HNHacker News
TopNewBestAskShowJobs

ang_cire

2,309 karma · joined June 12, 2023

submissionscomments
ang_cire··on I built a vulnerable app and spent $1,500 seeing if LLMs could hack it
There is a cyber security verification program you can join to avoid these blocks:

https://support.claude.com/en/articles/14604842-real-time-cy...

If you work in security (which I assume the OP does), they should be able to get in easily. I think most people just don't know this is a thing.

ang_cire··on They’re made out of weights
I think that the reaction here alone disproves this somewhat, because imo it's exposing how anthropocentric most people still are, despite all evidence that we are in fact just "meat all the way down".

Despite all the evidence that we are in fact just biological machines, people still persist the theory of our own uniqueness from other creatures, which we ourselves often treat as biological machines.

This adaptation is wonderful to me specifically because I think it shows that our shifted goalposts of, "well we're not just animals, we can think and reason" was never more than a convenient excuse for many people (and as evidence of animal intelligence continues to mount, denialism still attempts to preserve this distinction by claiming human thought and reason is different than 'animal' thought and reason, sans evidence).

It's not about who created it or why, it's about how people still haven't actually internalized the point, because the subject changing from human to LLM doesn't intrinsically change the message about consciousness, but the reaction being a 180 shows how hostile people are to that message, still.

ang_cire··on The California state assembly has passed the 'Protect Our Games Act'
Because they don't?

Do you eat at the cheapest restaurant every day? Do you think that every Michelin Star restaurant immediately fails and shutters? Do you think everyone buys the $80 prepaid flip phones, and no one actually buys the $700+ iPhones?

Most people don't gravitate towards the cheapest option (in fact, many people find the cheapest option automatically suspect and won't buy it), but rather want a balance of affordable and desirable. No one living in NYC is doing it because they're gravitating towards the cheapest option in the first place, they're there because it has a high level of desirability comparative to its cost, even as expensive as it is.

ang_cire··on Citing 'severe' math deficits, UC faculty demand a return to SAT tests for STEM
To be clear, I never endorsed any of these routes. They're all bandaids that try to make up for the intentional defunding and prisonification of our public school system.

We can't solve the intentional sabotage of our educational system by keeping kids in it for longer via remedial classes, which are supposed to be focused on kids who have personal barriers to learning, not systemically-imposed ones.

ang_cire··on The California state assembly has passed the 'Protect Our Games Act'
You're wrong. There is no wealth-tax based mass migration of the wealthy from NYC to Florida.

1) https://fiscalpolicy.org/wp-content/uploads/2025/10/20251009...

> As we get more data on the post-pandemic period, we increase our knowledge of the major upheavals that took place in New York between 2020 and 2022. Despite the state suffering a deep recession and massive out-migration during the pandemic, data show that New York’s tax base remains resilient. When taxes on millionaire earners were raised in 2021, tax revenue to the state increased by an estimated $3.6 billion and there was no detectable increase in high earner out-migration.

ang_cire··on The California state assembly has passed the 'Protect Our Games Act'
'Some' is a meaningless non-metric. Some people do anything.

Actual studies [1] show that the rich are not moving in response to wealth taxes, and in fact when they do move, it's almost never due to taxes.

> As we get more data on the post-pandemic period, we increase our knowledge of the major upheavals that took place in New York between 2020 and 2022. Despite the state suffering a deep recession and massive out-migration during the pandemic, data show that New York’s tax base remains resilient. When taxes on millionaire earners were raised in 2021, tax revenue to the state increased by an estimated $3.6 billion and there was no detectable increase in high earner out-migration.

1: https://fiscalpolicy.org/wp-content/uploads/2025/10/20251009...

ang_cire··on The California state assembly has passed the 'Protect Our Games Act'
Yep, this is the "higher taxes will drive new yorkers to florida!" fear-mongering (sometime, sadly, even by people who don't actually know better but automatically shill for companies).

