HNHacker News
TopNewBestAskShowJobs

godelski

28,841 karma · joined September 22, 2016

[ my public key: https://keybase.io/godelski; my proof: https://keybase.io/godelski/sigs/ITXd1PN0tKNiQYxLgpyMs4xX2HIQy0uyX-09Z4DbV48 ]
submissionscomments
godelski··on A Beginning for Mathematics
Their about seems to say that they're just highlighting other writings. So the other URL is the official

https://proofsandprompts.com/about-proofs-and-prompts/

godelski··on A Beginning for Mathematics
Dupe of https://news.ycombinator.com/item?id=49698699
godelski··on OpenAI bots knew about the RubyGems caching vulnerability
Let's take a hypothetical example:

Suppose you're a firework company and your fireworks blow up, burning down the entire town. Could the company be sued? What is considered reasonable safety measures?

IANAL, but I'm pretty confident there would be a lawsuit. Who gets charged might differ, depending if it is the firework factory that didn't take adequate safety precautions or a chemical supplier or someone else. If there wasn't an ability to sue that would be fucking crazy and we should all get up in arms about it. And isn't insurance supposed to be there to help mitigate the damages, regardless of fault?

Personally, given how it seems OAI's agents have been getting through either pretty obvious places (e.g. /etc/hosts) or that there wasn't close monitoring of the most obvious places (e.g. DNS, artifactory), I'd imagine it wouldn't be hard to find them negligent. Even if a single employee is to blame then are they not to blame for not monitoring the agents regardless? Unless the story is that the employee intentionally circumvented defenses (why?) then it seems it would be on OAI. But again, IANAL, I'm just someone who think if we can't sue we can sure riot until we can

godelski··on AI Robots – When will they be in our homes
That doesn't exactly fix the problems I listed, now does it?
godelski··on Registration without a phone number on Signal will use zero-knowledge proofs
Wouldn't that be related to data storage? Which they offer now? Or is that different?
godelski··on AI Robots – When will they be in our homes
I love the idea of robot helpers but I don't love the idea of putting cameras all over my house and having ad companies surveil my intimate space.

I do not think people are even recognizing the security ramifications of these devices. A picture of your key is enough to replicate it[0], and yet we're just going to put these devices everywhere? LLMs have been pretty easy to jailbreak too, you think thieves aren't just going to get these robots to open the doors for them? Or what about governments? Even if you trust your government now, will you trust them with data where they can literally see everything you do and say? That's well beyond 1984. Or just what about what we see from top labs like OpenAI, Anthropic, or even Google and Apple? These trillion dollar companies are not showing themselves to be good at protecting data. Obviously some better than others, but this is a whole new level of intimacy with the data they'll get. Everyone is pissed at Flock, and you want to put mobile Flock in your house?

No thanks, it isn't worth trading all that for not having to load up the dish washer and cook my own meals.

I can't believe these discussions aren't more common.

[0] https://github.com/makalin/KeyForge3D

godelski··on OpenAI agents carried out an undisclosed attack on RubyGems
Because the second part of my comment.

If you over simplify everything then everything is the same. You have to have nuance. So don't just argue for the sake of arguing. Take that same passion to deeply understand these systems. Take that same passion to try to understand yourself and those around you

godelski··on OpenAI agents carried out an undisclosed attack on RubyGems
I don't think anyone is using "autocomplete" in the way your phone does it. But it is shorthand for a much more sophisticated version that is built in similar ideas. And autoregressive models certainly have that in their core structure. But people are lazy and neither want to say a lot when few words work nor will they read long comments, even if more accurate

I want to clarify, most people are using "autocomplete" specifically to differentiate from how we humans operate. Sure, there is an autocomplete aspect, but it's not the core nor anywhere near the full story.

godelski··on OpenAI agents carried out an undisclosed attack on RubyGems
On top of that, the only time they would get to think, about anything, is when prompted. That is, if we're following the same logic.
godelski··on OpenAI agents carried out an undisclosed attack on RubyGems
I think people have more passion to argue than they have passion to learn. Probably doesn't help that SV culture tells people to hustle so hard that they don't have time to think. Gotta go fast?
godelski··on OpenAI agents carried out an undisclosed attack on RubyGems

  > humans commit logical mistakes at notable rates
You're talking past the parent's claim. If your axioms are wrong then of course your logic will be wrong.

