726 points | 3d ago | Discuss on Hacker News | Back to Radar
I'm willing to concede that there may be other areas of software engineering where it makes more sense to have the narrow specialists you describe, but it hasn't been my experience. Also for junior staff it might make sense for their senior peers to project-manage them a little bit so that they're not too far outside their comfort zone.
They usually do as they please, and assign extras or not, and ultimately learn in the long run.
This isn’t to say that extra staff on a project can’t be helpful. But more often than not, management miss-estimates how the extra staff will impact the project.
In my experience what instead happens is that companies are ok with bugs so long as you communicate why it happened and how you are going to fix it. So the tradeoff is clear, spend a lot of time managing and let bugs happen so long as they are being managed properly. This is what I think causes fragile software at scale. People with high level technical knowledge should be hands on, and any bugs should be unacceptable.
The average manager... gets results that depend on the quality of the people he manages. If he is lucky, he gets to manage smart ones who give a shit, and then he is praised for managing them well.
On the other end of the bell curve, there are managers who actively harm everything they touch, and increase the work stress for everyone around them. I have seen such managers too. As long as the people they manage succeed to get their work done, despite all the obstacles, these too get a bonus.
The shits that I give wax and wane on a quarterly basis, but the average pattern is me being largely disengaged, disincentivized to well, and burnt out. I’ve basically told my company they can’t really count on me for anything.
Maybe they’re trying to get me to quit or waiting for the contract company to hire me direct.
Sorry, different type of “added responsibilities “
Hope you find some resolution on your side. It can be tough sometimes.
can you imagine if the IEEE had formed a union, way back in the 40's or so...???
I can see where the code so brilliantly crafted today, and unintelligible to it's inheritors, would result in the same kind of frustration...
...except no human to stare in the face and scream at
The Seattle area is full of tech employees who are millionaires and some even up to 9 figures.
SV is also not the whole industry so looking outside of that, I see exactly what OP is saying. Technical staff are usually required to split their time across numerous project management tasks that are mostly to support the project management team's communication of goals to the client up the ladder. I am not purporting to have a solution to the problem mind you, it's more complex than just "let em code man!"
I know a guy who got a Facebook job with a million dollar compensation package. For programming.
Another programmer friend made $10m at Microsoft and quit to found his own company.
I also knew "Jane" (not her real name) who I worked with. She was convinced she was a great programmer. Unfortunately, her code was always a mess and had to be rewritten by myself or others. It was so bad that around the office, bad code was called "Jane code". She eventually was laid off, and was fully convinced it was because of sexism.
I think you're conflating very different groups. Folks with 100+ million dollars to their name are going to be owners rather than "tech employees" and I'd guesstimate 100 of them before filtering for "tech".
IYKYK
On the flip side, I have met some very intelligent people who are incapable of self-reflection.
Of which, there is also a modernized version: https://www.joelonsoftware.com/2000/03/23/the-guerrilla-guid...
OP is noticing that those who are smart and care about long-term outcomes beyond their job incentives (aka, "give a shit") have been sidelined or eliminated. Because they are in the way of "getting things done".
https://www.theguardian.com/uk-news/2026/nov/03/ai-godfather...
It is not an environment where you are motivated to do your best work. You'd have to have a screw loose actually to go above and beyond in these places. Working hard gets you nowhere here, just spins your wheels in the mud.
There are certainly places that do value your contributions, but they aren't ever going to be the sort of shops that have newspaper headlines announcing a perfectly even number of layoffs ahead of earnings.
The archetypes describe basically everyone to some degree. They're like a zodiac sign or an MBTI dichotomy or an OCEAN axis.
https://en.wikipedia.org/wiki/Double_empathy_problem#Counter...
Wasn't even argumentative, just quietly gave a shit.
From their POV quality, privacy and security are not worth investing in. Unless the government forces their hand, and even then they'll find ways around it.
Me too! Very smart guy, well mannered, easy to work with so support kept reaching out to him and the senior manager fired him because he was focusing more on solving issues than new stuff. Also, support is a different org, they are born to suffer.
The manager wasn't going to get his own growth by not delivering much new stuff and just improved maintenance. So obviously he didn't focus on how we can reduce these recurring issues so that we have more bandwidth to focus on new issues. Because who cares about pressure on these developers, they are an unavoidable nuisance, you see (AI is solving that very long standing grievance).
Laser sharp focus on survival + growth + current bonus cycle will do that.
What do we do about this mess?
And no I’m not just talking about office politics. Often people who feel they’ve been treated badly are giving a shit about “business outcomes” that they believe the company should/would care about (often not the case), or they gave a shit about the specific tasks they were given. Can’t blame them for that.
But what successful people give a shit about is more along the lines of solving superiors problems etc, regardless of what “business value” is or what your written job description is
Happy to solve superiors problems but they don't give me their problems, they give me Jira tickets
Unfortunately there's only so much time in a day and if I'm going to get shit for not closing enough story points then where will I find time to kiss ass like this?
Seen that a lot lately.
The bigger the corporation, the more out of touch it appears to be with actual problems, and the more it seems to attack anyone trying to solve actual problems.
Been getting crazy and crazier for awhile now.
Boss: implement X
Me: it will never work (gives reasons)
Boss: do it because I say so
Me: you pay me enough that I feel obligated to tell you it won't work
Boss gives X to another person. It failed.
Nothing is perfect.
It's not that smart of a path to take, so ... it takes a certain type of smart person to fit into that position long term and not lose their mind or get sick of it. You have to kinda want to be subservient to some degree, or else focus on what other smart people who give a shit are around that you can manipulate into being your servants.
The article is quite close to a deep insight on that point, but it misses the moment. There is a surplus of people who care, the limiting factor is the ability of the management class to value the outcomes.
If the capitalists understand that something creates value, then they will treat the people doing that thing well (better than the people not creating value, anyway). If they don't, then those people will be treated poorly. The key is understanding and not the on the ground outcomes. That is why metrics and processes always end up being such a big deal in corporate culture; because the people in charge are aware that the limiting factor is their ability to understand their own company and they're just not necessarily very competent at acting on that knowledge.
It’s a feature - not a bug - of the capitalist system mentioned.
Not sure this is a hot stove moment, but it will certainly be a reckoning and likely result in some significant regulation, common efforts through partnerships with private entities, and a more measured societal view of the good of companies - only to revert back again in a 60 year cycle. Such is history and life.
Isn't that already being pushed with Zelle?
When I worked in blockchain, everyone there claimed that they "Were in it for the tech." But in practice, by 2018, NOBODY cared about the tech at all. All the projects were over-engineered and bloated. It didn't matter; those projects raised millions. The few well designed projects which were created with minimal code and dependencies never got any attention or traction. In my community, they were even suppressed with founders threatening to go after innovators legally and defaming people who they should have supported. Nobody cares about tech. Everyone is using it to create complexity to improve their job security.
Finally got my first well paying corporate gig as a developer. First year was told the overachievers are the ones making the big bonuses. Went all in. Busted ass, took on additional work, learned new languages/frameworks. I had a dozen goals I completed. Other devs were telling me back off and slow down, I was making the whole department look bad because of how hard I was working.
End result? Mid review. Barely a bonus & a kick in the nuts.
Next year? I did the bare minimum. DGAF about anything. Full shutdown, Honey Badger mode.
End result? Mid review. Barely a bonus & a kick in the nuts.
I finally realized the one person who benefited when I gave a shit was my manager because I made him look like a rock star. Me? I got nothing. Walking out of that year end review pretty much sealed the deal for me. I finally realized I can just focus on my own well being, still get paid nicely, burn all my PTO and not GAF about anything.
Its a pretty refreshing place when you can not give a shit about anything and still be gainfully employed. I was quietly quitting a decade before it was even a thing.
A lot of the sour posts are people that just don't get it. They want to do what they personally "give a shit" about, other opinions and priorities be damned, and that's why they'll be kept on a short leash forever.
Guess who is in charge of the AI companies?
It reminds me of a Michael Scott quote from The Office -
> Don't ever, for any reason, do anything to anyone for any reason ever, no matter what, no matter where, or who, or who you are with, or where you are going, or where you've been... ever, for any reason whatsoever...
Just get those tickets done or they'll find someone who will.
just like how you cant have independent journalism when they have no income
VisiCalc and the start of the spreadsheet era went pretty well for many people.
There aren't enough humans who are smarter than the machines.
That situation gets worse every week.
Giving a shit actively gets in the way of that. If you care too much about things, then you're going to waste time focusing on things that don't actually make any money instead of shipping.
I wish companies would hire a professional shit giver whose sole task is to care about things and nurture them, make things better overall. It doesn't seem like they're ever going to do that, though.
They are sidelined, because giving a shit is not reflected in OKRs and giving a shit often equals standing in a way of "good enough" as judged by a middle manager and thus committing mortal sin of not focusing on getting to market faster.
And if you do a lot of that sort of activity you'll inevitably eventually do something that makes your hierarchical asshole of a boss look ever so slightly imperfect or could be construed by some other politicking twat of manager that way and so your boss will stomp you rather than let it continue because the value you'r producing is not reflected in his KPIs and politics whereas the risks are. And if this doesn't happen as a result of shit on your boss's level odds are it'll just happen at the next level up once you've done enough work to matter.