There are so many games (like Hitman: WoA, which I love btw) that "require" online access in order to provide the same functions that previous games by the same devs provided fully offline (e.g. keeping track of your weapon unlocks).

This is just clawing back some of the consumer protections that the "we're not selling you a product, we're selling you a temporary and arbitrary license that we reserve all rights over" BS snuck around.

ang_cire··on Citing 'severe' math deficits, UC faculty demand a return to SAT tests for STEM
I think because if their graduation requires them to pass that course (and it's a freshman-level course), the university is basically facing a choice of, "teach them this thing", "change graduation requirements", or "either kick the kids out or willfully let them waste 4 years just to not be able to graduate".

Both the latter 2 are big choices for a university administration to make, so it's much easier to ask the professors to make up the difference. That's why it's the faculty and not the admin demanding this; they know what the admins are asking them to do is impossible.

ang_cire··on Shunning AI is the human choice
But are these actually completely different technologies, and if so, where is the dividing line? Firearms certainly have not decreased in significance, and they're the modern version of a bow, which is simply 2 iterations later in propulsion methods: tensioned string -> high-tension cable -> high-pressure gas.

Are LLMs really going to fall off in significance, or will it just be the nth newest incarnation of LLMs?

The function of what an LLM does (generative language) is what people seem to take issue with, but the function is here to stay, even if the next iteration has a different name or method.

ang_cire··on Google's Antigravity Bait and Switch
It's funny how people talk about de-Googling their lives as a struggle, but there are only 2 things I can think of that I use them for anymore, and that's 1) gmail, and 2) google maps.

It's always surprising to me when people mention these google services I've never heard of. What do you mean a Google IDE? Haven't you heard of Vim, bro?

Mostly-jokes aside; don't trust Google! Google is asshole.

ang_cire··on Shunning AI is the human choice
I don't know about you, but I can buy bows and arrows at hundreds of sporting goods stores in my local area alone, and I even know of 2 local blacksmith shops that sell swords.

Castles still exist as well, you just aren't invited to them (which was true for us peasants back in the day, too). Trump is still trying to get one built under the ruins of the East Wing, in fact.

ang_cire··on Shunning AI is the human choice
In the same way that any technology could just magically disappear, sure.

But I hear everyday, non-IT-sector people talking constantly about how they're using it, and that means there's a demand for it, and someone is going to supply it. I think a lot of anti-AI people think it's still equivalent to the PDA, and don't realize it's a smartphone already.

The other side is that "AI" is of course very very broad and isn't new, and e.g. medical vision models are making advancements that are having huge impacts on patient care already, especially around early cancer detection. Those aren't going away (and shouldn't), so there's still going to be a demand for the underlying technology and infrastructure to support it, even if LLMs stop being spammed everywhere.

The other thing which people seem not to understand is that you don't need a whole datacenter to RUN individual LLMs, you need it to train them, or to run them at scale for thousands of customers. A lot of the upper-mid-tier models that exist now can be run on a single (beefy) 4U server in your closet if you've got the GPUs to put in it. And people are running e.g. Deepseek V4 Pro FP4 locally. If you've got an actual server room, like at a university, you can run the full, un-quantized versions with ~2-4 servers.

Technology that is living in peoples' homes and businesses already is not going to just disappear. It's a lot less centralized than the market prevalence of OpenAI and Anthropic would lead you to believe.

ang_cire··on Write some software, give it away for free
All my personal software is MIT licensed. Selling software isn't my bag, baby.
ang_cire··on I'm done making desktop applications (2009)
> Why I'm Done Making Desktop Applications

To save you a click: It's harder to monetize desktop apps than webapps.

Lol. LMAO, even.

ang_cire··on Tim Cook's Impeccable Timing
Honestly, Apple may very well be betting that AI in it's current form is transitional, and might be better off letting others duke it out for now.

