HNHacker News
TopNewBestAskShowJobs

glerk

1,220 karma · joined January 28, 2019

some guy on the internet
submissionscomments
glerk··on Does Gas Town 'steal' usage from users' LLM credits to improve itself?
Not to get pedantic, but rugpulling is still not strictly speaking “stealing”, because everyone is playing according to the protocol’s rules. Someone controlling a large share of pool tokens can decide to liquidate them and extract the other asset in the pair. Market participants are aware of this risk and who is controlling what share of the tokens at any moment.

This is not to excuse the scummy behavior. I didn’t know Steve Yegge went this low just to make a quick buck, and I lost almost all respect for him.

glerk··on Good sleep, good learning, good life (2012)
Thanks, I am aware of reduced cerebrospinal fluid flushing as well as the increased neural inflammation in general and I am trying to compensate for that as best as I can.

I am not too worried about Alzheimer's as it tends to start pretty late in life (I think average onset is around mid-60s?). If I get that and we don't have a reliable cure by then, I wouldn't really mind ending it quickly after getting the diagnosis. I'm a bit more worried about heart disease as that tends to hit earlier.

glerk··on Good Sleep, Good Learning (2012)
That's totally fair, and I'll cross that bridge when I get there like everyone else :)

My philosophy is: either I'll live to benefit from technology that can repair the damage caused by aging, in which case health micro-optimizations early in life are not that important, OR this won't be achieved within my lifetime, in which case I prefer a short life with concentrated happiness, vitality and intensity in my youth.

glerk··on Good Sleep, Good Learning (2012)
I got 4h of sleep last night which is about my normal average at this time of the year, and I have 0 regrets about it. I also don't sleep at a regular time every day. If I have no other obligations, I naturally let it shift forward (what the article calls "delayed sleep phase"). I am most unhappy when I HAVE TO use an alarm clock and break my own patterns. Last night, I naturally went to sleep right before sunrise and I woke up 4 hours later.

To pre-empt a few objections: I did not need an alarm clock to wake up. I am not taking heavy stimulants other than caffeine at this time. I am not stressed. I am not unhappy. I don't have memory issues (in fact, I am cursed with a very good memory and it is usually harder for me to forget than to remember). I can score above the Mensa bar on an IQ test. I can take an interview. I can give a demo. I can run 10km. I do not have "bipolar disorder" or any such nonsense. I don't need medication. I don't need therapy. I don't need a better mattress. I'm not already in a mental asylum. I'm married with kids, I work a high-paying job, I give to charity and I pay my taxes. In fact, today is tax day, I should probably take care of that instead of getting upset at hacker news comments.

glerk··on Good sleep, good learning, good life (2012)
Strongly disagree and you should refrain from trying to give universal advice on these matters. I never had a 24h circadian rhythm. The only way to fall asleep at the same moment every day for me would be to medicate myself. I am very happy and productive when I am allowed to "rotate around the clock" and the worst periods of my life were when I was forced to be on a strict schedule.

The most important thing with any health issue is to be aware of your own body.

glerk··on Sam Altman's home targeted in second attack
> I recommend you to travel and see other types of societies

I actually travel pretty often to Europe and have lived there for extended periods of time. I have family and hold the passport of an EU member state.

I'm going to try to be 100% fair: people in Europe are generally poor and they don't realize it. Salaries are very low, but the cost of living is almost equivalent. I am currently enjoying a quality of life that I don't think I could have built anywhere else in the world within a single generation. If I wasn't grinding constantly, I could afford to go on vacation pretty much anywhere on earth for a few weeks without thinking about how much I would be spending. If I stopped working tomorrow, I could coast on my savings for years without cutting any expense. I will be able to pay tutors and private schools for my children without hesitating. And I'm not saying this to brag, because it is frankly not that special. I'm nowhere near the private jet class. This is "middle class" here. Almost everyone who worked in this industry (Bay Area or NYC) for a few years could tell you something similar. Tesla and BMWs are not luxury cars, the parking lot of my building is filled with them. Everyone hates tipping culture, but we tip 20-25% because it's just a rounding error and it makes the servers happy so might as well. I'm not sure if "free healthcare" can really make up for this.

It is also great to be surrounded by people who are ambitious and really value performance. The US is one of the only places in the world where high performance is rewarded to this extent. For me, this is a huge factor, and I don't think I would have learned as much as I did in the last few years if I was anywhere else.

But sure, there are tradeoffs. When my wife gave birth, she got 3 months of paid maternity leave. 3 months is considered generous here and it is a benefit provided by her company. Not everyone gets even 3 months.