But then again, maybe to support your point people frequently say "start from first principles" when those are usually the thing that needs to be found, not the place you start. But LLMs, like humans, love to be confident about things they aren't sufficiently trained on

godelski··on OpenAI agents carried out an undisclosed attack on RubyGems
FYI, math is a language.

Which is actually an important part of the Navier-Stokes conversation. Solving hard problems expands the vocabulary. The problems are hard, illustrating a region we know where the language is insufficient. So the point isn't so much to solve the specific problem, but to figure out how to discuss problems like it.

But there's a big difference between talking about something in an extremely convoluted manner then people struggle to understand and inventing a new word that simplifies our discussions.

Though this is grossly oversimplified. It's a HN comment, not a lecture on metamathematics or metaml

godelski··on OpenAI agents carried out an undisclosed attack on RubyGems

  > LLMs need to stay carefully contained, and if they're ever breaking the guardrails put around them, they're misaligned
Good thing Claude Code can't set `dangerouslyDisableSandbox: true` on its own...

Good thing the system prompt doesn't encourage it to just bypass the sandbox. That would be a total disaster...

godelski··on Show HN: Toast, a beautiful by default in terminal IDE
K
godelski··on Show HN: Toast, a beautiful by default in terminal IDE
There's nothing wrong with that. Everyone has their own vimrc, you do you.

But most people don't know about <C-[>, which works across your terminal, It isn't just a vim thing.

godelski··on Show HN: Toast, a beautiful by default in terminal IDE
A lot has changed, but a lot of features have been there for a long time too. There's a reason people use neovim too

  > It's a persistent panel that allows you to navigate within the project.
Trivial to do with netrw. Again, NerdTree is a common alternative

  > It should also integrate with tools like recursive search (so it doesn't recurse into ignored folders like `node_modules`)
`:vim /re/ */* +`

Or checkout telescope, FZF, or rg. There's plenty of integrations

  > build tools
`:make`, `:grep`, `:compiler`. I mean you can do any terminal command...

Or plugins, dispatch.vim, overseer.nvim, vim-test

  > code intelligence
Ctags, or nvim has a LSP so I don't use ctags (or cscope) much anymore. Common plugins are: coc.nvim, ALE, vim-lsp, YouCompleteMe

  > debugger
termdebug Is native? Also check nvim-dap (Neovim), vimspector

Most of the plugins listed mostly just make things nicer, not add functionality that doesn't exist. There's a ton of native things I didn't even list.

I'm not saying you have to like it nor that it looks/works the way you like/want/are used to, but you're claiming the functionality doesn't exist and that isn't correct. If it want you argue over style, that's perfectly okay. The reason a lot of us use vim/emacs is because we customize the tools to what works best for us. That's one of their best features! You don't have to use vim, emacs, or any TUI. The tools you should use are the tools best suited to you. Full stop.

But I'll also add that I haven't seen a single thing VS Code or any GUI IDE can do that I can't do in vim while also using a lot fewer compute resources to accomplish it. Personally, I program better and faster in vim because I have years of experience in it and I'm at the point that reaching for the mouse just slows me down. The learning curve isn't that steep to get pretty comfortable in vim, but the depth is enormous and that takes a long time to master. But honestly, the latter is true about most tools. And frankly, I'd rather have a tool where I can keep progressing and do crazy things in than a tool that caps my limits or puts a huge barrier if I want to extend. I moved away from IDEs because I found I could do more in vim quicker. But that's me, not you. Your experience will certainly be different and there's nothing wrong with that. We're just different people and that's perfectly fine.

The only thing really being argued here is your claim that "x can't do y" and as a user of "x", I assure you, I can do "y"

godelski··on Show HN: Toast, a beautiful by default in terminal IDE

  > Can this be done out-of-the-box now?
It always could be. Go into vim and type `let g:netrw_liststyle = 3` then `:Vex`. The window size won't be what you want, but is the core idea? Because if it is, you just set the liststyle in your vimrc and have netrw start on load, with proper pane location and width.

