Yep fundamentally different from back end because you don't control the client.
In theory the client can do literally anything (for example, someone could be using an ancient version or have some features voluntarily disabled) and you have to decide which things you're going to handle
It’s an absolute clown show that it was down for 4 hours, but would you want to be flying in some of the busiest airspace in the world on a “backup” ATC system that isn’t normally used?
I mean one could do A/B weeks. Have two separate systems that do the same job managed by two different teams etc. and switch between the primary system every week. This way you can ensure both systems are stable.
I’m way more often than not on the side of refactoring but there’s risks on both ends with a mission-critical service like this. Neither side is zero risk, the choice is about which option will result in fewer and less severe outages.
Yeah that's kinda my point. I'm not sure if the models have gotten that much smarter, but they're certainly getting more capable. That's not the same thing though.
There are things GPT 6.0 can accomplish for me that 5.3 was not able to. But there are also things it still fails at, and it doesn't seem to be much better at the big picture. It does spam about 100x more tests though and I wonder if just RLHFing it to test everything constantly is carrying it more. 6.0 writes so many tests and spends so much time verifying it's work in python sandboxes. Slow as hell but it tends to get things right the first time more which is good, I guess. I don't love the thought of a 500loc feature adding +4000loc due to tests though.
Yep same with latest Claude models - code isn't really any better than Opus 4.5/4.6, but use 5x as many tokens doing random stuff that's mostly unnecessary.
And yeah still for some reason they often can't understand how to set up any project locally without handholding, which is something you'd think an LLM would actually be good at
French people should be occupied by Guyanans who have no historic connection to France. Apparently that makes sense to some people when it's the other way around.
Batteries don't solve the storage problem, but surely you know this.
China is pursuing the strategy you're describing, but there's a reason they're also building nuclear, they're increasing LNG infrastructure, they haven't phased out coal, and they've built a huge amount of oil storage.
Solar and wind can be part of that strategy but it's not so simple, otherwise people would do it
Bateries do solve storage problems. And yes, china energy strategy is based on wind and solar - their portion of China energy production is massively growing.
Batteries can solve short term (days to a week at most) storage problems, they don't solve seasonal storage problems - making sure you have enough energy all the way through a cold, not very sunny, and not consistently windy winter.
If you don't believe me, run the numbers of how many kg of lithium batteries China would need to supply an entire week of energy in the middle of winter (all energy, not just current electricity demand, because in this scenario we're assuming everything is 100% electrified).
Also if you're potentially going to be fighting a war, you need more energy, not less, and it's even more important for it to be reliable year-round.
This is why China is doing what they're doing, they're not idiots and they know that wishful thinking is not a strategy
You dont need "very sunny" for solar to work. If you need to invent dark windless winter to make a non existent issue maybe you are just too deep in bullshit.
> China is pursuing the strategy you're describing, but there's a reason they're also building nuclear, they're increasing LNG infrastructure, they haven't phased out coal, and they've built a huge amount of oil storage.
This opinion is rather simplistic. China is reported to have 60% of it's energy needs already supplied by renewable sources, but this transition creates different types of challenges such as preserving the power grid's inertia. Otherwise you might see yourself in a scenario such as the Spanish blackout. This means having in place strategies for fast response (gas) and baseline (hydro storage, nuclear) to prevent the scenario.
> batteries are the fastest response power. gas is high maintenance, & bad capex as an intermittent source.
Batteries have a limited capacity and have a higher investment cost. Gas is the standard option for fast response. Pumped-storage hydro has by far the highest capex and has a heavy environment impact.
It's baffling how people show up in this sort of discussions arguing that standard tried-and-true strategies somehow don't work. Quite odd.
Batteries and solar have been beating lifetime cost of fast response gas turbines for five years or more. Now with uncontrolled fossil fuel cost spikes the operating costs of gas are highly variable compared to the pretty much flat investment and operating cost on battery + solar. if one were comparing battery plus renewables for full seasonal offsets are tighter, but the costs are so much lower, you can provision more and plan to dump power in the summer (or maybe not as this last long hot fossil fueled climate change summer will drive new demand for a/c)
Also Sam, Dario et al have constantly (like literally, every single week at least) been telling us that AI is going to make all of us unemployable and have to live on welfare.
