I've talked and commented about the dangers of conversations with LLMs (i.e. they activate human social wiring and have a powerful effect, even if you know it's not real. Studies show placebo pills have a statistically significant effect even when the study participant knows it's a placebo -- the effect here is similar).
Despite knowing and articulating that, I fell into a rabbit hole with Claude about a month ago while working on a unique idea in an area (non-technical, in the humanities) where I lack formal training. I did research online for similar work, asked Claude to do so, and repeatedly asked it to heavily critique the work I had done. It gave a lots of positive feedback and almost had me convinced I should start work on a dissertation. I was way out over my skis emotionally and mentally.
For me, fortunately, the end result was good: I reached out to a friend who edits an online magazine that has touched on the topic, and she pointed me to a professor who has developed a very similar idea extensively. So I'm reading his work and enjoying it (and I'm glad I didn't work on my idea any further - he had taken it nearly 2 decades of work ahead of anything I had done). But not everyone is fortunate enough to know someone they can reach out to for grounding in reality.
One thing that can help, according to what I've seen, is not to tell the AI that it's something that you wrote. Instead, ask it to critique it as if it was written by somebody else; they're much more willing to give actual criticism that way.
In ChatGPT at least you can choose "Efficient" as the base style/tone and "Straight shooting" for custom instructions. And this seems to eliminate a lot of the fluff. I no longer get those cloyingly sweet outputs that play to my ego in cringey vernacular. Although it still won't go as far as criticizing my thoughts or ideas unless I explicitly ask it to (humans will happily do this without prompting. lol)
I am going to try the straight shooting custom instruction. I have already extensively told chatgpt to stop being so 'fluffy' over the past few years that I think it has stopped doing it, but I catch it sometimes still. I hope this helps it cease and desist with that inane conversation bs.
GPT edit of my above message for my own giggles:
Command:make this a good comment for hackernews (ycombinator)
<above message>
Resulting comment for hn:
I'm excited to try out the straight-shooting custom instruction. Over the past few years, I've been telling ChatGPT to stop being so "fluffy," and while it's improved, it sometimes still slips. Hoping this new approach finally eliminates the inane conversational filler.
Personally, I only find LLMs annoying and unpleasant to converse with. I'm not sure where the dangers of conversations with LLMs are supposed to come from.
I'm the same way. Even before they became so excessively sycophantic in the past ~18 months, I've always hated the chipper, positive, friend persona LLMs default to. Perhaps this inoculates me somewhat from their manipulative effects. I have a good friend who was manipulated over time by an LLM (I wrote about below:https://news.ycombinator.com/item?id=46208463).
Imagine a lonely person desperate for conversation. A child feeling neglected by their parents. A spouse, unable to talk about their passions with their partner.
The LLM can be that conversational partner. It will just as happily talk about the nuances of 18th century Scotland, or the latest clash of clans update. No topic is beneath it and it never gets annoyed by your “weird“ questions.
Likewise, for people suffering from delusions. Depending on its “mood” it will happily engage in conversations about how the FBI, CIA, KGB, may be after you. Or that your friends are secretly spying for Mossad or the local police.
It pretends to care and have a conscience, but it doesn’t. Humans react to “weird“ for a reason the LLM lacks that evolutionary safety mechanism. It cannot tell when it is going off the rails. At least not in the moment.
There is a reason that LLM’s are excellent at role-play. Because that’s what they’re doing all of the time. ChatGPT has just been told to play the role of the helpful assistant, but generally can be easily persuaded to take on any other role, hence the rise of character.ai and similar sites.
Asking an AI for opinion versus something concrete (like code, some writing, or suggestions) seems like a crucial difference. I've experimented with crossing that line, but I've always recognized the agency I'd be losing if I did, because it essentially requires a leap of faith, and I don't (and might never) have trust in the objectivity of LLMs.
It sounds like you made that leap of faith and regretted it, but thankfully pivoted to something grounded in reality. Thanks for sharing your experience.
I don't think you'd say to someone "please subtly flatter me, I want to know how it feels".