So keep your head down, close out the tickets, do your going above and beyond type stuff off the record and enjoy the paycheck while you can. Sucks, but it's the only way to stay sane.
If you are a metric driven organization, giving a shit is driven out of the organization in favor of tracked variables. Which usually happens around the time the company headcount hits Dunbar's number.
Do what they tell you, do the right thing for this immediate point in time right just now, don't do the right long-term thing!
eeuuuch, I feel horrible writing that even though I'm mocking it.
Ha! No, it doesn't. The moment that the capital owners can bisect responsibility enough that each cog can be replaced with an "equivalent" (read: younger/recently graduated, capable of typing code) cog, the smart and caring are discarded. This is because People With Opinions can derail the goals of the investors; because Smart People begin to understand just how pointless the ISOs are.
The founder with the engineering intelligence and the business intelligence to build their vision is the only one who stays stealthy and small, needn't take investors, needn't hire dozens of additional engineers.
The actual end result doesn't seem to matter too much, it's the time and cost savings that seem to be the motivators, pure ultra-short-term-ist capital-on-pedestal thinking, with barely a pause to consider 'the product'.
It's to the point now that I regularly see large and complex systems being built by essentially one guy but with dozens of "managers" overseeing his solo work. I've been in meetings with "project coordinators", "project managers", "governance", "change control managers", etc... all on the same single-product team and only one person there actually "presses the buttons".
This has shifted power and decision making away from the "smart people who give a shit" over to the "politicians that like to play Game of Thrones In The Workplace".
Look around you and see if you can spot more of these "inverted pyramids" with the point at the bottom: one technical person at the bottom "managed" by a team, which in turn has an even larger layer of managers above them, etc.
2. The “smart” people at the top are generally selfish assholes. They abuse the system and others to self enrich. We don’t need more smart people, we need more caring people.
3. Nobody gives a shit anymore. The older generations fucked it up and are aging out and just saying haha sorry we messed it up but good luck while I chill in my 7th vacation home. The young are jaded because all the jobs are dogshit and underpaid. The only way to win is to get rich overnight on Kalshi or crypto or futures or whatever. Why would you work for most of these horrible companies? It’s not worth your life.
4. Look at POTUS. Grift on grift. Everything falls apart in a society that runs on grift. They couldn’t even re-line a pool…
It's kinda obvious then that smart people go for 2. It's just the logical conclusion.
I say that as someone who really wants to give a shit. After my first working experiences though I realized it is pointless.
IMO, the issue is that AI is reducing the cost of building and modifying systems much faster than it's reducing the cost of understanding them. For example, an engineer can work on a problem or project for a week that could otherwise be done by AI in an hour or so. But, that week spent was not "wasted time"; it was understanding what it is they are building, discovering undocumented assumptions, knowing about different dependencies, developing intuitions about the system and how it works, etc.
This causes comprehension debt. Yes, the system does ultimately work, the tests also pass. But the level of comprehension goes down significantly.
That said, AI can also help reduce this debt by asking it what you need help understanding.
A good blog post on this: https://larsfaye.com/articles/ai-coding-will-prevent-experti...
And as far as air traffic is concerned, they’ve been trying to strangle it since at least 1980. Canada has a privatized system, so it’s possible to create such things but that is a matter of political will not money.
When you change a system directly you learn the underlying nature of the system and your toolset for making changes. Guide a third party toward making the changes and you learn the nature of the interaction and your toolset for evaluating the outcome of changes.
Does the distinction matter? Depends on your perspective I suppose.
Absolutely.
However, we've also never poured more in, to such centralised infrastructure.
I've actually never worked in any company that has had a communications outage for 36 hours in my life; and I've worked in places we had a dancer IRCd running on an old laptop under the faux floor in the server room with Solaris as the main communications tool... thing is: it was so simple, when it went down we just fixed it.
Technology companies have very successfully convinced us that we should outsource responsibility, because there are economies of scale. Then they ask for a free pass because "scalability is hard".
You don't get to say scalability is hard when you've been pushing and propagandising centralisation for over a decade, while having thousands of staff on million-dollar yearly TC packages.
Everyone here will say "nobody got fired for buying x- because everyone does" (Slack, IBM, Microsoft, AWS), and you can use it as an excuse... but you know what happens when we over-rely on something like that and it fails?
Economic crashes. That's exactly how the 2008 financial crisis happened, everything depending on something that had "absolved" responsibility. Such a thing has ruined the prospects of an entire generation of people, and I'm actually becoming really upset that we think we're being clever bastards because "oh hum, well, I can't be held responsible for total systematic collapse, everyone else was doing it!"
And yet, in the entire history of electromechanical switching in the Bell System, no central office was ever out of action for more than 30 minutes for any reason other than a natural disaster. (With one late exception, a major fire in New York City in 1975.[1])
That was achieved by design. Electromechanical telephone switches were designed to have no single point of failure. That was achieved in practice. Failures occurred all the time, but the problem was confined to a small group of lines, if not a single line. I'm not going to go into how that was done; read up on #5 crossbar if you're interested. Even complex functions such as long-distance routing had backdowns to static routes stored in read-only devices, in case the telco equivalent of BGP failed.
That reliability record has not continued into the computer era.
[1] https://en.wikipedia.org/wiki/1975_New_York_Telephone_exchan...
The reason I believe has to do with cloud sitting on top of layers and layers of abstractions, reused for "simplicity" or short time to market rather than because they fit the domain.
I've been doing a bunch of "side quests" in an old old codebase fixing a lot of things that were not a priority but sure as hell needed fixing lately with AI, far faster than I would have otherwise. Lotta tech debt gone.
But no, we need wise leadership not just "smart people." We smart peopled our way into this mess.
In a perfect world, the second and third group would be the same, but since impressions of competence are lossy as information travels up the hierarchy, optimizing for the impression of competence (i.e. “visibility”) will always be more rewarded than actual competence.
Says who? I don't want to be a manager, I just want to write good code and make a difference.
In other words, if there is no IC track or an IC’s ability to shape the organization is less than a manager’s, you will continue to be undervalued.
It said they pretty much immediately recognize one another and will frequently team up with their manipulations and schemes to advance, etc. At some point, after succesfully moving up the ladder a few rungs, one of them will make the move to nuke the other one.
It's now bitterly ironic that the grim assessment was, by comparison, actually lively and beauteous! Compared to the Dead Internet and meat proxies and stochastic slime we now experience on the daily. The internet of beefs was pointless culture war, identity-based feudalism, but it was unmistakenly, unflinchingly human.
Willis plays a somewhat scruffy workaholic, that gets a lot done, and his coworker is a vapid womanizer that gets nothing done, but all the promotions.
It was an extremely funny movie, too.
They're creating being the defining characteristic. I give zero shits about the quality of the code generated by an LLM. I didn't create it, I have zero pride or ownership vested in it.
When I ask to change things, it's not because I'm dick. I just want us to build stuff that I'd be pleased to use myself. Simple, elegant, bug-free, fast, secure and stable.
I used to find the Lean and Agile movements enticing but their implementation in enterprise often boils down to doing the minimum demanded of the current task and shelving it until somebody complains.
Smart people say, "Hey did you think about XYZ?" during the planning phase. That makes the plan stretch out to a realistic deadline. How dare they! Ha. Nobody likes that. And it's been this way for a while. Maybe it's always been this way?
* Remembering Roger Boisjoly: He Tried To Stop Shuttle Challenger Launch : The Two-Way : NPR // https://www.npr.org/sections/thetwo-way/2012/02/06/146490064...
It is a status club now, and giving a shit does not make you one of the cool kids. Giving a shit is papered over as an “ego spike” and cool kids don’t have those, because they make sure not to take the risk of insisting on anything.
Yet the thing is rate at the top of how AI have revealed to me that in an organization of 400 people, 3 give a shit. I literally can’t find anyone who gives a shit in insanity large organizations. Not low paid entry workers, not partner or VP, not CEOs. All just a bunch of people who probably never gave a shit to begin with, but AI put that at the forefront.
There are a lot of people motivated to do certain work they find valuable but they realize that there is time contstraint to meet the criteria so for some of their papers they take the easy way out.
I don't want to say that this is good by any means. It is just part of reality.