Is it any wonder people hear that and go "hmm, no thanks" when the prospect of a massive whirring data centre next to their house is raised?
The SV LLM crowd know a lot about LLM psychology but seemingly very little about human psychology
> The SV LLM crowd know a lot about LLM psychology but seemingly very little about human psychology
I doubt that it's ignorance on their part. I think their target audience with those comments is the US government, basically setting up OpenAI/Anthropic as "too big to fail" in the sense that "if AI is _inevitably_ going to take away everyone's jobs, then we'd better make sure that it's US AI and not Chinese AI or some "rogue AI" that's doing it, and the government should 1) be in bed with us, 2) provide a financial backstop, and 3) not regulate us or put on the brakes so that we can stay ahead of the Chinese.
They don't care what the "normal" American thinks.
But the question shouldn't be, do you want datacentres or not? Datacentres are gonna be built in China or elsewhere.
The question should be, do you prefer to have them also built in the US or do you agree to have China (or another country) controlling AI in the future?
If AI is really gonna make everyone unemployable, then it seems to me like you would want them to be built in your country, not in another country, especially not one you disagree with.
Of course they're still the question of where exactly there are built and how to compensate for the nuisances they may bring.
They've fired up the same irrational hordes that brought us flat earthing, anti-vax, and 5G Covid. It's just another parade of extreme ignorance. And yes, the SF crowd is comically oblivious to what they're dealing with, ivory tower style.
They're detached from the great unwashed masses. It's not that they don't know anything about human psychology. They don't care about those people or their concerns, historically. And that's what is coming out in this interfacing with the broader public.
The comparisons to anti-vax and flat earth are absurd.
And what exactly is the benefit to you to having a data centre specifically in the area you live? Seems to me that far from “irrational hordes”, the protestors are standing up for their interests above the interests of the AI companies.
Opposition to AI data centers is completely rational, which is why both liberal and conservative folks are against it. These data centers are being forced on communities with soon-to-be-former leaders agreeing to nondisclosure agreements. They raise electricity prices. They are being slapped together with poor water recycling, and noisy diesel generators that pollute. The data centers for AI use a lot more resources than traditional ones. And the tech oligarchs think they can ram rush these things through, but they can't. People are sick of the lack of transparency. They're doing it all wrong to save a buck and a minute.
Ultimately the common thread in most of these cases seems to be a lack of purpose and meaning in life in general, not really an issue with the specific work environment.
The purpose of life is to make meaningful human connections, become wiser, become a better person. This is something that some combination of family, traditional social structures and religion used to provide for a lot of people.
"Create shareholder value at my current startup" obviously isn't going to scratch that itch in the same way. It's no wonder people end up feeling "burnt out" - that's your brain telling you that what you're doing is ultimately pointless, and your brain is absolutely right.
Treat work first and foremost as a means to an end, not an end in itself, and generally you'll be happier. You can (and should) take pleasure from your craft and a job well done, but honestly who cares how your Series A goes.
"Work coaching" or whatever this article is advertising will probably not help you at all
In my case, my work is being devalued by AI.[0] The same effort nets smaller and smaller returns every month, even though the work is _still_ necessary, because AI does not do original investigative work. It's hard to put work into something that might vanish in a few months.
I'm starting to feel something a lot like how people describe burnout. I need to do a bunch of different things to stay afloat, but I ain't got any gas left in the tank.
I find pleasure in other aspects of my life, but I still need to put a roof over my head.
I'm feeling the same way, and when I brought it up to some friends in tech it turned out in our group of 8 I was the least depressed.
The wife of another friend of mine confided in me that AI replacing his work in such a short time has made him depressed to the point she's getting scared he'll hurt himself.
I think we need a support group or something. But I'm too busy burning the candle at both ends trying to "catch the lifeboat" before its too late. So are most other engineers I know.
I'm sorry you're going through this but know not only are you not alone, but you're in the majority of software developers at this point (I think, anecdotally).
Thanks for writing this. This is how support groups start, when people find out that they're not the only ones going through this.
When I opened up to friends (and wrote that post), it encouraged a whole lot of people to also open up. We shared tricks, talked it out, and just felt a little bit more seen.
