Completely agree. Consumers won’t pay for anything online, which means every business model is user-hostile. Use the web for five minutes without an ad blocker and it’s obvious.
Atlas may not be the solution but I love the idea of an LLM that sits between me and the dreck that is today’s web.
Complete unrelated to what you actual wrote about, but... This is the second time in a week that I hear dreck used in english. Before, I never noticed it anywhere. First was from the "singing chemist" scene in Breaking Bad, and now in your writing. I wasn't aware english adopted this word from German as well. Weird, that I never heard it until just now, while the scene I watched is already 15 years or so old...
As another commenter noted, it was loaned from Yiddish rather than German, although the two languages are very closely related. There are many Yiddish words in English that have come from the Jewish diaspora. Common ones I can think of are schlep (carry something heavy) and kvetch (complain loudly). Since this is Hacker News, the Yiddish word I think we use the most is glitch. Of course there are also words from Hebrew origin entering English the same way, like behemoth, kosher, messiah...
Well, while I can never be sure about my own biases, I submit this particular instance is/was not an "illusion". I truely never heard the word dreck used in english before. Why do I know? I am a german native speaker, and I submit that I would have noticed it if I ever heard it earlier, simply because I know its meaning in my native lang and that would have made spotting it rather easy.
I also believe noticing Baader-meinhof in the 90s is rather unsurprising, since RAF was just "a few years" ago. However, "dreck" as someone else noted is documented since the early 20th century. So I dont think me noticing this just recently isn't a bias, rather a true coincidence.
I think we should note that this is a situation created by the businesses more than the consumers. By offering "free" products they drove paid products out of the market. They didn't have to do that. But if I'm going to be as fair as possible, it's only regulation that could have stopped such a niche feom being exploited. One business or another would have eventually figured it out.
Why would you trust an LLM to make the right decisions on your behalf? You think it can't be tricked by dark patterns? LLMs can't even follow direct orders much less detect deception or ill intent.
I guess you did say "the idea of" not "the reality of".
It depends completely on the term. You can make a great case that we've already reached AGI. You can also make a great case that we are decades away from it.
> Given the power dynamics, the burden of proof is on Shopify to prooave it wasn't retaliating at the behest of, or in a misguided attempt to defend DHH's honor.
That’s just a way of saying “I don’t have any evidence of what I’m claiming”
> Charts like this are misleading because they don't take age into account. It's not really noteworthy that old people die of heart disease and cancer.
I think this is the whole point of the article. The news does not cover reality as it is, it selects information that is noteworthy and drives clicks/views/engagement/ad revenue.
This is why the news has been shown to increasingly misrepresent reality:
It also has to do with “deserving” death, or injustice. Someone who is obese dies of a chronic illness, or a smoker, etc. doesn’t register as news, or even the cause themselves, because the vast majority of obese people and smokers know themselves that their lifestyles lead to illness and early death.
But dying from a criminal act? It’s undeserved and arguably more easily preventable than grand lifestyle changes across the whole population. If a felon with 50+ arrests murders someone, a “quick” adjustment in laws could prevent it in the future
>It’s undeserved and arguably more easily preventable
is it though? Crime has been with us since the dawn of civilization. It's easier to tell a story in which crime is personalized and framed as preventable but in reality there's always new modes of crime, new criminals, always the incentive for people to steal when they can, and so on.
When societies manage to "stomp out" crime they're no less brutal than when they attempt to stop a pandemic. I think what a society frames as aberrant is just a reflection of the kind of public morality they endorse. A society of pirates probably thinks scorbut is more undeserved than being punished for theft.
> When societies manage to "stomp out" crime they're no less brutal than when they attempt to stop a pandemic.
I don't think this is necessarily true. Because when trying to stamp out a pandemic, everybody was negatively affected. When trying to stamp out crime, it's overwhelmingly suspected criminals that are affected. Not long ago El-Salvador had literally the highest homicide rate in the world and was just a domestic war-zone. It's now rated as the 8th safest country in the world [1].
What did they do? Dramatically bump up the penalties for gang membership, round up gang members, and throw them into militarized prisons. Their President now has a 90%+ approval rating, and is one of the most well regarded leaders in the world. Obviously there's a 'First they came for the Communists, And I did not speak out, Because I was not a Communist...' type concern here. But in this case, it seems that first they came for the criminals, crime plummeted, and everybody lived happily ever after.