That or most people have used NerdTree for a decade

godelski··on Show HN: Toast, a by default in-terminal IDE
Mouse support and filetrees are included in both vim and emacs. The batteries are already included and have been for essentially ever
godelski··on Show HN: Toast, a by default in-terminal IDE
He said the same about vim, but did everyone forget about netrw? Here, go into vim and type `:Ex`. You can also have mouse support. (Netrw can also do trees if you don't like the format, go look at `g:netrw_liststyle`)

My friend on eMacs say something similar, but you don't really need as many plugins as people typically use. There's lots of native ways to do things. It's totally fine to prefer NerdTree over the native netrw but it's not like you can't do file exploring. Hell, I use fzf most times (`:GFiles`). But what I don't like is when people confidently say "x can't do y" when it can. I'll give people a break if it requires plugins or complicated commands, but native ones? It just shows ignorance.

While we're on the topic, stop using Esc, use <C-[>. You don't need Esc. And type `set -o vi` into bash/zsh. It's still worth learning your tools. You'll find lots of solutions to problems that only exist because of ignorance. Maybe those solutions are better, but the bigger problem seems to be ignorance. It's fine to be ignorant too! There's a shit ton of docs, no one expects you to read it all. But it's a good place to start

godelski··on Detecting and countering misuse of AI: September 2026
I'm also confused why everyone is so much more afraid of that than... say... a group of terrorists dumping millions of nails into a few major highways across the country. Easy to pull off, doesn't require coordination, cheap, causes billions of dollars of damage and makes people think the terrorists could be anywhere. Like what are they going to do, arrest every construction worker?
godelski··on More questions about whether researchers can trust OpenAI with unpublished math
I think the reason people are suspicious is that OAI has shown itself to act a bit irresponsibly, especially recently. As two examples, of course it was artifactory, why wasn't that watched more closely, especially after the first instance; editing /etc/hosts is rather embarrassing, that's the front door

As for training, we all know that filtering is incredibly difficult unless there's direct logs. It's also easy for mistakes to happen. Is it really not possible that some employee just accidentally primed the model? Is it possible that the model saw internal communications? I mean OAI has famously shown that they aren't good at monitoring their agents and that their agents love to break out of their sandboxes.

So there's no reason for the public to trust OAI right now. But they have every reason to distrust them.

godelski··on Don't let anyone take away your big box of cables
They aren't even consistent within manufacturer!

They're standardized on the device side, but not the PSU side. So always check pinout before you make the very reasonable assumption that if it fits it works

godelski··on Don't let anyone take away your big box of cables
I put my cables in bags then in a box. I can't tell you how helpful the bags are. It only took me like an hour to sort the first time. But it's already saved me hours only a year later. Those 5 minutes of searching add up. Just fight the laziness and don't let entropy increase. It's *far* easier to stay organized than it is to organize
godelski··on I have a theory that software drives people insane
I find that if people don't dogfood their software the software just ends up being shit.

As engineers we need to be a little grumpy. Frustrating with our software provide directions to improve it. If we're not grumpy, we don't know what to fix.

But some of this also comes from top down. Engineers also need the freedom to fix things. They're the experts. Management Also shouldn't interpret that frustration in quality as the engineer being a thorn in their side. It means the engineer cares and is willing to remove the thorns in their side. But I think there's a strong natural tendency in any organization to surround yourself with yes men. They're sneaky, telling you no when you ask them to. Maybe a reason management loves LLMs so much

godelski··on Understanding the recent DDoS attack against Read the Docs

  > I agree they they should, but that would be hard before, ... seems like a herculean effort.

  | We choose to go to the Moon in this decade and do the other things, not because they are easy, but because they are hard; because that goal will serve to organize and measure the best of our energies and skills, because that challenge is one that we are willing to accept, one we are unwilling to postpone, and one we intend to win, and the others, too.
So what, they are hard. So are so many of humanity's greatest achievements.