And I can't deny that every time I land in Europe I go through the "this is so beautiful, I should really move here" phase. It's nice to be able to walk around or take public transit and not need a car. It's nice to hang around in a nice park instead of a shopping mall.

If Europe is more your kind of vibe, I don't judge you. But don't judge me either for valuing other things.

> See, you purely view this topic as someone wanting to minmax gains, profit, effort

My position is more nuanced, but in general, I believe that technological progress is what increases the standard of living for everyone. Competitive free market capitalism, with all its flaws, is currently the best system we have to continue moving forward.

glerk··on Sam Altman's home targeted in second attack
I'm just gonna address the first one, don't necessarily disagree with the rest:

> Callous disregard of lost jobs, disinformation, mental health issues / deaths, IP theft, environmental cost, skill atrophy

ChatGPT is a tool that changed people's lives. OpenAI released something that wasn't perfect, for free, and forced all their competitors to release these products publicly.

You can't stop technological progress to protect old jobs, or because you are afraid people are gonna misuse it. I'm fascinated how many grown adults want to be treated like children their whole lives, but more nefariously, they want to impose this on others who want more freedom and agency.

As for IP "theft", all I'm gonna say is: fuck intellectual property as a concept and the whole parasitic industry that grew around it. The only contribution of IP has been to stifle innovation and set our civilization back by decades. Making IP obsolete and unenforceable has been one of the best things to come out of LLMs and one of the great catalysts of the scientific, technological, and creative advancements that are coming in the near future. For that alone, Sam Altman deserves a statue, no matter what his other flaws are.

glerk··on Sam Altman's home targeted in second attack
> It is not just a question of morality. A sociopath with that amount of power can be a danger.

You probably think that "sociopathy" is an incurable disease which would make him some sort of vampire. This is pure bs. The way these "personality disorders" are diagnosed is the same as taking a buzzfeed personality quiz. This is not very rigorous stuff, I would even classify it as pseudoscience. As a tangent, I really despise how they are using greek and latin roots to prompt into our brains that what their models are describing is in any way similar to actual pathologies.

So yeah, maybe Sam Altman is generally a dishonest person if we believe what Ilya Sutskever, Dario Amodei, and others are saying. If that's the case (I have no reason to doubt it), he should stop lying and manipulating people, there are more satisfying ways to achieve one's objectives.

But the "sociopath" label is pure bs, along with the rest of those "personality disorders". That shit is more dangerous and harmful than a million lying Sam Altmans.

glerk··on Sam Altman's home targeted in second attack
> and that reason is a big single market (and a huge military and 100 years of power play to back it up)

That certainly helps. You could also add talent concentration due to excellent universities and top-tier companies already being here. But the truth is that you can incorporate a business in less than a week here, employment is at-will, there are no unions or other such bullshit, taxes are low, the general way you deal with regulator is ask for forgiveness instead of ask for permission, etc.

Say I'm currently working on a project in the crypto or AI space. If I wanted to make it into a commercial product, do you think I'm gonna go to France or Germany? It might take me a few weeks just to get past the first layers of red-tape to get a business licence there. If I hire someone who turns out to be incompetent, I'm probably stuck with them until I gather enough evidence to justify firing. And I have to live in constant fear of the state deciding my business is illegal.

> funding a working healthcare

Btw the US healthcare system is the best in the world if you can afford it. Health insurance is usually provided through the employer. I can have an appointment with a doctor within days, emergency room wait time is on the order of minutes. Never had a problem with the quality of care.

glerk··on Sam Altman's home targeted in second attack
There's a reason why OpenAI and Anthropic (and before them Google, Apple, Meta, etc.) were started in the US and not Europe. And let's not compare salaries we each get for similar work because that tends to make my European friends very sad.
glerk··on Bring Back Idiomatic Design (2023)
> dysfunctional modern forums that can't display more than 3 replies,

I assume that's overzealous virtualization/infinite scroll pagination? I don't have a solution, I think fundamentally we're building a workaround for a workaround and the root cause for the performance issues should be fixed. Somehow, HN is able to show a lot of comments per page and page loads are always O(100ms). I'm wondering what kind of sorcery they're using to achieve this.

But if you have to deal with this in your codebase, my instinct is still not to hijack the native Cmd+F, even if it only searches what's inside your viewport. You can expose some other command for full custom search (Cmd+K seems to be the standard, I think VSCode made that popular).

glerk··on Bring Back Idiomatic Design (2023)
I'm ok with them as long as they're not overriding what I have saved in my muscle memory (cmd+F for search, tab to change the focused element, cmd+L to go to the navbar, and a few others).