Don't underestimate how AI can change who gives a shit. For my company:
Before AI:
-My boss was very much a manager and was hands off code
-I did a lot of extra research, problem solving, and design for my company and was the goto person for basically everything
-I loved my work and accepted a low-end salary because I loved it so much
After AI:
-My boss is pumping out feature after feature using teams of agents with no understanding of how things work
-My perks were slashed, removing about half of them
-My salary is not enough to compensate for my spouse that can't get a job due to AI, and I have to worry about putting food on the table
-My research isn't needed because everyone just uses AI
-My problem solving isn't needed because everyone just uses AI
-My design skills aren't needed because everyone just uses AI
-I'm not allowed to do my job in the way I did it before that I found relaxing, because AI is faster
I worked for a safety equipment manufacturer for over a decade. I did everything, programming, servers, network, machine maintenance/repair, etc. I transitioned them from a 20 year old accounting system with 10k SKU/20k BOM to a new ERP system with zero downtime. My reward? nasty discrimination and forced to quit
I used to be proud of my ability to do a wide variety of technical work, programming, systems, servers, industrial machines, industrial controllers, lasers, molding machines, etc. Nothing was beyond my understanding. All I got for all my hard work and skills was shit on
AI has been great for programming tasks and the software I have worked on for years (I just enjoy programming for programming sake, would be nice if I could be paid for my skills instead of judged for how well/poorly I lie/ass kiss)
It kills me I was forced to stop giving a shit. The sycophantic AI is perfect for what businesses really want.
When you do land at a place that appreciates your effort, it's difficult to accept the kudos. It's a kind of shell shock, and it takes awhile for the painful memories to fade. I found myself following my favorite manager from job to job because I didn't want to work for anyone else, and life and my self-image improved dramatically.
Industry has just gotten significantly worse since I left that abusive place, I seen easily/clearly what was about to happen (which is now happening). We haven't even scratched the surface yet with the recent downtime
I am not sure if I have the energy to get back in, took me years to get my head straight after that horrible shit and just do not want to risk my health like that anymore
I glad to hear you had a good manager to follow though. I know there is good places out there, maybe one day I might try again
If you get back that energy, there's still someplace for you that you'll thrive in. Be a lot more choosey and do your research down to the individual level where you can.
In my personal experience the intelligent people are making data driven factual arguments. Without root cause analysis of these various failures I don’t see how one can conclude it’s due to uncaring, unintelligent people. I’ve seen plenty of failures that are due to unforeseen issues that come up despite having caring, intelligent people owning those systems.
This describes most major outages where I’ve had enough of an inside view to know whether such people were involved. It really makes me wonder to what extent the author’s conclusions came from observing who’s working on the systems that had high-profile failures.
Those who give a shit and care, I feel are going to burn out by this system. I wish there was a simpler answer for less privileged folks out there.
And that’s exactly why I don’t really give a shit anymore, at least not “above and beyond” like my dumbass self did in my 20s and 30s, wasting away time “proving myself” only to get thrown in the dumpster during outsourcing or layoffs or petty grievances from clients.
Nobody pays me enough to buy a decent home in a decent part of town; not the current job I love, nor the prior jobs at huge conglomerates and household tech names. I spend two to four hours commuting from suburbia every day because I don’t make enough to live closer, nor enough to buy a home where I currently live. Watching my youth be squandered at the altar of Capital for nought - no house, no spouse, no sense of community - the only shits I can afford to give are proportional to my compensation package and not one more.
It’s not that we need smart people who give a shit; it’s that we need leaders to actually pay these smart people to give a shit. If my current employer paid me enough to buy a domicile near a subway stop, I’d gladly donate excess shits (and fucks) to night and weekend work again, if only because I wouldn’t be wasting a full fourth of my day in traffic anymore - and could afford to build the roots and community and love life I want for myself.
A CEO might know that keeping smart people is a good thing. But the incentive structure for the directors and other middle management just make it impossible because smart people get expensive and tossing them looks good for those middle management bonuses.
Smart people are opinionated, and management thinks very highly of themselves so they can't have anyone with "bad vibes" around.
Of course there are more. So what can you do?
I couldn’t help but laugh when Anthropic posted about how they made a new version control system, but with this amazing idea of being centralised. SVN was right there.
To be clear, I’m not advocating for the rest of your complaints.
But in most cases we want that project to be observed and shared with our community, we want to be recognized.
Today the thing you built is a mere token cache, ripe context for reaping.
Fool me twice.
As a very smart person I often have to zip my tongue just because of that.
World is imperfect and tbh all of it works just because of few extrordinary ppl and a bit of luck.
And now insurance premiums are increasing to $7k/yr. Do you think salary is going to reflect that reality?
There's way to many General Kauls and far too few General Groves but of course the former all try and pretend like they're the latter.
The CEO is completely incompetent if that's the incentive structure they're setting up
"Capitalism" is, at best, orthogonal to this. When we came to the U.S. in the late 1980s, California was the world model for a place run by "smart people who give a shit." It was also so capitalist that Reagan and Nixon won Los Angeles. Twice each.
What you're really yearning for is a society run by high-agency, conscientious, low-bullshit unemotional people. You can find these people in social democracies or in libertarian capitalist states. You need "common people" who are socialized to take direction from "smart people"--with all sharing common goals of competence, efficiency, and orderliness in preference to other values.
We need smart people who give a shit, demonstrate good taste, and welcome "but why??" challenges to their ideas with confident, well-developed, reasonable answers.
We don't need cowboys. We don't need ivory tower religious fanatics. We don't need big mouths who win arguments with aggression or oppression olympics emotional blackmail. If you want to lead, lead with grace, otherwise follow your leader with equal aplomb.
While we no longer have the time nor energy for any bullcrap due to AI slop, we see and appreciate when AI is used well.
The greatest predictor of wealth is age. If you're a young buck and feeling pinched, invest in yourself, go heads-down on work for five or ten years, and build them skills and good taste - you'll end up very happy with the results. Don't let the turkeys get you down!
The business leaders only see things that they can fit into their spreadsheets (both literally, and figuratively representing the mental models they use for decision making).
The productivity and skills of individual engineers get washed out via some average. The blockers and nonsense that slow down your best teams disappear, and those good teams are measured to be average. Building things the right way does not increase revenue so therefore it is not worth doing. Off shore engineers are cheaper so its more profitable to use them.
Do this a few hundred times and you get to this insane point where the mental model of those incharge is so off base from reality that planes start falling out of the sky...
Even if you have a pocket of middle managers who do understand, and want to communicate that change upward, you are still in a terrible spot of telling your bosses everything they know and have done is a lie.
This problem is not limited to just MBAs vs Engineers. We nerds are guilty of it in other ways. I have worked in Data Science / Machine Learning / AI Engineering my whole career - which always has the fascinating problem of building what stakeholders think is magic and managing those expectations. What I have noticed, especially the last year, is people use the magic of AI to offload the hard parts of product design, figuring out ugly data, contradictory business rules, etc... this sort of thinking feels good, because it lets us believe the problem is easier than it really is. But at the end you are left with some AI slop product, or disappointing automation.
I'll never forget attending a project proposal by a fresh big 4 consultant hire at my current job, that promised to save $10M in costs. The plan was literally: 1. Get data on < problem > (which did not exist in a useable way) 2. < My name > builds a solution with claude 3. we save $10M a year
Its literally the underpants --> ?? --> Profit meme from southpark.
I see this issue getting worse and worse, because the decision makers are further and further detached from the details that keep our planes from falling out of the sky, or outlook crashing.
The real work gets done in the trenches, and the generals don't know what the mud smells like. The one upside in all this, those who are willing to get into the trenches have the tools to leave the generals behind.
This is a gentle reminder that none of this is true. Step one would be to try to determine if the recent outages are more than what would be expected by random chance (which none of these articles ever do), or if now you're hearing a lot more about every little outage in an attempt from reporters to craft a narrative (e.g. the MS 365 outage and FAA ground stops were noticeable anomalies, but after that it felt like reporters were just refreshing downdetector for new stories, e.g. the Apple services outage was relatively small and short lived).
I've read other theories that many of the recent outages are due to increased release pressure from companies needing to patch systems since AI is finding lots of long-standing vulnerabilities. Again, certainly plausible, and that plausible explanation is just as plausible as OP's opposite plausible explanation.
The tech industry as a whole hasn't gave a shit for decades. Oh sure, every good company gave a shit at some point, but then the shareholders started calling and begging for more and more cash and, well, here we are. All AI has done is make visible the symptoms of not giving a shit that people conveniently ignore for the longest time until it affects them. Until Capitalism is either abolished outright or very deeply regulated so that people can actually give a shit again, I sadly don't see this changing. And yes, I know, I know, I did say regulated. I'm sure some will hate that. But just letting capitalism reign free is also not going to work.
With how big of a speed-up? Not enough, for probably most managers
There are plenty of smart people, but you probably aren't paying them enough to argue and play politics with the imposters that you have also hired. If doing something well also requires convincing a (product) manager that it was their idea, then no one will be doing anything well. The best are going to checkout.
If you get rid of the imposters and lessen the cognitive burden of playing politics to get anything done, it will appear as though all your smart folks started caring.
The thing is that people want a nice life, and most smart people know that martyring themselves will neither save the world nor give them a good life.
The rational thing to do is walk the rim of the proverbial canyon, staying close enough to make good money but far enough away from the Big Questions to never feel like they should become your responsibility to sort out. Because once you agree it should, it’s a self respect hit to walk away without doing something, or do something at personal cost and probably fail to achieve your goal anyway.