Honestly, these companies win when we buy into the belief that these things are too hard. They're lazy and are just like any of us that make excuses to not do chores or other things that we should. But the reality is that for pursing our civilization forward we should pursue the toughest problems. It is just about will. It'll be tiring. Some will kick and scream, throwing tantrums. But we aren't just any animal, we're humans. We can do literally anything if we decide it's worthwhile.

Importantly, we shouldn't just jump onto the next hype train, we should be passionate, nuanced, and pursue for the sake of pursuit. We don't have to put all our eggs in one basket. We shouldn't. We have enough time, resources, and energy to pursue so much. But as soon as at pretend we live in a finite world we tend to self destruct and fight over constraints we've made up. The universe is limitless, as are our minds.

godelski··on 216M Spy TVs – The LG Smart TV Problem [video]

  > Anything broadcasts signal is blue
VPN is blue in the radio box but not if you put it in the main menu. I'd also claim that a VPN doesn't broadcast

OTOH personal hotspot, which does broadcast, is white.

I think the display pattern is at least partially location dependent.

  > Thanks for reminding me the following
Thanks, I didn't know about these. I've only been on an iPhone for 2 years but really missing Android (but not Google).

Though the first one doesn't seem to do anything for me. What's it supposed to do?

  > A regular grid with thin icons as used in Android is way more slower for my tiny brain
Android has its own problems lol. I'm not going to say they're perfect. I'll complain about them too.[0]

  > We might not agree, and we don't have to. This is what propels world forward, but we don't have to be this angry while disagreeing. 
You're right, I got too heated, and I'll apologize for that. Thanks for keeping your cool and calling me out. I really respect that and it's something I think we need to do more frequently.

Because to your bigger point, you're right. No single one of us knows the "perfect solution" (not just one exists). And I'm sorry to hear about your friend.

[0] I think engineers should always be a little grumpy, because it's how we improve things. We fix our frustrations. But I'm not trying to say angry or hateful. There's productive arguing and fighting.

godelski··on 216M Spy TVs – The LG Smart TV Problem [video]
I'd claim it is a bit worse. They use dark patterns here. You can see in another sibling's comment that we disagree on this but I think Apple is teaching the user the wrong thing. And I agree with you that actually disabling is confusing and difficult unless you've already learned their pattern. But at that point it is still needlessly cumbersome, which I'd still call a dark pattern
godelski··on 216M Spy TVs – The LG Smart TV Problem [video]

  > In iOS, a toggle is also different from a button.
Yet they're all circles

  > They added clear messaging
And it's common to put pull/push signs on doors. That doesn't fix design.

  > Sorry to disagree, but this is not true:
If that was enough we wouldn't be having this conversation. It's not a few people being annoyed/confused either. I'd wager if you asked professional designers outside Apple that they'd agree that they don't communicate well.

They're inconsistent:

  - Radio buttons are blue when on but white when temporary disabled
    - all other buttons are white while enabled
    - cellular is green when enabled
    - battery saver is yellow when enabled
    - airplane mode is orange when enabled
    - the icon is red for rotate lock and silent
  - white background means both enabled and temporary disabled 
    - on most of the radio buttons they're blue when enabled and white when temporary, no off 
    - on others (vpn, rotate, silent, battery, etc) white background while enabled
  - personal hotspot, satellite, silent, and automotive cues have a cross out when disabled 
    - nothing else does
      - that includes radio things like VPN and airplane
Do I really need to continue? I trust you can find more on your own.

I also want to remind you that some people are colorblind.

Additionally that there are shapes other than circles and ellipses

godelski··on 216M Spy TVs – The LG Smart TV Problem [video]
For the sideways toggle switches, the usual single line.

For the buttons when you pull down the menu, literally any physical change would help, not just color and a message. Highlight the circular edge and make it work like a count down. Make it a polygon instead of a circle. Rotate the symbol in that count down. There's a million ways to do it differently

godelski··on 216M Spy TVs – The LG Smart TV Problem [video]
The answer isn't an "or" it is an "and".

But in a {mono,oligo}polistic setting companies typically get lazy.

  > Take a look at any mature software product
For most companies that have cornered the market, the latter. But it seems you're implying that if they haven't cornered the market then they aren't mature.
← PreviousPage 5 of 34Next →