And again there are also completely reasonable human rights concerns, but the thing I think people often disregard in this calculus is the human rights of the literally 99% of people who were previously being terrorized, killed, robbed, and so on by the 1% of people. They have obviously rounded up some people who are innocent and they are making efforts to resolve that, and I hope those who are truly innocent receive fair compensation for the distress. But if one views human rights as a net (gains, versus losses), then El Salvador is vastly more humane than it was in the past.
Uh, El Salvador? The place where we just sent 200 largely innocent Venezuelas (and unknown numbers of others) to be tortured and raped extrajudicially -- a miscarriage of justice that Bukele was fully aware of and openly joked about on TV? The country that trotted out Kilmar Ábrego García for a dog-and-pony cocktail hour to pretend that he wasn't being held indefinitely under brutal conditions and without access to legal council? The country that pushed to impede a US probe of actual MS-13 gangsters due to likely corruption in the Bukele government? *That* El Salvador is our shining light?
Yeah, I guess crime is trivial to fix if you automatically get a life sentence at the gulag for having a tattoo or mouthing off at an authority figure (for certain definitions of crime).
A country headed by devils cannot, by definition, be humane.
If people were being imprisoned for "having a tattoo or mouthing off at an authority figure" there wouldn't be 90%+ approval, and way more of the population would be imprisoned. It's easy to create a lawless dystopia. It's also easy to create a 'over-lawed' authoritarian dystopia. Neither tend to be much liked by the population. The balance is in cracking down hard on criminality while avoiding overstepping that into complete authoritarianism. I think one of the easiest ways to measure this is simply by looking at approval ratings and so, for now at least, El Salvador is very successfully maintaining that balance.
---
Actually you just sent me down a rabbit hole watching videos of modern El Salvador. Here [1] is a great one with some dude just interacting with the locals, and discussing the changes. Lol, I seriously want to go there now. It looks like an amazing place to visit.
My wife went to El Salvador a few years back. She did not talk to men: men do not talk to animals in that country, that would not be a masculine thing to do. She talked to a lot of women. You get an entirely different view of society from the inside, and it rarely matches the view some American man with a film crew can get.
No he didn't, at all. AFAIK there were no formal approval polls at the time, but Stalin himself acknowledged his 'popularity' with his rather famous quote, "I know that after my death a pile of rubbish will be heaped on my grave, but the winds of history will sooner or later sweep it away without mercy."
And indeed as soon as he died there was a widespread movement of 'De-Stalinization.' For some more fun context - this is also the era that Gorbachev grew up in which is probably what planted the seed in him for disdain of the Soviet System that he would eventually play a key role in collapsing.
It is well known that Stalin enjoyed widespread popularity in the Soviet Union, even during the Purge years. But fine, substitute Putin if you wish. Russia is a country riddled with crime and corruption, completely undemocratic, with journalists and political rivals getting murdered left and right, and mired in a cataclysmic war with a once-friendly neighbor. Yet Putin polls extremely well.
Entrenched authoritarianism does not tolerate dissent. This should not come as a surprise. And I don’t have my citations on hand, but I’ve read a number of articles where Salvadorans express the same sentiment: ratings are inflated because people are scared to speak out.
I’m sure this is more than acceptable to some of those not currently in the government’s crosshairs. Unfortunately, there is no off-ramp for this flavor of authoritarianism. Bukele will remain dictator and El Salvador will no longer be a democracy.
The poll I referenced on perceptions of criminality was from Gallup. Approval ratings are typically carried out by groups like Cid-Gallup, which is based in Costa Rica. In other words - foreign organizations. And you can also see countless videos on YouTube of people, including Americans, treking the streets of El Salvador and talking to random people about the changes that have happened. Nobody's holding back anything. People love it. This is real. You're obviously going to be able to find stories against it online because it directly contradicts the globalist and Progressive ideology on how to handle criminality. And there are countless groups backing those sort of ideologies with vast sums of money fueling them.
The effects in El Salvador have been so dramatic that you can immediately reject most tales that try to undermine it on a macro level. For instance if it were true that they were only capturing small fries while letting the big fish run unchecked, you wouldn't see this dramatic decline in criminality. The gangs are being completely dismantled. If the gang leaders are or were working as collaborators then that's great! Make people even less motivated to even think about joining gangs when the leaders themselves might be snitches.
