> But we have no test suites that answer "is this code maintainable" or "does this feature make sense in the context of our business", so AI still fails those tests.
I've been struggling with this because I do believe maintainability of code is important. But the way it is going, if AI is writing all the code and no human needs to read it anymore, does it really need to be maintainable? Does it matter if it looks like alien code since we can throw it away when using the next AI model? I have no answers of my own to these questions...
In my work I use AI for code generation in two distinct ways:
- For throwaway prototypes I pretty much vibe code, only making manual tweaks occasionally if the agent gets stuck or something looks wrong visually.
- For production code I stay in the loop, reviewing and iterating on the code with an agent.
This leads to me having experience with “messy” prototype code and “maintainable” production code. Whenever my prototype code lives longer than expected it gets to a point where the
agent starts introducing regressions or churning on changes.
This happens much less often with the “maintainable” code. From my perspective it seems pretty clear that LLMs are faster, more effective, and less error-prone in a well structured codebase
> But the way it is going, if AI is writing all the code and no human needs to read it anymore, does it really need to be maintainable?
It's more important than ever, because code is just a way of expressing intent. If the code becomes overly complex, it de-coheres, and intent is lost across the system you are building.
This is why vibe coded projects turn into slop and eventually stop iterating: at some point that actual intent is too vague and contradictory because of the many layers of code that have been added.
This can happen with modern fintech because of greed. There's a reason they can offer such cheap services. The customer takes a risk in return. Now that risk has materialized.
I see your point about greed. Thanks. Let me still contrast that: GPT6 has 99.9 in ARC-AGI 3 and multiple bug-bounty programs closed due to the sheer amount of automated attacks and reports.
And they are "FinTech". "Oh, that email looks legit, let's just hand out the data.", like they have never witnessed phishing from the old days... am curious about the story here. That PR-spokesperson is more than damaging...
Here is one of the replies I got during my conversation with their agent (unsure if human or automated):
"Your personal data must be held until it is permissible to erase it in accordance with the law. Rest assured, it is totally secure and only held for this purpose."
This was in the same conversation where I sent them the article.
> "I have checked our records and can confirm that you have not received any notifications or communications regarding any security incidents or data breaches in the past 30 days.
> We take your privacy extremely seriously. All data transmissions between our mobile apps, servers, and third parties are fully encrypted, and your personal information is stored in secure data centres with restricted access. If there is ever any security incident that impacts your account, we will always contact you directly with instructions.
> Are you asking because you recently received a suspicious email, text message, or noticed an unusual transaction on your account? Let me know, and we can investigate that together."
Breach yes, but if they cannot 100% sure identify if your data was given out falsily, then they cannot say. They're not allowed to disclose that they provide your information to LE. So they can only inform you directly if they're 100% sure the specific information request response was sent to false entity. This is very hard to do.
This was a targeted attack towards specific individuals, probably carried out by a state actor or someone after data of very valuable individuals. Unless you're one of those (oligarch, etc), you're probably fine.
Trying to find a way to tip toe around KYC, whilst keeping their customers safe, has also turned out to only use insanely stupid methods, though. So why did they already do that?
FYI it turns out that humans are pretty bad at comparing faces to ID documents. Like, really quite bad.
Automated methods, like the ones Revolut use, are significantly more effective at KYC than a Jane Doe working a 9-5 at a bank. In no way is it “tip-toeing around KYC”, and while really unfortunate leaking a selfie is pretty low down on the list of “bad stuff a bank could leak”.
The implication that the solution to this is to somehow convince your direct competitors to do inferior in-person KYC for you is the most ridiculous thing.
>The implication that the solution to this is to somehow convince your direct competitors to do inferior in-person KYC for you is the most ridiculous thing.
People work with their competitors all the time (see Netfix vs Amazon). Whats ridiculuous is the claim that a scammer would prefer to show up physically at a bank and risk being exposed instead of operating remotely.
>leaking a selfie is pretty low down on the list of “bad stuff a bank could leak”.
don't some of them require a selfie while holding legible official documentation?
> Whats ridiculuous is the claim that a scammer would prefer to show up physically at a bank and risk being exposed instead of operating remotely.
Of course they wouldn’t prefer to show physically. What does that mean though? Are you saying no scammers showed up physically to banks, therefore banking fraud rates are less? Do you have a source for that?
> don't some of them require a selfie while holding legible official documentation?
You can of course do KYC as stupidly as you like (zoom calls anyone?) - Revolut (and their providers) obviously separate document presentation from the liveness check (and fyi this is a short video, not a selfie. The selfie they are talking about is just a capture from the video)
Is your argument supposed to be more convincing because you added the word "patently"?
>What does that mean though?
It means that when you find a way to bypass purely online identity verification checks executing fraud at scale is easier than the physical alternative. As you would say, this is patently obvious.
In the USA they already take passport photos. Being able to receive mail addressed to a name is the closest thing to a national ID the USA has. They're already depended on for identity verification quite a lot.
Some post offices in the US also function as a so called notary public. Basically, they can verify your identity and attest that it's you who sent/did something.
This is used quite often for important things that don't have offices themselves.
Try opening one now. Today it's hard to get a hire purchase contract as an existing custoner (already known and verified) without photos of the ID and selfie.
Likewise, opened a couple in the past few years and never saw this. That said, bank lobbies have tons of ambient cameras anyway, so they don't really need a selfie.