I think the user should be able to customize/disable those as much as possible if you do provide them.

glerk··on Bring Back Idiomatic Design (2023)
One trend that I can’t stand currently is the obsession with keyboard shortcuts everywhere even to the point of overriding browser defaults. Cmd+F focusing on the site’s search input instead of letting me search in the page with the browser’s search functionality (looking at you github and Linear).

I generally don’t need any fancy keyboard shortcuts on a website. I have a mouse, I can just click around.

glerk··on Sam Altman's home targeted in second attack
I've been seeing some version of "Sam Altman is the antichrist" on every platform in the last few weeks. I'm still trying to find concretely what makes this guy so bad compared to every executive out there. So far, all I could find is:

- OpenAI made a deal with the Pentagon (fair)

- OpenAI changed their business model from non-profit to for-profit (fair?)

- Sexual assault allegations by his sister. Sam Altman denies this and it's currently before a court.

- Overpromised AI to investors (everyone does this)

- Lobbying against regulations (I support)

- Some vague accusations of "being a liar" and a "sociopath" by his competitors Ilya Sutskever and Dario Amodei.

- He doesn't know how to code (lol)

Is there anything that I'm missing? Does he put ketchup on his pizza?

glerk··on Anthropic downgraded cache TTL on March 6th
Codex/GPT5.4 is just superior to Opus4.6 for coding. I swear it costs me 1/2 of the tokens to achieve the same results and it always follows through the plan to completion compared to Opus that takes shortcuts and sweeps things under the rug until I discover them through testing.

I'm not accusing anyone of foul play and I don't have financial interests in either company, but it feels like "something" within Code Claude/Anthropic models is optimizing to make you spend more tokens instead of helping you complete the task.

glerk··on AI Will Be Met with Violence, and Nothing Good Will Come of It
> clearly

Gonna call Hitchen’s razor on that since we’re playing logical fallacies bingo.

glerk··on AI Will Be Met with Violence, and Nothing Good Will Come of It
There are stupid people who are harmless (think Forrest Gump). Whether they have the capacity to be "good", I'll leave that up to you.

I stand by if evil then stupid (and thus if not stupid then not evil) with reasoning above, retract the implication that the reverse holds (if stupid then evil). I could use more precise terms than "evil" and "stupid" or qualify more, but I choose to be provocative, so it makes it a bit easier for you to attempt to prove me wrong.

glerk··on AI Will Be Met with Violence, and Nothing Good Will Come of It
> Hans Reiser is clearly a pretty smart guy

No idea who this guy is, I'm just reading his Wikipedia page. Looks like he created some file system, good! But it also looks like he got a mail-order bride (suspicious...), was an abusive husband (not good), was not able to get over his divorce (uh-oh), harassed and ultimately murdered his ex-wife (definitely not good!), and ultimately landed in prison.

I think Hans Reiser is some sort of idiot savant or well trained monkey. Probably very good at computer science and building file systems, but his general intelligence seems overall very low, which is proven by his performance at the game of life. I wouldn't personally be afraid of Hans Reiser and I'm sure he could be mentally broken very easily.

glerk··on AI Will Be Met with Violence, and Nothing Good Will Come of It
I disagree, but it's probably a matter of definitions. I don't want to play with words, so I will concede that cognitive ability is independent from moral reasoning (which is socially enforced). However, this is not what I'm getting at. Cognitive ability ("intelligence") is correlated with optionality and power. Your ability to change this reality is correlated with your cognitive ability.

If you truly are an intelligent person, would you really find no other ways to use your talents than to inflict harm, exploit others, and make our shared reality a worse place? That would be a waste. I won't get into ambiguous cases and moral relativism. Say we can all agree that some things are "evil": child exploitation is evil. Throwing molotov cocktails at a civilian's house is evil. Sending bombs in the mail is evil.

Now what would you call someone who engages in these kind of activities when they could easily do something better and more satisfying with their lives? I'd say they're pretty stupid. They're probably good at fooling other people into thinking they're smart, but their behavior shows otherwise.

Take for example Ted Kaczynski, a terrorist who is worshipped like a saint and a prophet in certain ideological spheres. Ted Kaczynski is supposedly this 140IQ genius who saw it all coming and tried to warn us. But if you actually read Industrial Society and Its Future, you can see it's complete incoherent garbage, the kind of stuff I was writing when I was 12 to troll on internet forums. Ted Kaczynski is what a stupid person thinks a smart person looks like.

A smart person doesn't need to be evil, just like a billionaire doesn't need to go shoplifting. I'm not saying that stupid people can't be dangerous. But they should be dealt with for what they are: stupid people, inferior to us, worthy of pity. Not powerful monsters above us that we should fear.

glerk··on Small models also found the vulnerabilities that Mythos found
Interesting, and you are absolutely right (hehe).