I could go into detail about why Putin is popular, but it requires treking back to the 'wild nineties'. I'd love to discuss that if you're interested, but I think it's somewhat tangential to what we're talking about.
Anecdote: brother’s brother-in-law is Filipino, seemed more than happy with Duterte's crack-down on drugs (he attributed to weed the kind of mental breakdown the USA would expect from meth). Duterte's approval rating while in office was +45 to +81, so I think that support was representative and not merely a fluke.
Duterte was arrested this year by order of the International Criminal Court.
I want to add something here about Duterte's current poll ratings, after his arrest, but I have no idea which news sources are actually reliable. The first few "sources" I tried looking at had both AI-generated adverts (the "look at this gentle bear climbing on beds in the maternity ward!" kind) between each paragraph, and pre-AI slop (kawaii IQ tests) on the sidebar, so I don't trust them.
Gang violence among gang members is a life style choice. For children involved in gang violence it’s much more problematic. An adult who freely chooses to make their living on the street is a bit less unnerving as the lifestyle can lead there. When an innocent person is shot during gang violence, it is much more newsworthy.
I would also like to stop gang violence but this often means “throwing the book” at gang members, which is often disliked by many activists.
I myself live in a safe area of a major city, and there are gang murders in my neighborhood occasionally. It makes my relatives and friends ask how I can live here. But a grown man shot in his car at 3am over a drug deal doesn’t make me feel that less unsafe, and I have kids here
"Gang violence among gang members is a life style choice. For children involved in gang violence it’s much more problematic."
Gang membership is skewed younger and often includes "children" (depends on definition) 14+. Makes it a little tricky about lifestyle choice when dealing with minors.
The problem with "throwing the book" at gang members is that it doesn't work.
Nobody joining a gang is making a rational reckoning of the risk/reward of getting caught by the police, partly because they don't plan to get caught and partly because the much larger risk is getting killed.
And the people getting arrested and prosecuted are primarily not the people calling the shots or driving recruitment of new members.
The best way to put a dent in gang violence is to disrupt gang recruiting, and one of the better ways to do that is to improve societal safety nets so joining a gang is less attractive.
> "The problem with "throwing the book" at gang members is that it doesn't work."
Then how would you explain El-Salvador? They went from the homicide capital of the world in 2015 to the 8th safest place in the world [1] in less than a decade. And "all" that they did was dramatically bumped up the penalties for gang membership, round up gang members, and threw them in militarized prisons as opposed to the typical gang reinforcement retreats. Crime dramatically plummeted and you ended up with a president with an approval rating upwards of 90%.
From my perspective in modern times we've trialed both soft and hard systems on crime. The soft systems in general have had very poor results except in places that already had no issues with crime (e.g. Norway), whereas the hard systems have demonstrated phenomenally positive outcomes. Places like the US have a major problem with things like privatized prisons that create a commercial incentive for incarceration, but I think these are tangential to the topic.
That entire article is about whether people "feel" safe. It doesn't seem to have any stats on whether or not they are actually safe.
But back to the actual thread: the majority of gang violence is against other gangs. This is unlike other high-crime areas (for example, places with high rates of carjacking) where criminals are targeting people just moving through the area.
Here [1] is their homicide rate. Overall criminality has declined proportionally as well. The numbers continue to decline as well. In 2025 it looks like they're looking at an overall homicide rate of ~1.3.
I think perceptions of criminality is a very important metric because it controls for the possibility of numbers being juked. If everybody thinks crime is going up, but the numbers say it's going down, then it's possible there's some sort of collective delusion. But it's also possible that the numbers are being juked, or that various biases (like declines in rates of reporting) are driving a numeric decline in crime even as crime rates climb.
Not really. In modern times I think we are increasingly missing the point of why we started collecting all of these data to begin with. And that's to aim at giving everybody a more pleasant life. And that is going to be determined solely by their own subjective experience and perception.
Of course you're right that subjective experience will be biased, but it will usually be biased in a relatively fixed way. And so changes in this overtime create arguably the most valuable measurement that exists. Like during the previous administration, trying to brow beat people into believing that the economy was awesome because 'look at these totally-not-fake numbers' was just so dystopic.