If Wasmi supports all the Wasm proposals that you need for PGlite and if Wasmi supports all the WASI features you need (Wasmi only supported the standard WASI features without extensions), then Wasmi should work for your use-case. :)
This! I live in Western Europe and have never owned a car. I mostly travel by public transport or bike. I rent a car about once a year, but could do without that as well. Before people tell me that I am an exception: I know other people like me in various other places in Europe! It is possible and actually good for your body to have to move, at least to the train station, carrying groceries etc.
>This! I live in Western Europe and have never owned a car.
I'm also in Western Europe and almost always needed a car for my commutes to work, unless I was willing to accept a 1-2+ h public transport commute each way, because tech employers where I live love locating their offices in the sticks where land is cheap because they're broke and cheap, and the city hasn't significantly updated public transport infrastructure in ~30 years, but the population and size of the metro area grew by like 3x.
>Before people tell me that I am an exception: I know other people like me in various other places in Europe!
The thing is, Western Europe is not a country, but a collection of various countries with various economies, cities, and infrastructure, and various people with various housing and career locations, some close and well connected to public transport, some not at all, and from my circle of acquaintances, yes, you are the exception, as most people I know get to work by car.
So, you can't possibly throw such blanket statement over everyone in Western Europe, unless you're massively obtuse or living in a bubble without realizing it.
Like, I'm sure most people in Munich, Paris, London, Madrid, Vienna, etc can easily get to work on the excellent public transport they have, but I don't live there and don't have such amazing public transport, so then car it is for us.
>It is possible and actually good for your body to have to move, at least to the train station, carrying groceries etc.
Bruh, I only need a car just to get to work in reasonable time, not to move or carry groceries.
I get my groceries from the supermarkets 5-10 minute walking distance to my place, but I don't have the luxury of employment jobs within the same distance, so car it is, unless I want to be wasting my life switching connections between slow and infrequent trams, busses and trains which often suffer delays.
And regarding the "body movement" as a pro for public transport, I'd rather have more free time to move in gym and to the workouts my body condition needs, rather than waste my time walking and sitting/standing on public transport, as that's not an actual workout but a waste of time, as all you can do is browse HN on the phone with the hand that's not grabbing the bar. Driving is also a waste of time, but at least it saves me almost 1 hour each way which is massive time save that I can use how I like.
I'm an OSS maintainer and to me it's not just about the review itself. Being greeted by a wall of text for every little small thing is counter-productive. I hate going through 2 pages of text for each PR. It usually shouldn't take more than a couple of sentences if you understand the issue and the solution.
But most important for me: lots of time the PR just adds even more code, although other options do exist (ie sometimes REMOVING some code). You have to know the codebase well in order to find those objectively better solutions.
How do you handle them? I'm not facing this problem as the team I work with is very senior and have good taste and discipline. But I can imagine it will be a problem at some point, and I frequently have to personally tell Claude to rewrite it's vomit in English. That's probably step one for people submitting poorly written PRs, reject them until they are written clearly and concisely. And if they are too big, also rejecting them and telling them to go back to the drawing board and submit smaller more focused change. But I'm not int his position so I'm taking an educated guess.
I've seen it. Walls of text with stereotypically worded non-summaries that just repeat all of the code in words, mutating values all over the place instead of the obvious canonical one place that touches related values...
Yeah you can use LLMs, but don't let me notice it from the quality of the output.
I've noticed that LLMs seem to be especially bad at things relating to space, position and movement. I guess they have to synthesize that part of human intelligence entirely, it's not in the words.
Or, the SaaS-less approach, if a issue description is too messy/long, close it with "Please reopen with proper and concise description focusing on the issue" then lock it. Eventually people catch up and stop with the slop, just like in real life.
But you have to be able to say "No ...", rather than just slapping another subscription on top of an already broken workflow.
If you're maintaining OSS, that's understandable, and you're free to say no, but in the corporate world, that's not realistic, AI is here to stay, if they don't harness it they would just be left behind.
even if the AI gets good and stops writing sloppy stuff, it's still gonna write a lot of stuff, and you're gonna review it anyway, and take responsibility and ownership, and it's still gonna take you more time, because the bottleneck is now reviewing and understanding the code.
I agree that the workflow is broken, but only on the reviewing side, AI is a tool we use to make products just like any other we used in the past, punch cards, machine code, assembly, ...etc. AI is just the new tool that sits on top of the code as the next level, no one codes with punch cards, no one writes machine code anymore, we used to write the compiled language and don't care about how it's compiled or turned into machine code, same with AI, although it's not there yet and still requires babysitting by engineers, but that's our new job now, and we need to learn how to use it and make our lifes easier.
Sure, but that's irrelevant when someone says "I'm an OSS maintainer" and the context is explicitly about reviewing code submitted by others in a FOSS context, where you can say "No" and don't need any SaaS in order to do so.
That's not true at all of the corporate world. If your team is mass producing slop and you don't have processes in place to get it under control, you've got a big problem on your hand.
If any engineer sent me a 20,000 line refactor I'd immediately reject it and tell them to go back and start making changes incrementally at minimum. More likely I'd force them to have a whole design discussion with the team to make sure that what they are doing even makes sense.
What happens if they push out slop that significantly increases your infrastructure costs?
What happens if they push out slop that significantly increases the number of bugs or outages?
What happens if they push out slop that has no observational metrics, dashboards, or tooling?
In every case you push back on the team and make them fix their shit. I don't care if they are using LLMs or not. They are responsible for their work being sufficient quality. If they aren't meeting those standards, then they need to step it up.
I've been struggling with this because I do believe maintainability of code is important. But the way it is going, if AI is writing all the code and no human needs to read it anymore, does it really need to be maintainable? Does it matter if it looks like alien code since we can throw it away when using the next AI model? I have no answers of my own to these questions...