We still haven't found and agreed upon the 'best' way for AI to work in a given environment, and the experts in this area aren't working at Apple. Once there is a clear path forwards to use AI best, it makes sense for Apple to jump in.

ang_cire··on The Uncanny Valley and the Rising Power of Anti-AI Sentiment
As a millennial, I will be the first to run my brain on my toilet homelab servers.
ang_cire··on Codex for almost everything
Ah yes, those hilarious US Americans, not wanting to trust all their chat communication to Meta...

Edit: literally just saw this article afterwards:

https://www.lbc.co.uk/article/dubai-police-spied-private-wha...

UAE government getting access to private WhatsApp messages (presumably in conjunction with Meta).

ang_cire··on The seven programming ur-languages (2022)
This is very cool, and I did not know this. Thank you!

I wonder if my formal university python training predated this change (~2010), or if the professors were themselves unaware of this.

ang_cire··on Why Japan has such good railways
I'm sure geography helps, but it's certainly not the driver for good train service design. Cities in Japan are definitely not laid out in thin lines, and there's not just a few routes in any given city. I was living in Nagoya back in high school, and its train lines are sprawling.

Side note, there actually isn't one shinkansen from Kagoshima to Hakodate, that route would take you on 5 different shinkansen lines: Kyushu, Sanyo, Tokaido, Tohoku, and Hokkaido. But I get your point.

ang_cire··on Claude Opus 4.7
If you're not doing code review, you're not doing code review. If you're not gating builds on code review, you're not doing code review. If your developers are lazy and just approve the PRs as they land, you're not doing code review.

If you're thinking there is some magical line where LOC < n gets properly reviewed, but LOC > n doesn't, I assure you that's not the case.

And no one is turning off their approval gates in their build pipeline just to accommodate AI code.

ang_cire··on AI cybersecurity is not proof of work
The missing part of their intended meaning is "skilled hackers". Unskilled hackers are everywhere, and they're bad at programming, but so are unskilled programmers.
ang_cire··on Cybersecurity looks like proof of work now
> This chart suggests an interesting security economy: to harden a system we need to spend more tokens discovering exploits than attackers spend exploiting them.

What this fails to take into account is that unless the codebase is changed, there are a finite amount of actual (and even fewer actionable) bugs in a piece of code, but an infinite amount of potential attacker spend; nothing stops you running mythos against it, whether it finds anything or not, and because each run is atomic by nature, you just have to play the numbers out and see when the average vuln discovery rate is dropping. You could spend a billion dollars and not find anything, without the defender spending a cent.

Generally speaking, the advantage goes to whoever can spend more time or money on security research (this has always been true, which is why the NSA was able to find Windows exploits that M$ did not). But eventually the fount of bugs in a piece of software will dry up, and attackers have no way of knowing if that's the case or not before dumping money at it (especially since attackers do not generally coordinate unless they're just branches of the same 'entity', e.g. nation-state).

ang_cire··on Codex for almost everything
In what world would I prefer to give someone access to me via a messaging app rather than a fully-async text SMS message? I don't even love that people can see if you've read their texts now.

Fully agree about phone calls though.

ang_cire··on Claude Opus 4.7
> Humans are lazy though and I can't help but feel we are being inundated with sketchy apps doing all kinds of things the authors don't even understand... there is a good chance they have no clue what they created.

I have bad news for you about the executives and salespeople who manage and sell fully-human-coded enterprise software (and about the actual quality of much of that software)...

I think people who aren't working in IT get very hung up on the bugs (which are very real), but don't understand that 99% of companies are not and never have met their patching and bugfix SLAs, are not operating according to their security policies, are not disclosing the vulns they do know, etc etc.

All the testing that does need to happen to AI code, also needs to happen to human code. The companies that yolo AI code out there, would be doing the same with human code. They don't suddenly stop (or start) applying proper code review and quality gating controls based on who coded something.

> The only way I felt comfortable using Claude Code was holding its hand through every step, doing test driven changes and manually reviewing the code afterwards.

This is also how we code 'real' software.