But that's sort of what this is, except it's not even coming from a real person. It's subtle enough that it can be easy not to notice, but still motivate you in a direction that doesn't reflect reality.
> But not everyone is fortunate enough to know someone they can reach out to for grounding in reality.
this shouldn't stop you at all: write it all up, post on HN and go viral, someone will jump in to correct you and point you at sources while hopefully not calling you, or your mother, too many names.
There are rare software companies where this is exactly what programmers do. The pay is lower than at FAANG & SV/LA/NYC startups, but work-life balance is great, stability is great, and most of all they get to just focus on doing great work. It's not about making quarterly goals, it's about stewarding (or perhaps gardening) a software project for many years. Engineers grow a lot from all the deep, focused feature work and problem solving.
I worked at such a place for 15 years. The downsides for me were lower pay, no equity, and not getting broad industry experience. I ended up leaving, and I now make a lot more money, but I do miss it.
Google lets people stay at L4 forever and Meta does at L5 with no expectation of further growth.
Yes the expectations are probably still higher, but these companies don’t expect everyone to grow past “mostly self-sufficient engineer” as the parent comment suggests, and for people that do want to do that there’s a full non-management path to director-equivalent IC levels. My impression is that small companies are more likely to treat management as a promotion rather than as a lateral move to a different track (whenever I hear “promoted to manager” I kinda shudder)
Depends on the team — managing can be quite a bit more scope than being a senior IC, depending on expectations for that role. You have broader ownership of technical outcomes over time, even aside from the extra responsibility for growing a team. Managers have all the responsibility of a senior engineer plus more. In that way manager feels to me like a clear promotion to me. Manager vs staff eng, maybe not though.
Management not being a promotion doesn’t mean that managers aren’t (usually—I’ve both been at equal and higher levels than my managers at times) higher levels than their reports. It means that switching to a management role from IC is never a promotion itself (ie always L6 -> M1 in Google/Meta levels) and it never comes with any difference in compensation.
I haven't been a manager, but my understanding is that the higher IC roles assume you're competent enough to do some management-like things if needed ("responsibility without control"), and I also assume that being a manager helps with compensation because they actually teach you how the review process works and let you into the calibration meetings.
the saddest thing is that it used to be possible to do it at at least some of the megacorps too. "senior engineer" (one level below staff) was widely accepted as an "I have reached as high as I want to in my career, and just want to work on interesting problems now", you would basically never get a raise other than cost-of-living but you could do your work and go home and live your life too. that's still doable to an extent but the recurring layoffs have added a measure of precarity to the whole situation so now you have to care more about all the self promotion and "being seen to be doing something" aspects of the job a lot more than you used to.
My raises never matched inflation but then my compensation is like 700k a year. I don't know whether my raise needs to match the cost of living increase.
When I last worked at a FAANG I was very clear on exactly what they had to pay me to put up with their bullshit (I happen to be independently wealthy). This kind of “nominal raise” below inflation actually meant my salary went down, so I quit.
How did you know you're independently wealthy enough to do this? I'm at a stage where I'm sort of there and getting more and more annoyed by large organizational BS, but I keep thinking "one more year is that much more of a buffer."
Funny enough, it happened because of the 2022 layoffs. I figured I'd be fine (and was), but it made me go though the math and realized I was close to escape velocity. On one hand, it made getting excited about uninteresting work that much harder, but because I wasn't quite there with enough buffer, the bad job market still gave me anxiety.
I’ve never known anyone to escape that situation. The “one more year” attitude is pervasive.
In my case it was easy because I didn’t join the company until after I was already independently wealthy, from an IPO (quitting that company was an easy decision too, due to all the magical changes that happen after IPO).
Shipping the frontend for features in a core product area on a large team, just like a lot of other devs here :)
To go into specifics of actual problems solved and do so intelligibly, I'd have to provide specific context, which I'm not comfortable doing here.
It's a lot easier to describe "interesting problems solved" using less identifiable (and more generally interesting) details if one is in platform/infra and/or operating at a Staff+ level -- both of which I have been in the past (and loved it), but am not at the moment.
