I like it! But you should do biochemistry - atoms, acids, proteins, organelles, cells. I think people's intuition is WAY off and this game could help. In fact I one-shotted this: https://sites.google.com/view/scalecraft/full-page-embed (with argon)
It doesn't have to be that way. You can use llms to greatly improve the quality of the code. You do this by focusing on one part of the design use the LLM to explore alternatives, use it to poke holes in what you have. I think the engineering cultures that embrace this way of using AI will far outperform everyone else in the long term.
I just had a discussion with the AI recently about whether to duplicate the code something used or refactor it so both places used the same code. The decision was to duplicate the code, but make notes (and a ticket) to refactor it later - because the time to test the possible regression cases of the original code wasn't available right now.
I could have worked through all that on my own, but having the AI as a rubber duck to talk to - plus the edge cases it was able to point out - was really helpful.
> I just had a discussion with the AI recently about whether to duplicate the code something used or refactor it so both places used the same code. The decision was to duplicate the code, but make notes (and a ticket) to refactor it later
I wonder if a refactor is even necessary.
Concepts like KISS, DRY, etc. are largely human-centric. They're meant to make code easier to write, understand, and maintain - for people.
Duplicating code was traditionally "bad" because it increased maintenance costs, risked divergence, and added cognitive overhead for human developers.
But with coding agents, some of those tradeoffs start to change. An agent can find and update duplicate implementations in seconds. It doesn't need to maintain the same mental map of a codebase that a human does.
And if we have good specifications and comprehensive tests, behavioral divergence should be detectable and correctable. In fact, duplication can offer an advantage over shared abstractions: isolation. Changes to one implementation don't risk breaking unrelated consumers of shared code, and each implementation can evolve independently.
Maybe we won't need as many high-level abstractions, or eventually even some of the architectural patterns we consider best practices today.
If the cost of writing and maintaining code approaches zero, perhaps we should optimize less for human readability and more for runtime performance, memory efficiency, isolation, and correctness.
We might even regain some CPU and memory efficiency by eliminating layers of indirection, virtual calls, and abstractions that primarily existed to make code manageable for humans.
Perhaps the optimal architecture for AI-maintained software is fundamentally different from the optimal architecture for human-maintained software.
A lot of our software engineering best practices are based on the economics/ergonomics of human maintenance. What happens when those fundamentally change?
> Duplicating code was traditionally "bad" because it increased maintenance costs, risked divergence, and added cognitive overhead for human developers.
Until proven wrong by coding agents, this is still the case even for agents. Not sure why you are assuming this doesn't apply for the machine
I cannot read a 50-page T&C document. I mean, obviously, I can physically do it, but I don't have the patience, and I'd lose track of what I'd read. Getting an LLM to summarise a 50-page T&C document is simple, it doesn't lose track, and it spits out a summary in seconds.
I cannot read a 50K LOC code base at once. I mean, obviously, I can physically do it, but I don't have the patience, etc. I need mental shortcuts like abstractions layers to hold even parts of the code base in my head. LLMs don't. They can read that in seconds and hold the whole thing in context.
This doesn’t really support the argument that KISS et al don’t apply to LLMs. They may not ‘need’ these principles for the same reasons as humans, but will still benefit from the context having higher information density.
> but will still benefit from the context having higher information density.
That's an assumption.
And is that what we really care about anymore? Maybe we are ok accepting some "inefficiencies" in the code for the LLM but we get better CPU utilization or cache/RAM usage
So is that ;)
But yeah, the link between context length and performance reduction is well researched. Automatic context management has been getting better, but it’s not a panacea, especially in very large codebases and long sessions.
> Maybe we are ok accepting some "inefficiencies" in the code for the LLM but we get better CPU utilization or cache/RAM usage
If by ‘inefficiencies’ you mean choosing hardware-friendly architectures over those that correspond to a human-friendly world model, then I would agree with you (this was already required in my domain pre-LLMs). Even still that does not stop one from adhering to DRY, KISS, etc as the abstractions required are often either free or close to free at runtime.
>because the time to test the possible regression cases of the original code wasn't available right now.
The models don't know that they're not human when they estimate. It's worth telling them that all estimates need to be made in the context of an agentic AI doing the work.
It's funny, I've had similar conversations, where the agent would recommend the 'lazy' solution because the robust one took too long. More often than not, it was assuming human-hours of work though. Once reminded that a clanker would be doing the work it's conclusion shifted to the robust one.