So for example, we tend to overestimate threats rather than underestimate them. Yet in El Salvador we now have the overwhelming majority of people (at the 8th highest rate in the world) say they feel safe walking alone at night. That is just an extremely informative datum. I'd also add that people's actions are based on their perceptions. Gallup hits on this in a reasonable way in that survey linked earlier:
---
"In our uncertain world, it’s not enough to make sure that people are safe. They also need to feel safe. When people feel safe, they devote time and energy to learning opportunities and to their relationships with their families, communities and workplaces.
Feeling safe fosters trust in these relationships. This trust forms a foundation for collaboration, cooperation and social development, which makes communities more resilient to challenges such as natural disasters, economic downturns, political conflicts or health crises like the recent pandemic."
Sure, throwing the book doesn't get you 100%. But am I supposed to believe that increasing the penalty for doing the wrong thing doesn't decrease the frequency of the wrong thing? Having everyone you know in your crime circle being in jail vs. roaming free certainly has an effect on your decisions to join/stay.
Probability of punishment seems to matter more than severity of punishment. This follows from economic and game theoretic models and is backed up by empirical studies.
It turns out that when your big worries are "not being able to afford rent and food" and then "getting shot", the difference between 5 years and 20 years in jail, or the difference between a 10% likelihood or 50% likelihood that you get caught don't really factor into the decision-making process.
This sounds great until you’re minding your own business in your house while these people engage in gun battles out in the street and you or your kids catch one of the strays.
If you google this you will find plenty of examples that made the news, and not all of them do.
Here is an article including two such examples. One kid was sitting down eating dinner and the other was sitting in a car. They were both shot totally incidentally during shootouts they had nothing to do with.
People are injured by celebratory fire all the time. That said, getting hit by a stray bullet of any sort is very very rare, which is the reason the stories stick in your head. Children get killed and injured (or injure others) by playing with unsecured guns as well.
The fact is that if there are guns around, there is a little bit of danger especially if they are loaded. Stricter gun laws tend to produce less gun violence and accidents.
That doesn't move mortality numbers much though. It's something like 50 deaths per year from stray bullets, vs 20,000+ homicides, vs 40,000 ish fatal car accident deaths.
That homicides make the news much more than car accidents, and stray bullets make the news at all, is kinda the point of the article.
It is theoretically possible but in the 20+ years I’ve lived here there has never been an innocent bystander killed, and maybe 5 murders I can remember. I live in a wealthy enclave of a major city. I’m just a city guy, I’m not concerned
Tremendous alpha right now in making scary posts about AI. Fear drives clicks. You don't even need to point to current problems, all you have to do is say we can't be sure they won't happen in the future.
I don’t think it does stand. During the Industrial Revolution, automated production replaced craftsmen and often produced a better product.
Specifically, interchangeable parts were a product of industrial automation. Craftsmen were not able to build parts to the same level of precision as machines, so true interchangeability only came when the craftsmen were replaced.
Building software extremely expensive and slow. All of us are balancing quality and speed.
The ultimate goal of building software isnt quality, its aggregate customer value delivered over time, and poor quality undermines customer value. We all have limited resources and are doing our best.
The author can go maximize quality while delivering more slowly/expensively. End users can choose which products they want. Something tells me users may choose products that have more defects but deliver more customer value.
The AI bubble will be like the dot com bubble. There is too much hype and lots of investors will lose money. But at the same time, lots of successful and world-changing businesses will be created.
Does that mean the AI emperor has no clothes? I don't think so.
It might be like the dot com bubble, where eventually a few winners emerged that went on to be hugely valuable. It could also be like the tulip bubble, where eventually prices collapsed and never recovered. Or for a more modern example, the NFT bubble.
I'm not saying that all bubbles build real businesses that have a big impact on the world, I am saying specifically this AI bubble will do so.
I've already seen enough significant, real-world value being created by AI technology that I am a believer that it will have a big impact. It will also destroy a lot of investment money along the way, but that's just normal bubble stuff.
I'm not sure, I've never done it, but I think the idea is to have many tiny customer-specific databases and move them to be powered by sqlite very close to the customer.
But I'd love to hear more from someone more well-versed in the use cases for reliable sql-lite
Atlas may not be the solution but I love the idea of an LLM that sits between me and the dreck that is today’s web.