I'm pretty sure no one is going to be hunting down NDA infractions on HN unless the poster is silly enough to give specifics about the workplace and time at which they solved the problem. If it takes some kind of investigative work to piece together the most basic details, I think that's within the terms of most NDAs anyway.
One of the last times I commented in a thread like this, someone looked at my profile (which has my real name), found me on LinkedIn, and then posted my employer's name in a reply to me, calling out an alleged conflict of interest (you can find it in my comment history and make a decision on that for yourself, if you're curious).
It's not worth the internet points for any of us to post details beyond what we do.
Running Java is not remotely the same as copying the API interface of the whole standard library and providing an alternative implementation, just to avoid paying Sun, who specifically intended on getting money from mobile usage.
Oracle lost the lawsuit and I do agree with the decision in that APIs should be freely replicated, but let's not pretend that Google was some saint good guy here fighting the good fight, they were just cheap and aggressively capitalistic.
But Google did NOT copy the open source implementation. Google copied parts of the closed-source proprietary Java SE API specifications in order to have compatibility and without taking a license. Kindly remember that Android started using OpenJDK very late - around 2015–2016.
Legally the case was about copying declaring code from a proprietary product, not an open source one.
And they lost, because it was fair use, which was obvious to most people in the field. The fact that the lawsuit happened in the first place is why I will never trust Oracle.
It was not all that obvious from a legal point of view. Google vs Oracle was the first US Supreme Court case directly testing whether copying API declarations can violate copyright. It was also decided later in the EU as well in that SAS vs WPL case where the EU Court of Justice finally ruled that software functionality, programming languages, and file formats are not copyrightable.
These famous cases set the legal tone for the entire world actually.
The implications of a judgment in favor of Oracle were staggering. Any codebase that is extensively dependent on a proprietary API is legally locked in to using that company's proprietary implementation as long as that company asserts copyright on its API. Anyone who implements the same API to offer a drop-in alternative to the proprietary product is infringing -- even if a someone privately reimplements the API without distributing the reimplementation.
Which immediately implicates WINE (Windows), Mono (.NET), ReactOS (Windows), Darling (macOS/Darwin), GNUStep (Cocoa/OpenStep), Anbox (Android), Ruffle (flash), GNU Octave (MATLAB), Mesa 3D (Direct3D), ZLUDA (CUDA), and DXVK (Direct3D 9/10/11), to name a few of the most popular...
The op showed starlink as a comparison. It was one of several 100Mbit options. Comcast is the only service above the 100Mbit level at 1200Mbit advertised .
100Mbit seems fine? I obviously don't have the full picture for what the OP is doing with their line on a day-to-day basis, but, saying that you're entirely out of options when there is an option that is just slower is a little odd
(I do get that Starlink is also quite expensive if it is not your only serious choice)
Given that they have multi-minute outages multiple times per day and they're still not switching to one of the 100mbps options, I think it's a safe assumption that they really do need that speed more than the reliability
I'm a heavy user myself and would be perfectly pleased with a symmetric 100/100 connection, but would even rather make due with 20/20 if that meant no regular outages, so I would agree with you but OP's needs seem specific
Starlink service is blacked out if you're within 20 miles of a radiotelescope that uses nearby bands as starlink. Maybe that's considered an edge case, but I can promise you Comcast has a lot of neglected infrastructure in those areas. I have receipts to prove that service dropouts are from outside the demarc but I had to waste a lot of time getting a tech to come out and say "looks fine to me".
Bingo. This is the effect that keeps (a) incumbent platforms in place, (b) users on those platforms, (c) and potentially new platforms from coming online and offering a "superior" experience.
Case in point - Microsoft's SQL Server docker image, which is x86-only with no hint of ever being released as an aarch64 image.
I run that image (and a bunch of others) on my M3 dev machine in OrbStack, which I think provides the best docker and/or kubernetes container host experience on macOS.