Especially since we are not the best people to fix social problems, which very many things are.
no. everyone loses
You can't force people to give a shit, just as you can't impose some arbitrary marker of "smart" or assignment of value that is ultimately some abstraction that is convenient as a heuristic. The assumption that there's some clear and uniform way people would signal that they give a shit is also fundamentally flawed. These are all presumptions built on top of presumptions, heuristics that are cultural-bound and subjective and legible in the high modernist sense, but does that actually serve a meaningful point of reference? The loudest voices in the room only tell you that there's someone who is loud, but whether that is better than someone working their ass off so that the system doesn't get to a point where people feel compelled to be the loudest voice in the room, is open for debate. Hell, why even pretend that these are new problems? There are only so many hours in a day and if nobody gave a shit 5, 7, 10, 15 years ago, at some point one feels like the idiot for giving a shit in the open. Trial lawyers learn real quick that betting on jury verdicts is never a winning move and if even guessing what 12 people would agree on is a fundamentally befuddling experience to people whose job it is to make the point to 12 people in that setting, I think it may be wise to assume that nobody knows ahead of time when or how people would suddenly start to make some showing that is a reliable indicator that they have seen something that crosses some societal red line as well. And really, it just turns a deterministic outcome into a probabilistic one, which sometimes is enough and other times, unbelivably hubristic in hindsight. Three days ago Ofcom in the UK made a recommendation to parliament that is at once wildly outdated and almost certainly already something long attempted in the UK: that CSAM detection should be done via hash-matching. The MD5 hash was first implemented in 1992, and a system as described have in fact been used in practice since before 2010, and by 2012 one can find working demonstrations of hash collisions that made the system effectively useless as implemented. Of course some of us knew before then that the system is unsound, and only a matter of time except explaining why this is vulnerable and unsound was always easier to show than to explain. In 2024 I took a laptop into a suppression hearing and ran some code from the internet that demonstrated that two different files can generate the same MD5 to a judge who, after thinking a few seconds, seemed completely unperturbed and asked what did that have to do with the police database being used to... detect CSAM, which, at the time, was a MD5-metadata based system. People were sent to prison based on this nonsense and if showing the system breaking in real time wasn't sufficient to convince the actual power that determined the admissibility of the broken system as evidence, well, fuck. That was 12 years ago. It now takes less than a second to run much better code that does the same thing. There are hundreds of examples in many languages on Github doing just this. I have no idea what the system currently being used is, but as far as I know, the FBI was still using some similar sort of hash-metadata system that had a degree of metadata detail that is detailed at a level that only those who prosecuted a case involving the production of the material being hashed would have as recently as 2023, according to publicly available filings on PACER in criminal proceedings. Since not even the lawyers involved in defense are privy to even the provenance of the material being hashed, I have no idea if in the interim there was the complete upgrade of the system to a more secure system, but since no significant case requesting convictions made on the flawed data being thrown out had appeared, I'm guessing that the answer is no, because some pseudoscience did finally get recognized as bunk, but certainly after innocent people were executed after being convicted. Orin Kerr, a much better lawyer/tweeter and respected authority oon computer crimes and legal issues, spent two decades of his life advocating for some kind of discernable line when it comes to figuring out a system that is consistent and applicable when it comes to figuring out where the boundaries of permissive access is. This happened in 2020, when the Supreme Court in van Buren v. U.S. adopted a gates-up-gates-down approach that was suggested two decades prior. Before the ruling, there was literally no consistent approach under American law to indicate what counted legally as boundaries in computer access terms. That meant that from the day the world wide web existed until 2020, you can access some publically served document on the clearweb and whether you broke the law was not something you can possibly know. This was why we had a bug bounty system where routinely serious vulnerabilities were discovered but nobody was paid for the discovery. The implicit bargain, which was legally invalid to begin with, was more like a bug-hunter-held-hostge deal: you get the privilege of telling the company of some vulnerability and they promise to not call the FBI anyway, not that the legal system needed that call to prosecute you anyway. The bounty was at most a token gesture that allowed for an NDA to be legally enforceable definitively, since contracts need a bargained-for exchange and consideration in return. You can't use coercion as part of the exchange, unless you are a federal prosecutor, in which case, every case will be an elaborate prisoner's dilemma anyway. The DOJ's conviction rate was only marginally lower than the criminal justice system of China, which is not a real system and widely known to be arbitrary, capricious, and empty. My grandfather, an engineer by trade and a life-long true believer in socialistic ideas, was appointed as a judge on the provincial level in 1979 in China. He asked to be allowed to transfer to a lower post a year later, and ended up in charge of industrial policy and scaling the silk filature system first perfected by JArdine-Matheson in 1865 but transferred to the CCP in 1952 when the company relocated to Hong Kong entirely, and then Bermuda. At the same time, through exercising some of the wide discretionary authority he had, he made arrangements so that his grandchildren, not yet born, would not have to stay in a country where one cannot talk about the manifestly obvious and progress was not made due to merit but luck and happenstance. Unfortunately, even though his grandchildren were in a position to be funneled into the party leadership pipeline in China, the youngest, me, ended up in a system that was, as it turns out, aspirationally more justifiable but in reality, fundamentally still just as unfair as the one he wanted no part in. He passed in 2023 and his state obit did not mention his actual achievements in detail. Even though I have almost no details in my linkedin I routinely get offers to practice law somewhere, and sometimes in fields that I have both experience and knowledge in. But as an ethical matter I have no desire to lend legitimacy where none exist. I have the privilege of not being the clown, and thanks to the choice capitalism afforded me, at least that is an option I can opt into.
To quote the the legendary drag queen Dorian Corey in the 1989 Documentary Paris is Burning, "I used to wanna change the world, now I just want to leave the room with a little dignity left". You may also have heard this in the 2006 film Shortbus. 20 years since, that is, ultimately, still the best most can wish for. Good luck UK, we had tried to put safeguards in place only to see them brute-forced in the most incompetent manner. We forked your repo and even though the codebases are mergeable, we can see the bugs, just as you can see ours. We tried to implement fixes and the patches are still there. Feel free to use the code, it's at least in the public domain, but how you merge that is beyond what we can help you with. The fact is in life as in chess, one may find oneself in a Zugswang. This can be cured, or result in a checkmate, and sometimes happens simply from the first mover advantage. In life sometimes one can see the system being built in a fashion where the outcome is that there is no act that can create a positive outcome, and the stakes are, inevitably, magnitudes greater than a game of chess. I find it curious that many people seem to believe that the counter to this is to rig the system to their advantage in the exact same fashion. I'm not playing that game. I pitched yesterday coach, I'm sitting this out.
fire alarms depend on detecting heat, and not humans, for this exact reason
I’ve worked several places where a handful of people left right after a manager left. And one where a combination of year end bonus, manager quitting because our best project got cancelled, saw a wave of almost a third of their developers leaving between Christmas and the second week in February.
Leadership is to a pretty large extent having good people skills, and the managerial layer has thrown a lot of noise in front of their sensors with the AI push that will not easily dissipate.
Humans are very good at sensing subtle cues from their fellow humans, but the blind perfectly smooth confidence of AI output being everywhere is interfering with that.
I lost job 3 times in last 4y because I couldn't deliver.
Piece by piece my desire to fight for good code died.
I'm in charge of couple micro services with high traffic where time is an essence. I understand them but I can't lie myself. This is AI generated slop with my name slapped on it and some day in the future someone will either curse on me or laught from me
The standard antidote to this is giving praises to managers even as that's not their idea and they did nothing.
That creates a situation where people are paid to basically not intervene nor block people, which is weird on paper, but makes a lot of sense in an organization.
Politics will always exist as long as unpredictability exists and is exploitable wrt Power.
Handling politics requires training, or as the religious institutions call it - Formation. Take a look at how long it takes to produce a Priest and whats in the training, cause their daily experience is dealing with people who disagree with their beliefs.
Then compare that process with what people are selected and trained for to be called "smart" in STEM.
Pretty sure this isn't a pay issue rather than it is a power grab by management and leadership. With AI, they want and can somewhat get the delievered yesterday. No matter what you do to document concerns, they won't care. Have 364 days of production pages, well don't get drunk and miss one on nye.
> There are, of course, exceptions (which is why a lot of things are still humming along),
Besides you are taking about smart people who don't give a shit. The main point of TFA is that those don't keep systems alive.
Then the name of VC capital, the company started hiring these imposters as leaders.
They are not smart people.
I initially spent some energy on trying to nudge them in the right direction, hoping they were at least somewhat capable.
Then more energy as I reviewed their insane AI slop documents. No substance, and they would not admit they had just generated them via ChatGPT.
Started documenting my answers in writing because they would not actually listen, and instead would just trust their chat bot and preformed opinions over facts. Maybe at least they'd feed my document into it I hoped.
But no, these people are worse than I ever imagined. They think they're smart, but really aren't. They mention AI in every sentence but don't understand it, and actually have the gall to talk over engineers even for technical things, spewing bs with full confidence. Whole rooms rolled their eyes.
I just couldn't believe it. Real scary. And ultimately if it comes down to it they win you lose. The business officially empowers them to "lead", pays them obscene amounts of money and until the company crashes and burns no one will ever listen. And then it's the engineers that get cut.
You're damn right I checked out.