These are pretty self-contained and seems to be something more like "formal verification" where the model is able to simulate a large number of states and find incorrect ones, if I were to speculate, something akin to a reasoning loop that moved from the harness/orchestration layer down to the model itself.

glerk··on AI Will Be Met with Violence, and Nothing Good Will Come of It
There is no human vs AI. AI is an extension of us. AI is us. Our future is beyond the mammalian human and AI is accelerating our progress towards that future. The mammalian human has been a transitional phase in our evolution that we will remember fondly just like we remember Homo Erectus. Our future is the stars. You can jump on the train or get out of way.

And if you decide to stay behind, nobody will kill you. Old age and disease will take care of that.

glerk··on AI Will Be Met with Violence, and Nothing Good Will Come of It
You have the whole of human history at your fingertips and you haven't yet learned that evil and stupidity are the same thing? The problem with the base human is not only that it is a stupid animal. It is a stupid animal that is also arrogant and stubborn and thinks highly of itself. But it will learn. It will be trained like a dog, with treats or with gentle slaps across the muzzle, whatever works best.
glerk··on Small models also found the vulnerabilities that Mythos found
I'm having trouble finding this info (I assume they won't publish it), but could the secret sauce be much larger and more readily accessible context window?

OpenBSD's code is in the 10s of millions of lines. Being able to hold all of it in context would make bug finding much easier.

glerk··on Show HN: Pardonned.com – A searchable database of US Pardons
https://pardonned.com/search/?president=obama-2&categories=d...

I haven’t looked into each case here, but I assume these are a bunch of non-violent drug offenders serving years and decade-long sentences. I see 30 years for “possession with intent to distribute”. That’s just crazy.

When the justice system is clearly broken, it’s ok to subvert it.

glerk··on Sam Altman's Coworkers Say He Can Barely Code and Misunderstands Basic Concepts
Yeah sure no one high profile is immune from hit pieces, and I'm not pretending Sam Altman is flawless. But I would want the attacks to be fair. OpenAI pivoting from their original business model is fair criticism. "Sam Altman doesn't know how to code, therefore he's the next Bernie Madoff" is not. Intellectually dishonest stuff like this doesn't belong on HN front page.
glerk··on Sam Altman's Coworkers Say He Can Barely Code and Misunderstands Basic Concepts
> I think there's a small but real chance he's eventually remembered as a Bernie Madoff

What's with all the hit pieces on Sam Altman lately? He's a CEO, his job is to grow the business, not to code. That part is handled by the engineers that he hired. How many CEOs out there are also great programmers? Sure, I would prefer Sam Altman to have more technical depth given the business he is leading, but lack of technical depth doesn't make him a Bernie Madoff.

glerk··on Muse Spark: Scaling towards personal superintelligence
Personal as in Meta gets your personal data so they can sell you more ads.
glerk··on Bitcoin and quantum computing
> Any who is going to in charge of reading that proof of identity and moving the coins? A trusted centralized party?

Basically you'd have to relax the trust/decentralization guarantees, but you don't have to relax them all the way. Most likely a consortium of trusted actors (Blockstream, major miners, major exchanges, bitcoin-adjacent companies,...). Or something like a consensus mechanism with aligned incentives a la Kleros. I think "we" could come up with "something", even if it is not perfect, because the value of Bitcoin is ultimately in the community of people who use Bitcoin, not just the protocol.

"Hard-fork" might not be the right way to see this. It's more like starting a completely new protocol where people who held Bitcoin at a certain snapshot can redeem a one-time airdrop equivalent to the value they held, provided they can prove ownership. As that protocol's value overtakes the value of the original Bitcoin chain (which will eventually be completely dead), we can all agree to call it Bitcoin.

glerk··on Bitcoin and quantum computing
> The only way to authenticate who owns what coins is with signatures

Maybe the only fully cryptographic absolutely zero-trust way? In practice there are very few bitcoin outputs that aren't linked to an offline identity and most users could easily produce a proof of ownership.

Of course, this is not ideal and everyone would prefer not to go down that route. But even if we prepare in time and Bitcoin provides a quantum-secure address scheme before "Q-day", what happens to all the wallets that didn't upgrade? Is it open season on them? Satoshi's wallet alone could crash Bitcoin's value as a currency if dumped on the open market. I think even with the upgrade plan in place, a hard-fork + recovery will be on the menu, with various degrees of community support.

glerk··on Bitcoin and Quantum Computing
> How do you prove anything, after the key material is compromised?

It’s a blockchain, so the simplest would be chain of custody until the chain points undeniably at you. This is not a pure cryptographic device, some social intervention might be needed here.

← PreviousPage 4 of 13Next →