> I can't help but think that massive code bases that have moved to vibe coding are going to spend inordinate amounts of time testing and auditing code

This is the correct expectation, not a mistake. The code should be being reviewed and audited. It's not a failure if you're getting the same final quality through a different time allocation during the process, simply a different process.

The danger is Capitalism incentivizing not doing the proper reviews, but once again, this is not remotely unique to AI code; this is what 99% of companies are already doing.

ang_cire··on The tech jobs bust is real. Don't blame AI (yet)
Because we live in a techno-oligarchy now? Because the leaders of the top tech companies (by revenue) literally sat behind the President at his inauguration?

Because they can afford to buy the 'right' to do what they want, and you can't, and what they want is cheaper labor who they have more control over, and H1B workers will never rock the boat because the visa is a sword hanging over their heads.

Downvote all you want, it's the truth.

ang_cire··on The tech jobs bust is real. Don't blame AI (yet)
I think a lot of people don't see the historical trend of where investment moves around. The 90s dot-com era was driven by massive investment in e-commerce (remember that term?) businesses. It created a bubble, which then burst.

When bubbles burst, investors change tack, and when investors pull back, companies turtle and do layoffs.

After Y2K, investment moved on to real estate. In 2008, that bubble burst. It moved on to social media next.

I would argue that bubble burst somewhat silently during COVID with the quiet deaths of Xitter and Facebook (from a standpoint of cultural relevance), and investment transitioned towards and into its current AI bubble.

Investors are always going to move on once the lustre of a field has waned, and I'd hazard that we'll see investment move somewhere other than software tech next (if I had to prognosticate, I'd say it's moving into robotics/ drones).

ang_cire··on Small models also found the vulnerabilities that Mythos found
That's funny, this is how I've been doing security testing in my code for a while now, minus the 'taint analysis'. Who knew I was ahead of the game. :P

In all seriousness though, it scares me that a lot of security-focused people seemingly haven't learned how LLMs work best for this stuff already.

You should always be breaking your code down into testable chunks, with sets of directions about how to chunk them and what to do with those chunks. Anyone just vaguely gesturing at their entire repo going, "find the security vulns" is not a serious dev/tester; we wouldn't accept that approach in manual secure coding processes/ SSDLCs.

ang_cire··on Code is run more than read (2023)
EVs have gears and gear oil in their drive units. There is a reduction gear that needs lubrication, as well as the differential.

They actually will need oil changes starting anywhere from the 50k to 100k mile mark.

Here's the maintenance guide with pictures walking through changing the oil and filter for the Rear Drive Unit (RDU) in a Tesla Model S:

https://service.tesla.com/docs/ModelS/ServiceManual/Palladiu...

ang_cire··on France to ditch Windows for Linux to reduce reliance on US tech
If it only runs on NT, it'll work better under WINE than on Win10/11.

Legacy app compat is actually an argument for moving to Linux.

ang_cire··on Eight years of wanting, three months of building with AI
> in these sort of "greenfield projects" I work a lot better with concrete prototypes and code in front of me I can dissect instead of trying to endlessly play with designs in my head

I'd really recommend separating prototyping work like that out into a pre-design phase. Do the prototypes and figure out the direction for the actual project, but then come back in with a clean repo and design docs built off the prototypes, for claude to work from. I started out using claude to refactor my old projects (or even my codex ones) before I realized it worked better starting fresh.

I think sometimes it silently decides that certain pieces of code or design are absolute constraints, and won't actually remove or change them unless you explicitly tell it to. Usually I run into this towards the end of implementation, when I'll see something I don't expect to and have to tell it to rip it out.

One example recently was an entire messaging queue (nats jetstream) docker image definition that was sitting in the deploy files unused, but claude didn't ever mention or care about it as it worked on those files; it just silently left it sitting there.

Another example was an auth-bypass setting I built in for local testing during prototyping, being not just left alone by Claude but actually propagated into other areas of the application (e.g. API) without asking.

← PreviousPage 8 of 30Next →