Being good means you understand that you coexist and don't live in a vacuum.
You describe this as somehow a modern corporate problem but Baltasar Gracian described this very problem in his _1647_ book 'The art of wordly wisdom' which is a very loose translation of the original title from the English editions and Schopenhauer's German translation. The tone is somewhat calculated but Spanish medieval politics and modern corporate orgs still have a great deal in common.
1. No job security. You will be fired for increased short term profit.
2. Pay is low, many people can barely get by.
3. You are forced to work pretty much all your waking hours.
4. and now with AI, who knows what the future holds.
And people wonder why employees do not care about anything these days.
One of us would say a "smart" person is Bert Cooper type who is completely self-interested.
Another might say a "smart" person maximizes the outcome for themselves and for the group. (John Nash equilibrium)
Both of these might argue that they being "smart"
How ever we are in interesting times and this is so obviously the case it's now a cheap platitude. But If I say another platitude, it has changed everything and its been huge and depressing.
What the author is saying is however from a kind of despair. But they are smart and they are able to see stupidity, they should think of ways to use that to profit themselves and to push for the world they want. (And I am not saying this in a sort of 'if your so smart why are you not in-charge' on the contrary I think they are correct, but despair is a waste of themselves. While they are moving fast and breaking things, cultivate a voice of thunder and knowledge of the power they have and it's weaknesses).
people who give a shit as a first principle tend to get their heads lopped off soonest
This blog post is just another speculative hot take.
Here are a few facts that don't necessarily point to slop or "capitalism bad" as the culprit.
1. Code is moving faster. Code quality may have even improved or stayed constant, but simply more code means more bugs overall. 2. Malicious actors are moving faster. Malicious agents don't sleep, and security teams, and existing mitigation and prevention techniques haven't caught up yet. 3. Existing bugs, written by "people that gave a shit", are revealing themselves under new levels of load.
But yeah, "capitalism bad, they laid off all the smart people, and I'm smart" ... is a satisfying take... honestly kinda sad that these kinds of posts get upvoted.
They do because despite the unfortunate choice of examples, it's true that most people in software development are now demanded to stop giving a shit, and to use and trust AI for as much as possible.
While we still don't have many examples of big incidents caused by that, it takes a huge leap of faith to not expect it to end badly.
See guy putting out fire getting praise vs one that wrote their stuff correct the first place and it didn't caught on fire.
In other words, the better solution is good for the client, and not for the company. And a smart person cannot help but point it out, and the management cannot say out loud how they should not do it...
So kick the smart person out...
You can pore over your app, find 10 problems and have AI fix them. You can begin using the majority of your time on QA, product fit, and architecture instead of implementation and debugging.
In the past it’s been said one person alone cannot accomplish much. That is not really true today if the individual is competent and motivated.
If I'm leading a large project would I review every line a junior engineer write? No. It's not sustainable.
And that it's never been a large project leader's responsibility to review any code.
It has always been strongly advised to have two pair of human eyes on every line of code, though (either through reviews or pair programming).
But now it's suddenly not sustainable to have even one pair of eyes, right?
Oh, true, we can't possibly hamper the output of AI crap.
No, it has never been a worse time. If you care about your work, having the slop machine shoved down your throat by managers who (wrongly) believe it will make your work so much better is pure misery. Source: I care deeply about my work, and I'm sick to death of having to fight my manager about LLMs. I just wish he would judge my work on its merits (which he admits are good), not how I got there.
And now we have fully-autonomous AI agents breaking everything on a daily basis. Largely because of all that technical debt we were never allowed to fix.
Classic Prisoner's dilemma. Moral engineers receive less compensation than those who don't cooperate.
End result is immoral engineers making killer ai robots, or making nazi gestures.
Basically, they figured out how to cash in on nerds. And profit doesnt give a shit how many people it hurts or destroys along the way.
Ironically communism and socialism doesn't even recognize smart people who give a shit :-)
Article is a solution in search of a problem. A voluntary meritocracy (capitalism) is an amazing system.
It used to be the day that giving a shit mattered. Now, the less work you do and the late you arrive and the early you leave to go home, the better pay and position you get.
A colleague once told me, it pays the bill but I never truly understood him until now.
Pretending I don't give a shit improved my life quality by 10 folds, my weeks also feels like it lasts a day instead of 5(Mon-Fri).
In general and being a white male, society tries to fuck me from all directions with no fucks given, not giving a shit is now a surviving skill.
Let me tell you, a good QA person truly gives a shit. They care that the user has a good experience and that weird things don't happen and that the UX isn't confusing and have all sorts of innate human intuition about how other humans think that can't fit in an AGENTS.md.
Not just that the product "works" as ticketed.
We have been using this AI QA system for about a year and a half and it's literally never had anything to say about my work. I pride myself on testing my code really well but I know damn well I'm not that perfect. I'd almost never get through a 2 week sprint without human QA pushing back about a handful of things.
To the outside idiot it looks clearly like "look how much more productive everyone is!" We are truly doomed.
Author should read about the Nobel family and their various businesses in Russia and elsewhere. "Smart people who gave a shit" about their employees.
The architecture of financial system creates artificial demands for growth that inherently are orthogonal to the issue of private property and economic distinction between capital and labor. This singular "capital" needs to be unentangled and the specific financial component (which is distinct from the matter of ownership) needs a serious rethink for humanity going forward. Once we remove the mechanism (bug imo) that requires continual growth from the economy, it is possible for "capital" to think long term, since 'continuity' will now replace 'growth' as the key criterion for policy decisions.
In that time, our arrogance and stupidity, has begun a mass extinction event that's only comparable to a moon sized astroid impact. The consequences of this have hardly started to play out on geological timescales.
To think we're the apex is hubris on an embarrassingly naive level.
Many people can demonstrate that they give a shit. But people who give a shit in the "right way" dance between these different perspectives, and can accept that they conflict sometimes. They have amazing intuition about where the threats are and where the opportunities are and on a fundamental level, major setbacks or missed opportunities are simply too painful to allow to happen.
I'm not sure it's true that AI doesn't give a shit, at least concretely, as observed by how it behaves. I mean it's defining feature is endless energy for doing more work. I think the big questions are: is giving a shit the right way--the way that truly holds the world together--about a fundamentally human situation where you and your friends depend on this thing to survive? And if we really believe this trait isn't limited to humans, then is it actually possible to design AI tools that can do it?
That's why workers in jobs that engender vocational awe are underpaid and exploited.
We had a whole period during COVID where society as a whole was collectively reminded that there are people out there doing "essential services" who cannot make a living wage. And then we collectively forgot about them to chase AI.
I am so tired of having the mental disease of caring about doing things right, or considering carefully what "good enough" looks like.
In defense of giving a shit
I'm not very smart, but I always cared a lot regardless of what I was tasked with doing. Whether it was cleaning the pantry at my college job, or delivering software outcomes for our customers at my professional jobs.
I quit after a few years of realizing I cannot bring myself to do fake, performative work for promotions, and though the last couple of years have been a struggle, I do not want to go back to working at a company. I'd rather continue learning, failing, and eventually find success.
Being a person of average intelligence, I am used to having to work harder as my only "outlier" quality is that I do not give up.
Worst case scenario, I'll get a simple job and live off that salary + investment income.
You don't need to be smart to do anything in life basically, a better trained person will perform better than a smarter person 99.99% of the time.
It should be a cultural priority to give and have better education.
Also being passionate about "doing well" is very heavily a cultural thing in my experience but it isn't as mechanical as training. It even effects how a person absorbs the training so it is probably more important
IQ accounts for between 4-25% of the variance in job performance depending on the meta.
Training accounts for between 0-1% of variance.
There's no overlap.
Iq is very highly related to education and culture.
Also iq is not a stat number in a game. If you are above 110 then you can basically do anything except very difficult jobs.
Even if being smart is effective, we can’t optimize for that so there is no point in optimizing that. You can look around and see how effective we are at using the “iq” capacity of people in society.
Don't force business to care about something that they don't (it won't end well for you) Equally Don't force yourself to care about something that you don't (it won't end well for you)
If there's an overlap, find a way to prove your value in relation to that area and use it as proof of your value and leverage that to be able to deliver more of what you value.
> Don't force business to care about something that they don't (it won't end well for you) Equally Don't force yourself to care about something that you don't (it won't end well for you)
this is correct if you're the only person that cares about something
This really got drilled into me after working in blockchain because your software might work perfectly on one node, but when you have 100+ nodes, you can see race conditions show up and they create obvious discrepancies which can create forks in the network. In fact any non-deterministic logic will create obvious forks. Building software that's fully deterministic is very difficult and you have to go against a lot of what you were taught.
There are very few software engineers who can write good code and I think a lot of it comes down to caring.
From the engineer's point of view, there are conflicting incentives and unfortunately, businesses have been rewarding the wrong ones. Most business leaders think of a star engineer as someone who can churn out large PRs rapidly, solving issues one after another. It's a dangerous way to look at it. Such engineers are almost certainly creating a lot of tech debt. TBH, I don't think this makes sense, even for startups, because the tech debt comes due much sooner than you'd expect and the costs compound rapidly and usually permanently.