Now, I wonder if it actually agreed, or if it sensed that I preferred the robust solution and was trained to make me happy.
It’s jarring when the bot tells you something will take a “week of work”. A good reminder I guess that it doesn’t have self awareness as we usually think of it.
Yeah, I've seen that, too. In my case it was less about the dev time and more about the regression testing time (much of which would fall on QA and involved reading from external feeds, etc - so, complicated). But your point is fair. The AI is ... notoriously bad at LOE. Not that it's always wrong, but frequently enough (and by a large enough amount) that tradeoffs like that really need humans considering them.
Yes, this is very bad. I've alerted my team and we'll be deprecating its use immediately. Hopefully procurement can get refunds on the licenses we purchased not that long ago. Something tells me Postman is going to regret violating people's trust - the moat isn't so large that people must endure unethical, if not illegal, behavior like this.
It's a terrible look and also a missed opportunity. AI assisted coding can do wonders for quality, if you choose to take that path. It's mainly because the AI can help you explore alternative paths and quickly select the best one forward. It's crazy to me a secure email provider would openly promote an AI Slop Cannon approach. I mean what on Earth is so urgent about proton mail features?? Especially since their market are technologists and the product is a luxury product essentially. Completely wild.
It says you can't ban AI and shouldn't try. There is no general solution except a worldwide ban of technology itself. What you can do is local and incremental and based on good judgement, patience, intelligence, etc.
This sounds like a short-term strategy at best
You only need an agent to click things long enough to build an API on top of the thing it's clicking on. And running a full windowed virtual machine is expensive!
I'd also caution against saying things like llms are doing most of the work I don't think that's true. And where it is true it shouldn't be. These things are still new tech and they need a lot of hand holding and oversight still.
And its not just me, its everywhere on the internet for everyone. Pretty everyone is complaining about it on the X or atleast was yesterday. Perhaps since it broke many benchmarks there was too much load and that's why they did this and now as soon as it gets less attention quota will improve.
I did and they refused to help and said you used it heavily. Which doesn't make any sense to me since I gave me exact token break down. Quota has slightly improved since yesterday but still nowhere near claude code or other agents.
Why don't banks use PKI? It doesn't have to be in the users face. It might be that your app generates a keypair on startup and sends the public key to the bank server, and the bank server sends you a public key, and you talk to each other with a shared secret. Maybe even you have to go in physically to the bank with your device to set it up and attest the legitimacy of your private key. Personally I'd really love that.
They do often use hardware keystore keys for authentication and other purposes. That's a standard feature supported by all Android devices with greatly varying quality of implementation. There are even rare cases of banking apps using key attestation to provide evidence of those keys being in an actual hardware keystore which has been a standard feature on every device launched with Android 8 or later.
They want to stop people from using tampered or fake versions of their app which could be tricking the user into doing things. They also want to enforce using a device with a mandatory app sandbox and some level of security patches/protections intact. Most aren't willing to ban using highly insecure Android devices which haven't received security patches for many years, but it's trending in that direction.
Around 90% of banking apps are currently compatible with GrapheneOS. It would likely be more like 50% if we hadn't put a lot of effort into improving compatibility with a lot of workarounds for the bugs and problematic anti-tampering checks done by these apps. It would be closer to 20% if we not only didn't have that but also didn't have our sandboxed Google Play compatibility layer feature.
The remaining apps not compatible with GrapheneOS is mainly due the Play Integrity API device and strong integrity levels enforcing having a Google certified device running the unmodified stock OS. Google doesn't allow GrapheneOS to be certified. Certification also disallows many of our privacy and security features. It also isn't compatible with providing fast security updates and would waste substantial resources as it does for Android OEMs. Those apps need to be convinced to stop using the Play Integrity API or to use the Android hardware attestation API provided as part of key attestation to verify the hardware, OS and app are genuine. The latter only solves it for operating systems and hardware they choose to permit.
How do you see the not so distant future when the vast majority of banking and gov apps would require an integrity level higher thar BASIC_INTEGRITY? Which alternatives would you suggest to GrapheneOS users? A second device, perhaps?
We plan on having improved compatibility with banking and government apps rather than most banning using GrapheneOS. Those apps aren't banning it for any actual reason. We don't have to convince them to stop doing what they are but rather to add a bit of code to support alternatives meeting underspecified requirements.