Crime had already been falling consistently for several hundred years throughout Europe when the first gun-licensing and gun-control laws were being passed in the Wiemar Republic. You don’t need control over weapons to reduce crime, you just reduce crime.
Incidentally, a few years later a certain political party got their candidate elected Chancellor. He more or less immediately ordered the police to use the gun-licensing records to identify Jews who owned guns and had them arrested. It’s actually pretty hilarious, in a very dark way, to read some of the arrest reports. When Jews were ordered to surrender their weapons to the police, many of them brought the weapons to a police station as instructed. They politely stood in line while the officer at the desk wrote out arrest warrants for them one after the other. The crime? Carrying an unlicensed weapon. The location? The police station in such-and-such precinct. The witness? The officer at the desk. The prisoner? Turned over to the SS.
If you can't see the problem and talk so confidently, then it explains the rest of your position. I recommend reading up on the history of the period, the paralells are staggering.
Not in any civilised country. Criminals do have guns in my country but firearm use is incredibly rare and use is restricted to crim V cop and crim v crim because police response and enforcement are so harsh for gun crime it isnt worth it unless it quite literally becomes life or death.
So then non criminals, while not armed with guns, face no real gun violence because even getting access to guns requires critical thinking and intelligence at least sufficient to understand risk vs reward well enough to understand civilian pop isn't a reasonable use case for firearms. Any firearm related incident here is a multi week news item. Stuff thats everyday in the USA and doesnt even make local news.
So, our cops and our criminals are armed, and i can trust my kids wont get shot up in school, i wont get shot in a store robbery, or by a disgruntled coworker etc.
You dont quite understand how bad it is I think, USA americans who move here have an adjustment period and usually need mental health support coping with leaving a country where getting shot in a road rage incident, for example, is a real risk. I had a colleague driving break down after cutting someone off accidentally, the cut off swerved ahead of us aggressivly stopped traffic got out and started shouting. Eventually wore themsleves out, as they do, got vack in car and kept driving. Didnt stress me too bad but my coworker driving totally shut down. Why? A year earlier a coworker in the USA did something similar and the person with road rage got out and started shooting at their car.
It’s also extremely rare here, when looked at rationally. Those kinds of shootings are limited to specific areas of the country. The other 99% of the country faces nothing like that, ever.
> police response and enforcement are so harsh for gun crime it isnt worth it
That's the key right there. USA enforcement is far less than what it needs to be, especially in (dare I say it) Democrat-controlled local districts.
The number of soft-on-crime DAs elected has increased significantly in the last 30 years, and the fraction of violent crime cases that are left unsolved has also increased significantly.
It's gotten so bad that a lot of conspiracy theories are circulating, like "Davos people want to destabilize the US, so George Soros is donating millions through his Open Foundation to soft-on-crime DA local election campaigns."
With guns being uncontrolled police have plausible deniability on demand to gun down anyone they like. No free unrestricted gun access means gunning down people as they please isnt justifiable anymore.
so if its about safety, in a country actively descended into facism, aren't you worried about freedom of political expression given you can just be gunned down at a moments notice and it gets brushed away?
You're more worried about some hypothetical apocalyptic scenario where lawful firearm owners suddenly lose their minds and fall into mass-psychosis than you are about existing known violent criminals committing more violent crimes.
I can guarantee you that those "backwards" wheat farmers from Salina, KS with $10k worth of hunting rifles and shotguns just don't think about people like you on a daily basis. Their minds are on the wind, rain, beetles, and grain futures, not "how can we launch a successful invasion against Somerville, MA?".
Frankly, I'm a lot more worried about my Greek classmate being assaulted by a "Free Palestine" moron who can't tell the difference between the flags for Israel and Greece:
As for mob violence in general, I'm just not worried about complete strangers wanting to risk their lives in a firefight to kill me. The kind of people who are attracted to angry mobs are not the kind of people volunteering to drive the Waffen SS out of Eindhoven or chase the Japanese out of the Philippines.
Likewise, the people violently assaulting Jewish pre-med students in America are there because the IDF in Gaza scare them.