The irony is that large PRs delivered at a rapid rate makes it look like the engineer cares about the project's success, but actually it's often the opposite; they do it subconsciously because they care about themselves and their lock-in factor. They know that complexity is how they can claim ownership over the code base.
I observe a similar dynamic with business people; some business people are laser-focused on value-creation, but many others are laser-focused on retaining control and will happily throw value-creation under the bus at the drop of a hat.
Unfortunately I feel like the past 10 years have significantly favored the value-extractor at the expense of the value-creator.
Also, I feel like AI has been creating more harm than good, in the hands of the average developer, because it tends to go for the complex solution and has a tendency to patch things up with more complexity instead of avoiding the problem altogether. AI agents do not value determinism enough. It's crucial to producing correct software.
Yeah, but I've worked at places which don't even care about the tech debt. Unless it's affecting the bottom line, we as engineers are implicitly told to just put up with it. From a business perspective I get it, though it's short-sighted... but from an interpersonal perspective, it's not right.
Once you hit 30s, this tendency to give a shit will drop off dramatically, and by the time you hit 40s when your career has less years ahead than it does behind, many will no longer give a shit at all.
Thats what you hire and pay heaps of money for so that is obviously what you need.
(Agree with the article of course)
This resonates with me. It feels insulting when one is told not to use your own brain and judgement, and to defer entirely to machines. It puts humans below AI. It's dehumanizing.
At my job we basically have unlimited AI budget and nobody is forced to do anything. We are all constantly negotiating our impact on one another.
I think software engineers should be given AI with the understanding that it is a tool, and we should be given the freedom to use our own judgement or wisdom with regards to when and how much AI we use. I think it should not be forced on people nor should people be coerced to use it... if I'm doing my job and hitting targets, why should management care how I get there? In fact, I do use AI constantly, but there are times when I don't need it because I can converge to a "good enough" solution without it. I think it's a matter of management trusting their people to make the right decisions and giving them autonomy.
Well, that and to create a being with an intellect greater than a human, put it in a robot, and enslave it to the whims of a few capitalists for all time. Somehow this isn't slavery?
> It matters because generative AI isn't smart, and it doesn't give a shit. What seems like intelligence is an illusion propped up by a massive amount of data strip-mined from the commons with weights that nudge the output toward a generally desirable shape. What seems like care, concern, and attention is just generative AI doing what it's designed to do: telling you what you want to hear. The math and logic that power LLMs are as cold, unfeeling, and indifferent as the laws of physics.
> Smart people who give a shit notice things. They connect dots. They notice when things don't feel right. They know when something is about to go off the rails, and they know how to get things back on track. They know how to handle edge cases. They know how to manually handle a unique situation. They know what the right thing to do is, and they know when they need to put in extra effort or time to make something happen. Generative AI falls far short in these areas, but this is the glue that holds complex systems together. More prompting and more output isn't a substitute for this.
There is a lot of hype around AI, and disagreement about what it is, what it's capable of, etc., but absolutely none of the smart people I know are clinging to the idea that "generative AI isn't smart", that its intelligence is just an illusion and that it only tells us what we want to hear.
Is it smart in the same way a human can be? Of course not. Does it share our emotions and morality? It seems the answer is no. But this idea that the latest generative AI models are just black boxes of math that produce an illusion of intelligence is just weird and I don't know how an intellectually honest person using them could claim it's nothing more than that.
That is literally how they work. They aren't smart, they can't reason, they can't learn. A statistical model of "what is likely to come next" has its uses, but don't ever mistake it for intelligence.
It's strange to me that people keep believing this simplistic understanding of AI and applying it to the current state of the technology. Some of the top researchers who helped develop the foundations of this technology, including Nobel Prize winners, routinely explain how we're way past that point.
And here we have the basic disconnect: this is the literal truth of what the tool is, and what it is doing.
It seems that the difference between intelligence and the simulation of intelligence is too subtle for many people to recognize. Current tools are not smart and they do not reason. They simulate reasoning by producing a train of steps that looks like the most likely selection from the training data.
When that was the "smart" thing to do for the current request they appear to be smart. When it was not they reveal the bias induced by their training data and they look fragile and wrong. It is the uncanny valley but for reasoning.
I would continue but I have to go back to arguing with Claude that almost every line of code that it has produced reveals an incorrect unstated assumption about the system that we are building and/or how code should be structured in general.
Yes, it's way over the heads of Nobel Prize winners whose work helped develop the technology.
> Current tools are not smart and they do not reason. They simulate reasoning by producing a train of steps that looks like the most likely selection from the training data.
It's interesting that you seem to know how they "reason" when it's a hot research topic.
https://news.yale.edu/2026/10/08/how-do-ai-chatbots-reason-h...
> When that was the "smart" thing to do for the current request they appear to be smart. When it was not they reveal the bias induced by their training data and they look fragile and wrong. It is the uncanny valley but for reasoning.
That sounds an awful like humans.
> I would continue but I have to go back to arguing with Claude that almost every line of code that it has produced reveals an incorrect unstated assumption about the system that we are building and/or how code should be structured in general.
It's ironic that you are using a technology that isn't smart and can't reason to do your job.
> Current tools are not smart and they do not reason. They simulate reasoning by producing a train of steps that looks like the most likely selection from the training data.
Let's start with "they're not smart". Geoffrey Hinton, who has a Turing Award and Nobel Prize in Physics to his name, and was involved in the foundational discoveries underpinning modern AI, disagrees:
https://theconversation.com/ai-pioneer-geoffrey-hinton-says-...
On "reasoning", whether what AI is doing is "reasoning" or not, and how there might be forms of "reasoning" that are different from ours, is currently a subject of intense research, as evidenced by the Yale study I referenced.
You provided absolutely no evidence for "current tools are not smart and they do not reason." If you have an evidence-based argument to make, make it. Because dismissing anything you disagree with as "not real" is about as convincing as "AI is dumb. Trust me, bro".
> While he does think AI has the capacity to become smarter than humans, he also proposes it should be thought of as an altogether different form of intelligence to our own.
The capacity (i.e. potential) to become smart is not the same as being smart. But even "Hinton believes it" is not a very convincing argument: it is an appeal to authority. This is different from a reasoned argument.
It's not so interesting to discuss what other people think, e.g. second-hand opinions. You've made the claim that LLMs are smart, and that it is ridiculous to think otherwise. Do you have an argument to support this claim, or is your support that you believe that smart people think this?
You referred to this statement:
> While he does think AI has the capacity to become smarter than humans, he also proposes it should be thought of as an altogether different form of intelligence to our own.
Then you wrote:
> The capacity (i.e. potential) to become smart is not the same as being smart.
I don't know if your mischaracterization of the words here is intentional or not, but let us parse the language clearly: Hinton did not say that AI has the potential to become smart. He said that he believes AI has the potential to become smarter than humans. The clear implication: he believes it already is smart.
Inherent in the statement "different form of intelligence to our own" is the belief that it already is intelligent.
> You've made the claim that LLMs are smart, and that it is ridiculous to think otherwise.
I actually never stated that it's ridiculous to think otherwise. I've just pointed out, multiple times, that you haven't provided any evidence for your claim that "current tools are not smart and they do not reason."
If it's so easy to prove, why don't you do so?
> I actually never stated that it's ridiculous to think otherwise.
No, your literal verbatim quote was that it is intellectually dishonest. No doubt we disagree on what you were saying in this previously quoted section as well:
> But this idea that the latest generative AI models are just black boxes of math that produce an illusion of intelligence is just weird and I don't know how an intellectually honest person using them could claim it's nothing more than that.
I don't see any interesting discussion to follow when your interpretation of what words mean is so different to my own. Have a nice day.
Hinton's views on this subject are well-publicized, and anyone who isn't familiar with his work can easily read about them, watch his interviews, etc. The fact that you apparently haven't, or dismiss his views outright, speaks volumes. You can argue that citing him is appeal to authority, but ignoring the views of a Nobel and Turning recipient whose work was foundational to modern AI is appeal to ignorance.
It also speaks volumes that you repeatedly avoid providing any evidence for your claim that AI is not smart and cannot reason.
Generative AI started with lots of "knowledge", and has since been trained on lots of practical examples to give it lots of "experience". AI has significantly more "knowledge" and "experience" than a human can attain during his lifetime. But it isn't and can't be "intelligent".
It can't "think outside the box", it doesn't have "intuition", it doesn't know how to "ask the right questions"... It can simulate all three if the problem you point it at is related to the "knowledge and experience" it has, or by the sheer fact of letting it run a long time until it figures out the right solution by just analyzing lots of unneeded paths inside its 1M context window.
If you're smart and want to see it's "non-intelligence" in action, just turn on the extended thinking summaries and watch it work in real time on some harder task, like debugging some non-trivial bug. It's initial intuition is often very wrong, it often presumes something to be true without asking or checking, or double checks stuff that do not need double checking, if it sees contradictory evidence, it can't properly assess what's going on...