Peterson fallen in a profoundly sad, but also illuminating, way. He obviously had a bit of NPD even from the start, which is true of virtually every public figure. The crack widened and finally split with the help of powerful drugs - fame, money, and Xanax, with a side of audience capture. He quickly went from "liberal Canadian rationalist with the gumption to push back against the left on principle" to just another unthinking right-wing parrot. His view of strength hurt him here, because he was in a crucible that broke him and yet he clearly could not admit it. Contrast him with someone like Sam Harris who, IMO, has been in the crucible for longer but who is far less pugnacious and has managed to maintain his sanity and his nuanced principles despite the forces that want bimodal commentary only.
Sam Harris is not a good example. Sam Harris used to go on hour long crash outs about Trump on his podcast. It became impossible to listen to. I've spoken to people both online and offline who have said that they stopped listening to him because it became unbearable.
Then there is this infamous quote in 2022.
> "Hunter Biden literally could have had the corpses of children in his basement – I would not have cared."
I understand he is being hyperbolic, but honestly it wasn't that obvious though, because the aforementioned crash outs. He also allowed everyone to be quote mine him. It is an unforced error.
Generally I find Sam Harris tbh to be not particularly that interesting anyway. He gets a lot of mileage (like a lot of these personalities) from his elocution. He makes him sound insightful than he really is.
Well I guess we can agree to disagree because I think there's a lot about Trump to crash out about. It's difficult to understand how someone who so obviously a villain is so popular. It's disorienting and frustrating, like walking out of Star wars hearing people praise Darth Vader for how wonderful he was and how nasty Luke was to him. How The empire is a staple job Creator and the rebel alliance a bunch of filthy blooded miscreants who don't appreciate what they have...
If we are making Star Wars references. The whole lore of Sidious is how he and Plagueis manipulated various parties in the galaxy to gain power. In the story itself it explains how this can happen. The Jedi weren't really the good guys. They were an ossified monastic order that had lost its way.
Donald Trump and Hillary Clinton are both rival elites vying for power to enrich themselves and their circle via the apparatus of the Republican and Democratic parties. They are both reprehensible, for different reasons. Voting for Trump was a middle finger to both establishment parties in the US (the Republicans didn't want him running for the party). There are no good guys here. They were both shit.
The crash out was due to the fact Sam Harris couldn't wrap his head around the fact that people were fed up of public figures that are similar to himself. Quite frankly, it is embarrassing. He appeared to do very little introspection on the matter. His whole brand is him trying to rationally understand what's happening, and he did the exact opposite.
Cory Doctorow, on the other hand, is another left-wing intellectual that did understand the sentiment and has made an effort to understand people he deeply disagrees with. Guess whose ideas are more relevant currently? It isn't Harris; Harris is irrelevant these days.
> Bullshit. Trump is the logical outcome of republican party project. Vote for him was vote literally for that.
I don't understand why people like yourself come up with this stuff. There are loads of clips from 2016 where people were saying this. It's 10 years later, and I am still having conversations where we are denying the reality of what happened; it is bizarre.
> Republican politicians did not liked hia style, but did liked his project, supported and created project 2025, live the violence and all of that.
Project 2025 is a thing from the Heritage Foundation and wasn't a thing back in 2016, because it was published in 2023. You are just repeating stuff you've heard on the TV/YouTube.
The Heritage Foundation has been moving towards Paleo-conservatism since about 2021. The Paleo-cons were ejected/rejected from the Republican Party in the Buckley days.
I was talking about the 2016 campaign. Back in 2016 (not 2024), the Republican establishment didn't want him there. Pretending otherwise is asinine.
> The corruption was making Trump cool to these people.
People liked Trump because he's brash and loud and has an attractive wife who used to be a model. He has a ridiculous hairstyle. He flew in on a massive jet that has his name on the side of it.
He also trolled/manipulated the media constantly, which was pretty funny on the face of it. E.g., I remember he was going to talk about Obama's birth certificate or something like that, and he got CNN to broadcast his entire rally, and he mentioned it for like 5 seconds at the end.
I am English, I live in the UK. My mates and I would get some beers and watch the Hillary Clinton/Donald Trump debates late at night to see the shit show. It was really funny. "What is old Donald going to say this time?!".
