform and maintain great performance. Before now I've written Rust projects that heavily use Rc<> just so I don't have to spend forever thinking about lifetimes, when I do that I think "I wish I could just use Swift for this" sometimes.
You're right, though, that Swift remains Apple's language and they don't have a lot of interest in non-Apple uses of it (e.g. Swift SDK for Android was only released late last year). They're much happier to bend the language in weird ways to create things like SwiftUI.
If you think for a second that the Israeli state allows this company to sell its services to anyone who is opposed to their interests, you really don't understand how defense companies and states work.
My point is a simple one: the company was hired by someone. Was it the opposition party? To say this is entirely Israel's doing implies the Slovenian party that benefits just happens to have gotten lucky. The reality is likely considerably murkier than that.
Any company that sells this type of services exists as an extension of its parent state. Any contract it offers, especially to a political entity in another state, will be scrutinized by state authorities and allowed by them or not. Sometimes, those contracts will be forced on the company based on state-level negotiations.
Noone is saying that the party that contracted this company (if indeed it was a Slovenian party and not the Israeli state itself) for this service doesn't carry blame. But both the company itself and the state of Israel carry just as much blame for offering, permitting, and carrying out such services.
By your logic, if someone were to found a legal private paid assassin company in France, and then the opposition party in Germany hired this company to assassinate the German chancellor, you'd say that it's unfair for Germany to blame France for this.
Curious: can you show the research steps you took to reach this conclusion? Really curious how we can all easily determine which companies are and aren't CIA fronts!
No, AI is truly useful in software engineering. I was a skeptic until I started using it. No, it isn’t going to solve every problem out there, but it’s a force multiplier.
You pay understanding for speed. How much this trade is acceptable is up to you and the task you have in front of you. I cannot recommend it as a general solution.
This field doesn’t do well on long-term thinking. Even if all this turns out to be a net loss, it will be reinterpreted as a win and just an opportunity for even more of the same solution. There are numerous examples of this, e.g. the OOP craze. Tech is a stock market of ideas and HN is a trading floor. The “line goes up” logic applies - not merit.
You may not recall the crazy era of OOP where people would go bonkers with massive object trees trying to objectify everything and using operator overloading to do (dumb) things like adding a control to a window with +=.
That’s just falls. I’ve spent disproportionate amount on “understanding” awful tooling like Gradle and npm. There’s no value in it if you’re not an infra engineer. It would take me a couple of days to manually restructure my hobby app, now I can just say “extract this into another workspace/subproject” and be done with it in minutes. And that’s just one example.
I agree with this sentiment. I just also see AI-driven development in core business logic, where truly understanding what is going on is essential and yet completely disregarded.
Its previous head had a term that didn't expire until 2028 but he resigned after pressure from Elon Musk (who didn't like that he got fined), now a Trump-friendly head has been installed. What, realistically, would be the consequences if he lied? Likely none. Government officials lying on record is an every day occurrence these days.
It does not at all mean that this controller was overworked when this crash happened; that would be failed reasoning and misuse of evidence.
It just raises the question, which should be looked at.
It's scary that so many don't seem to know the difference. This is how misinformation starts and spreads.
You're 100% right, a "Trump-friendly" administrator has been "installed" so we can't trust the FAA's conclusions. The last guy quit so this guy is definitely going to lie.
I think OP's point is that sending your browsing data to a server, be it American or Chinese, isn't "no bs".
I see this recurrent feeling on HN that because the US does bad things we shouldn't care about other countries doing the same. I think we should care about all of them!
The feeling is that "no one" cares about it being sent to American servers, why should they suddenly care about it going to Chinese servers just because they're Chinese.
Not only that, USA is far more likely to send someone to kill you than china is. So between the 2 I'll take china (I'd prefer my data to not be sent to any foreign power).
reply