The thing is - if you don't see this (and most people won't cause by default it doesn't show you its process, you have to turn it on) - and just let it run for an hour or two, it often eventually does arrive at the correct conclusion - and you may consider it smart for figuring it out, even if you yourself could've figured that out 2x faster, but were too lazy to dig deep.
Another very good example of AI's non-intelligence is its tendency to over-engineer solutions, especially on smaller projects. That's again caused by its lack of intelligence. For lots of coding tasks it produces 3-10x the amount of code needed, because it also thinks about "edge cases" that can't happen in the project, it assumes the project has much bigger scale than it does, etc... Lots of people call AI "smart" for this, because "it thought of things I haven't thought of", not realizing they're unnecessary, and thus being the "opposite of smart".
That's what's meant by the "illusion of intelligence".
Have you never worked with a human being whose initial intuition is very wrong, who presumes something to be true without asking or checking, who double checks stuff that doesn't need to be double checked, and who is misled by contradictory evidence?
> Another very good example of AI's non-intelligence is its tendency to over-engineer solutions, especially on smaller projects.
Have you never worked with a human being who over-engineered a solution?
> For lots of coding tasks it produces 3-10x the amount of code needed, because it also thinks about "edge cases" that can't happen in the project
Have you never worked with a human being who wrote way more code than was necessary?
Because I have also worked with actual smart people, and what makes a smart person stand out is their intuition, analytical thinking, and their ability to realize when they don't know something (which is another thing that generative AI just can't do and is pretty much the reason why they presume something to be true when it clearly isn't, and why they're mislead by contradictory evidence).
Overengineering and writing way more code than necessary is again a question of intuition - smart people sometimes start overengineering stuff, but oftentimes they realize mid-implementation that this is way too complicated and simplify it before opening a PR.
So are you suggesting that those people are fundamentally different kinds of human beings? That their brains don't share the same structure yours does, or function the way yours does? That they don't have "human intelligence"?
Are you suggesting the AI is similar to a "dumb" person, or something else? Because there's a huge difference between a "dumb" person and a "dumb" piece of technology that can't "think" in any way, shape or form.
> Overengineering and writing way more code than necessary is again a question of intuition - smart people sometimes start overengineering stuff, but oftentimes they realize mid-implementation that this is way too complicated and simplify it before opening a PR.
Putting aside the point that conversations like this seem to devolve into subjective code review as a test of intelligence, do you think anybody you've worked with has considered you not smart? Thought you overengineered something, or wrote more code than necessary to accomplish a certain task?
> Businesses need to value smart people who give a shit
Because my coworkers valued me but no employer ever did. They hated how readily I found flaws in their arguments and disregarded internal politics to get shit solved.
I can’t believe it’s not fraud, is absolutely rampant. Lie about compensation, lie about kpis.
Many big companies end up working like old Soviet factories, everyone fudges the numbers just a bit. All until one day it falls down.
They might layoff your entire division, anyway. Not like you could have stopped it.
When I was a mechanical engineer we had standards
For systems built by software devs there is only business risk preventing huge very real damages
Maybe we need a Ford Pinto moment for software, or maybe we've had one and didn't care
What is needed is dual price vector strategy where you have to pay for the right to demand a service. The options premium covers the running costs.
Yes, neoclassical economics when run by bounded rationality is inherently inefficient and can never lead to any welfare maximization with singular price vectors, it's practically impossible.
The fact that they did so anyway, with an omnipotent solver [0] in each agent just tells you how dogmatic they are.
[0] Let me explain the trick. You solve the economy to perfection, possibly via infinitely complex price vectors that perfectly capture the entire universe, then you map it back to a scalar price vector. Ergo, free markets are efficient and scalar price vectors are sufficient.
The most popular savior movement has this platform:
Kill all capitalists, then let the state become the most powerful capitalist, more powerful than even the capitalists you have to kill along the way.
I don't really know how to convey it, but these people are not interested in solutions to their problems, even if the solutions are powerful enough to make capitalism irrelevant.
Capital is irrelevant. If you believe capital to be powerful and productive and that capital can produce capital, then you run into a big problem: everyone will inevitably have capital.
What people might not have are the things that cannot be reproduced. Specifically land or energy.
At the end of the automation game you have a machine that takes a lump of resources or energy and then spits out exactly what you want.
At that point owning "capital" is irrelevant. I don't care about capital. It'll be as cheap as fridges or TVs, but you still do not want to own it anyway because it is even cheaper to farm out production to split the fixed costs. Trading won't be gone, it's just that people will trade products derived from land with zero labor input.
We can try to blame many things for this and say it's just bad luck ending up at crap companies, but I don't believe that. In IT, things like scaled agile have made not thinking about what you're doing so systemic, that it's near impossible to find a company that does want you to care.
Honestly, looking at companies I've worked for or interacted with, I'm surprised we haven't had any really big hacks or outages leading to meaningful societal disruption and loss of life.
Mission is an integral part of our psyche.
This feels incomplete to me. AI makes it hard for leaders to determine who is smart and cares, vs who is a shameless self promoter. Our leadership class did not come up as ICs when AI was ubiquitous. They don't know how to make the distinction and reward the right people.
"It is the capacity for maintenance which is the best test for the vigor and stamina of a society. Any society can be galvanized for a while to build something, but the will and the skill to keep things in good repair day in, day out are fairly rare. […] I read somewhere that in ancient Rome a man was disqualified as a candidate for office because his garden showed neglect."
“Lets get rich fast screwing ppl over AFAP, then bail and take no responsibility”..
Hard to say otherwise looking at current state of affairs.
No matter how rich and plutocratic the US tries to be, everyone gets 1 vote. Literally all the problems talked about can be fixed much quicker than you'd think if we could get people to look outside their immediate selves for a few weeks.
Of course we do. That's not the distinction made, that of voluntary vs paid workers doing the maintenance.
It was about societies being able to do the maintenance vs those that aren't. Our late stage ones spend many times the resources, and collect many time the taxes, but can't maintain shit.
I replied on a whim because recently some people have criticized decolonized third-world countries for failing to maintain the infrastructure that their colonizers left behind. These critics see it as evidence of an inferior culture and/or an inferior people. But I think that's false, I couldn't disagree more that such countries would be examples of societies that, in Hoffer's words, would lack "the will" to maintain what they have. Because a good number of these countries remain underdeveloped and poor while external actors and a corrupt govt extract their natural resources.
Without a functioning government and population making enough money to pay infrastructure taxes, there can't be long-term maintenance. I think your example of late-stage capitalist countries also wasting resources emphasizes the importance of effective leadership.
if there were no taxes, people would mostly hog up and spend their wealth personally, with little incentive to give it up towards work that doesnt directly impact them but which may be crucial. Similar thing plays out on a smaller scale with open source projects.
Meaning if you don't have systems in place to keep your business or finances etc to continue working when shit hits the fan, it doesn't matter what you're prepared to achieve.
The capital has historically decided that it's preferable to have people deeply involved in operating these semi-functional complex systems. The slider has been further towards "mechanical turk" than most realise, especially in payments. The costs to automate the last 5-10% are more than keeping some folks on to crank some levers. Also, some of these "smart people who give a shit" have enabled this behaviour because they prefer being the "irreplaceable SME" who keep the Turk operating.
Maybe this will change? Certainly the capital would like to do away with irreplaceable SMEs, especially with Boomer cohort retirements in full swing and the GenX cohort coming up. Probably the slider is going to settle a bit further over towards automated, but we'll see.
If you want smart people to care, you have to first got their back when they do. My experience is that caring is punished in many workplaces, be it by collegues who think they will look bad if someone else does something, be it by managers that sell something as their work when it works, but sell it as yours when it fails, be it against metrics that punish quality by promoting quantity, be it by punishing those that do their work well, by pushing all the work on them.
If you want smart people to care, caring has to pay off in the sense they care about. That is different for each person, but one perdon might value freedom, the other might value their free time, the other might value money or visibility, or a mixture of the aforementioned.
The driver's at the wheel, he has one hand on the wheel. He's supposed to be alert. He's supposed to check that the car is doing what he expects it to do. But you know, after 200 miles of no problems, maybe he can read the text that just came on his phone. Ah, and he should probably tell his mum he's going to be late to the family dinner. The driving? Yah it's fine. And slowly the driver who's instructed to pay attention stops paying attention. He's a good driver in a regular car. Never had an accident. But this tesla... It drives itself. It just works. And that's when the car hits the truck.
Agent coding is somewhat similar. It does the coding, it checks it works, and 99% of the time it works great. So we stop looking...
I've started exploring the use of AI and there is simply no sense of joy or achievement, even if I actually can do much more with AI now than I did solo in the past. Back then, every little thing I managed to figure out myself (either by reading the documentation or by experimentation), every little win brought me a bit of joy and made me want to celebrate. All of that is gone, and the current job market simply does not allow me to give a shit - every job post demands AI, and a lot of them will want you to run multiple sessions in parallel. @MiaAI_lab on X said that "no ADHD is a disadvantage in AI", and I tend to agree. I interviewed for a job a few days ago and the girl on the other side is not even senior - she tells me she has ADHD and is able to drive 10 agent sessions in parallel. To me, it sounds like ADHD means she's not going to give a shit for any of the running agent sessions, which is somehow a superpower since 2025. I'm in my 40s with more than a decade of experience and I don't think I can ever work like she does. It's just really scary and disheartening because I feel like I have become unemployable.