I understand that doesn't appeal to everyone, but a lot of people like this stuff. I won't lie, I kind of liked it as well at the time.
Compare that to someone like Jeb Bush or even his opponents, Hillary Clinton; there was no momentum, no hype.
Malcolm McLaren is quoted as saying, "Pro-wrestling is real, and politics is fake." You need to keep that in mind when watching any political coverage.
> Not nearly in the same way. Not even close. And Trump voters knew that.
No, they didn't. What people like is that he was talking about stuff they cared about and they believed he was speaking for them.
It is 10 years later, and people like yourself are still misstating what people were saying and have done no introspection.
Trump voters aren't evil. You really need to do introspection because you still don't seem to understand why there were major political upsets, both in the US, UK and EU.
You are doing exactly what many people were criticising Sam Harris for. I was fed up of it back in 2017, and you and Sam haven't learnt anything.
It is one of the reasons why I don't bother talking to people about stuff like this, because you can't look past the surface, and you make no attempt to understand what happened.
I spent quite a lot of time reading and interactimg with conservatives and republicans. I am from that environment.
Claim it was rejection of both parties is self serving bullshit. It feel good sound good story and nothing more.
> which was pretty funny on the face of it. E.g., I remember he was going to talk about Obama's birth certificate or something like that, and he got CNN to broadcast his entire rally, and he mentioned it for like 5 seconds at the end.
That one is waaay more about "black man not valid American" thing then about "funny" in an abstract way.
> rump voters aren't evil. You really need to do introspection because you still don't seem to understand why there were major political upsets, both in the US, UK and EU.
Yes they were making bad choices, knowingly. Really, that is their choice. You need to do introspectiom if you still need to defend them and their choices.
There is really no reason to play this "only left of center can be criticised, impossible for far right and Trumpians to do something bad" game.
Dissapointingly, I got the exact response I thought I would.
> I spent quite a lot of time reading and interactimg with conservatives and republicans. I am from that environment.
I don't believe you. I am not conservative; I am not American, and I can tell you haven't.
> Claim it was rejection of both parties is self serving bullshit. It feel good sound good story and nothing more.
Why? Because you say so? I don't even understand how it is "self serving bullshit". People were fed up with run-of-the-mill politics. Even on the left wing you had many people that were radical (that have become less so over time) entering into politics around that time.
I also must have imagined reading all the comments and seeing all those videos that people posted around that time, specifically on the theme that they were voting for Trump because they believe he was something different.
I remember listening to many hours of people just talking in Skype, Google Hangouts and similar places at the time. I get really annoyed when people try to gaslight me about events that I was paying a lot of attention to at the time.
> That one is waaay more about "black man not valid American" thing then about "funny" in an abstract way.
I think it is funny to get CNN to broadcast your rally for you based on deception, because CNN wanted him to something racist and he didn't. He trolled them because they are predictable.
> Yes they were making bad choices, knowingly. Really, that is their choice. You need to do introspectiom if you still need to defend them and their choices.
Apparently you can read minds now. How do you know? I really doubt anyone said, "I know Trump is an evil man, but I am going to vote for him," or something to that effect. If you claim you have. You are lying.
> There is really no reason to play this "only left of center can be criticised, impossible for far right and Trumpians to do something bad" game.
I am not. The fact that you can't recognise that because I have take on Trump that is sympathetic to MAGA / Trump supporters is ridiculous.
Most Trumps aren't far-right either. You say you listen to Republicans and Conservatives, but you clearly aren't if you are claiming that.
I think, TBH, you are lying to me. So I don't want to talk any further. Go away.
I cancelled my subscription. Harris would always get lost in internet wars and definitely spent too much time reading idiots tweets and getting angry over them.
Sure, JD Vance is a cretin and Trump is vile. Can we please go back to intellectually interesting pursuits?
I agree, and I find it sad that so many people and companies are in a rush to take the human out of the loop. But for agentic/human hybrids not even all of those work out well - and the error modes have more to do with the "alignment" of the human. That is, humans with strong knowledge, an aesthetic sense for structure, strong curiosity - those humans do well in this loop. Humans that just want to "make money" using the LLM as a slave, interacting with it as little as possible, taking no interest in what it does (bad company leadership in miniature) get poor outputs. When the human is an invested, knowledgeable curator of code, someone who has a vision that goes beyond the external, you get some really great stuff.
reply