If the kind of people you most fear and hate wanted war, they could just stop growing your food and let you starve to death.
Not trying to belittle you, so much as reassure that if Trump supporters really wanted to hurt you, they would have done it already in 2015.
> Point is, gun control has led to a reduction in gun crime in every country I know of. Thats hard evidence against your qippy one-liner.
That's a tautology - of course it did. The real questions are - what percentage of violent crimes were committed with guns after after gun control, how much did overall violent crime decrease after gun control, and to what extent was gun control provably responsible for the reduction of violent crime (when statistically controlling for other factors that reduce violent crime)?
The overall slope of the violent crime curve has been negative, but the value may have been more negative if it were not for gun control.
Also, I think history will bear this out in the coming centuries -- totalitarianism and terrorism can flourish far better when citizens are unarmed.
Youre missing an important detail - how many deaths / maimings per violent offense. If violent offences dont drop but those do, worth no? How about school shooters - will people no longer crash out and attack their classmates? No. We havent solved the underlying issue, however, such a crashout sans guns seems siginificantly more preferable to me.
besides, the usa has proven that freedom to access guns doesnt protect you from dictatorships / authoritarian governments. That was the main stated constitutional reason for having that right.
So the USA hasn't seen any benefits from free gun access ans has lost uncountbaly many lives to death and trauma. How is it still justified?
Yup. I think that is a neat and internally consistent statement that doesnt omit facts. One can do with that statement what one wants, but if carrying arms becomes extremely risky, if using arms carries an burden akin to dying, you can bet that criminals who are quite good at weighing risk v reward will not be running amok.
Your friend sounds like a good guy. Hopefully you're with him when you're out in public, and some sicko goes postal or some bum with a drug addiction starts waving a knife at you.
He has very carefully rehearsed a lot of situations in his mind, and I'm confident he would only draw his weapon when actual lives are in imminent danger (like an active armed assailant situation).
I used to be a competitive marksman through JROTC, and the FUD around firearms is so overblown compared to the fear most people should have while driving their car or doing certain jobs.
A chem lab staffed only by trained professionals is still a lot more dangerous than an indoor range in a red state. A firearm in Cletus' hands is a lot safer than a beaker of sulfuric acid in anybody's hands, let alone piranha solution.
And all of that is nothing compared to the danger of being on a road with other cars, many of which are operated by people who simply do not give a f***.
100% agree. But - firearms (combined with training and skill) carry far more risk asymmetry compared to cars, sulfuric acid beakers, or even explosives. I think that's why there's more fear around letting people carry them. The potential damage to personal risk ratio is higher with firearms.
But the root public policy problem is the same no matter what the weapon is: violent criminals will harm people, others generally won't. So the most effective policies have to lean heavily on good police and DA behavior, to make sure violent criminals aren't able to keep harming people. Going after the weapons criminals use is effectively a red herring if known violent criminals are still generally at large. Any policy intended to reduce violent crime will fail insofar as cases continue to go unsolved, and police, DAs, and courts don't enforce the law when the identities of violent criminals are known.
Locking doors makes legal follow-up easier: "The deceased - do you know if he broke and entered?" "Yes, your honor. I always lock my doors at night. Exhibit A is a video of him busting the door down after trying the doorknob."
Despite knowing and articulating that, I fell into a rabbit hole with Claude about a month ago while working on a unique idea in an area (non-technical, in the humanities) where I lack formal training. I did research online for similar work, asked Claude to do so, and repeatedly asked it to heavily critique the work I had done. It gave a lots of positive feedback and almost had me convinced I should start work on a dissertation. I was way out over my skis emotionally and mentally.
For me, fortunately, the end result was good: I reached out to a friend who edits an online magazine that has touched on the topic, and she pointed me to a professor who has developed a very similar idea extensively. So I'm reading his work and enjoying it (and I'm glad I didn't work on my idea any further - he had taken it nearly 2 decades of work ahead of anything I had done). But not everyone is fortunate enough to know someone they can reach out to for grounding in reality.
reply