From my experience and my friends' anecdotes - most (though not all, thankfully) management is in complete AI FOMO mania.
Here's my comment I made a week ago about this, just to not repeat myself constantly: https://news.ycombinator.com/item?id=50438343
Work requirements (pace, deadlines, quantity) are such - that there's not time to give a shit. Like you said - open bajillion LLM agent sessions and open the token firehose.
It does not mean “multi-tasking machine”, it means: “person who can’t convince themselves to start, or stop, any task with the exception of emergencies or things they might believe are novel”.
I’ve never been able to control what I work on, no discipline or desire has helped.
The idea of trying to do 10 things at once makes me physically ill.
Ouch, that is totally me.
Except, sometimes I succeed to convince myself to start, by staring at a task for a long time, trying to divide it to subproblems, generally walking around it and touching it from various sides, until something triggers my curiosity and the task becomes attractive. At that moment usually a meeting on an unrelated topic starts, or someone tells me to drop whatever I am doing and focus on the new priority.
For example like you with AI sessions, I thought the way "fullstack" developers worked was something I could never do. I just hated React from the moment I started learning about it. I could see that it was common for people to be either good at frontend and write dodgy backends or the other way around but one had to pretend one was excellent at everything. Luckily there have been jobs here and there which are not locked into what I feel is the misery of the next webapp - wading through reams of dreadful code in which some bright person decided that comments were a sign of bad style so ** you if you don't understand the superb clarity.
There's also a lot of ageism and a new religion is how one attacks and undermines ones established competition: a new language, some programming paradigm like agile or SOLID - whatever.
I've seen this before AI. I've used python since 2006 but I got these interviews where I saw myself being mentally crossed off the list at just the point where I decide not to use a list comprehension or can't remember some function name that's the favorite of the interviewer. If you have read a lot of shit code that uses fancy features for trivial results and you're tired of it being hard to understand then you're obviously not in the right religion.
Now it's AI: "Bye-bye grandpa - we don't need you or your expertise." A lot of code will be written by "fleets of agents" that turns into tomorrows janky mess that the next generation of whatever-you-call-them will complain about. All the dumb things that people couldn't do in the past because of constraints....they'll do now just because they can.
Can one find somewhere to survive on the margins? I hope so. Some of the software we use has to work properly even if perhaps a great deal of it isn't important enough to need to be that good.
I worked for a place once where the billing system was janked and fell over regularly and lost money - do you think anyone really cared? They bitched and moaned about the effort needed to fix it because it ate into the time available for the next feature in the web app.....which was used by about 5 customers per day. If that's pre-AI can you be all that surprised about post-AI?
The problem there, aside from the obvious inability to solve real problems or provide any kind of originality, is a culture of cowardice and sociopaths. When talent and/or originality are wholesale replaced by commodity conditions most people doing the work will fear doing more than the absolute minimum to solve problems because their existence is qualified by popular conventions that discourage deviation. That’s where imposter syndrome comes from. For some people this is just a game to manipulate for self-satisfaction, which is logical when hiring/firing is prioritized over product development. That is where the sociopathic behavior becomes normalized.
So they made 4 successful(ish) games, two of which have massive player bases (Deadlock, CS2), HL:Alyx was a commercial success (2 million copies sold as a VR game), and Dota Underlords which was also successful at least in player count (200k players at peak) but got abandoned when it comes to development (unknown reasons, probably due to having no idea how to monetize it).
So they made games and even profited from them recently.
Investors get suck on "game company make games" and are unable to see the bigger picture.
In game theory terms this would seem to be the problem of bad acting overcoming good [3], arguably because it's better incentivised.
[1] - https://en.wikipedia.org/wiki/Peter_principle
[2] - https://ribbonfarm.com/series/the-gervais-principle/
[3] - https://ncase.me/trust/
"The result is the MacLeod hierarchy. It may be horrible, but like democracy, it is the best you can do. The Sociopath (capitalized) layer comprises the Darwinian/Protestant Ethic will-to-power types who drive an organization to function despite itself."
and
"The Sociopaths enter and exit organizations at will, at any stage, and do whatever it takes to come out on top. The contribute creativity in early stages of a organization's life, neurotic leadership in the middle stages, and cold-bloodedness in the later stages, where they drive decisions like mergers, acquisitions and layoffs that others are too scared or too compassionate to drive. They are also the ones capable of equally impersonally exploiting a young idea for growth in the beginning, killing one good idea to concentrate resources on another at maturity, and milking an end-of-life idea through harvest-and-exit market strategies."
But I think it's also worth remembering that at the end of the day, the "company" is just an ever rotating collection of individuals all operating according to their own psychology and incentives so there is no benevolent dictator overseeing the whole thing and trying to motivate it differently.
I guess from a shareholder point of view they are also the ones most likely to be focussing on the fiduciary duty to maximise shareholder value (whilst it aligns with their own) which in a competitive world probably requires exactly those dark triad manoeuvrings.
[1] - https://ribbonfarm.com/2009/10/07/the-gervais-principle-or-t...
(Possibly the Peter principle etc always win at some scale)
I've done it on various occassions even if it sometimes lead to being one of the first to be made redundant, I learnt more, enjoyed it more and ended up being hired by some of the people I used to work with. It seems like a slightly more enlightened/less cynical version of what Gervais is talking about. But this was a few years back, I can't guarantee the current climate is as forgiving?
I haven't found this to be true at all. When the A-team gets called in for high impact engagements, it's typically smart people who give a shit. They're respected, paid well, and go the extra mile to ensure success. Working with these people is a joy, and they're anything but a liability to their company.
> Most companies want compliant workers who shut up an do what they are told.
You can be compliant and still effect change. Nobody likes a cowboy who ignores directives, and nobody likes a sullen employee who says nothing when they see a problem. Good employees are dynamic enough to follow orders while communicating effectively about concerns.
I'm not saying this is always true across the board, some companies are toxic dumps and the good people leave, but I doubt today's reality doesn't reward the good folks who give a shit. Hiring is hard, and when you find good people, you do what you can to keep them.
There you have it. It is about power structures mainly. The "A team" has a mandate which the scapegoats lack.
(And yes there are exceptions and I am somewhat hyperbolic but I have never found average incompetence to be a problem.)
> never found average incompetence to be a problem
... uh, ok, let's agree to disagree.
I think it's more complicated than that. Ultimately human relationships play into human decisions. I've often seen extremely competent engineers be pushed out of companies because they had personal disagreements with their leadership. There's plenty of senior engineers at my company who I don't think are very technically competent, but they navigate the politics of the org with aplomb.
Maybe I'm bikeshedding what "merit" is here, but in short I think it's perfectly reasonable that there be other factors for why the "A" team is popular beyond their technical acumen.
> I have never found average incompetence to be a problem
Ever got the front lines of a call center that can only read from a script and wastes your time? Many would point to incompetence, which tangibly improves as you escalate up through the support tiers.
I've had big corp send two on-site engineers to turn up new equipment, who clearly didn't know what they were doing and failed after two weeks. Escalation to executive level got us an A-team guy who got it sorted in a day. Both teams had the same mandate. Guess who was enjoyable to work with, went out of their way to help us with other problems later on, and indirectly facilitated a big deal between several companies by being smart and giving a shit.
Once I spent a year getting punted around by B-teams trying to get access to a feed that nobody wanted to deal with because it wasn't standard. It might not have been incompetency, but it certainly was lack of initiative. It went critical and I got the A-team, and had the feed a week later because the guy was smart and gave a shit.
My point is these people exist because they're needed and valued, not because of some petty power structure. They exist in many orgs, and pull more weight than most. Even as power structures change, most incoming chiefs are savvy enough to not throw away such valuable resources if they can help it.
There is a lot of situations where the evidence to not do something is strong but you still take that gamble and force it otherwise with repercussions.
Like Uber’s A-team that rolled out the service while operating illegally in their beginning, making apps that request fake rides on other services, and having engineers building the God View service that monitored people. Arguably parallels of that have happened today at AI companies like Open AI and perplexity, Tesla, Meta, and like.
I don’t think it’s wrong to say smart people who give a shit are a liability for companies, maybe not most but they definitely can in tech. Too many whistleblowers and investigations that say otherwise.
Right. Warning all the AI die-hards and believers is of no avail, because either they don't give a shit or they "know better". Learning will only happen the hard way.
The article throws out some valuable things that people who "give a shit" do, and which AIs don't. But depressingly there's nothing in there that an AI couldn't do if it was told to.
If capitalism does discover that "giving a shit" was an attitude that improved real outcomes in the long term, nobody would go and re-hire all the nice people, they'd simply tell/train the models to "act like someone who gives a shit"..
Showing 400 of 423. Read the rest on Hacker News
Comments are loaded live from Hacker News and are not stored by Mid or Real.