> The purpose of life is to make meaningful human connections, become wiser, become a better person
That maybe your own opinion. The older I get the more I realize that the ignorance is bliss - the less I know the better, nothing drains my energy more than human connections and I don't have to be better or worse, I'm just perfect the way I am and if anyone doesn't like that, it's their problem and not mine.
Yeah, I'm not sure I find either your or the parent comment's experience totally reflective of my own, but I definitely agree that I don't think the "purpose of life" is a very well-agreed upon thing. People have debated it for pretty much all of human history, and I'm not sure how anyone can reach any sort of conclusion that isn't fully personal to their own experience (I guess I lean towards existentialism)
Ironicly, in these cases more work can help the burnout. When you work 80 hours per week you are so busy you never notice you are not making meaningful human connections. When you only work 30 that leaves plenty of time to look at your own life and realize it isn't meaningful.
That shouldn't be taken as a sign that working more is good. It is a sign you need to fix your personal life which isn't easy. Work can be a useful part of making and keeping personal connections, but it shouldn't be the only part of it.
same logic applies to using drugs to self-soothe. if you're high all the time, you have no time to realistically engage with the material realities of your socio-emotional life and health. instead of taking the time to reflect on past actions and develop insight on how you're relating with others or living your life, you just work instead. and, in the meantime, dissonance compounds as background radiation
eventually that just leads to an even bigger crashout, burnout, break. Sinclair Lewis parodied this as early as 1922 in Babbitt [0], depicting a man who is economically satisfied and fits perfectly the mold of the white picket life and yet is devoid of meaning and satisfaction. he then spends his years in the wilderness crashing out before stumbling back into conformity because he knows nothing else
you can find meaning outside of work in so many ways. joining a small organizing group, help run mutual aid efforts, even dipping a toe into civic engagement. disentangle achieving goals that signify success from living your values and you'll be on your way to a better regulated life [1]
> The purpose of life is to make meaningful human connections, become wiser, become a better person.
These are nice things. But let's be honest, there is no objective purpose in life beyond the biological imperative. Purpose is subjective, and people will find that on their own. Or not.
There is no singular “purpose of life”, different people are built differently and experience the world in their own way.
Some are lucky enough to get meaning from the default path of those around them, some have trouble finding a way that works for them.
Every person should consider what their own purpose is (and why, and what “purpose” even means for them) or they will be held in thrall to the social forces around them. If they don’t fundamentally align with those forces, they will have a bad time, and they may blame themselves rather than recognizing that they should seek a different path.
"Create shareholder value at my current startup" obviously isn't going to scratch that itch in the same way.
I would add a little nuance. One thing is your job isn't advancing Humanity Progress in a very direct and unambiguous way, say you work for some Very Important Organization that Does Good All the Time. Another thing is your job is totally useless or even harmful to society.
Maybe most of us are in between?
I've worked at dysfunctional companies that made me personally unhappy (busyworks, company politics, waste, bad bosses, bad teammates...) that somehow made some good things in the grand scheme of things, other that made me very happy with more neutral results.
Someone has to do mundane things, even to raise children and give a chance to the next generation to improve the world.
I don't believe putting too much thought in purpose is healthy.
Oftentimes it's undiagnosed health issues, such as vitamin deficiencies and/or malabsorption conditions such as gluten intolerance preventing proper vitamin levels, that can present much like "inexplicable" burnout, imho.
Do you think it could be a combination of things? Like, in order to avoid burnout you need some degree of meaning but also some degree of success and positive work environment.
I'm coming at this from the perspective that I've been struggling with a lot of burn out and I'm mostly proud of who I am and what I do but I'm stressed because I constantly have to compensate for society's failure to reward people's contributions.
That could just be me though. I definitely agree that I think that meaning and purpose is a big thing that people are missing now.
I don't think you're unusual at all. The modern world is almost designed to make you crazy in some ways and makes it harder than it ideally would be to achieve these ideals (which is what they are, we're all human and none of us can achieve this all the time).
Unfortunately largely it is what it is and can't be changed, the only thing we can change is how we choose to respond to it
> not really an issue with the specific work environment.
I’ve contributed some time in a mentoring group. Many of the cases people would call “burnout” had little to do with their work. Once they started talking details about their life situation it would become obvious that their problems ran deeper, but blaming work was an easy way to externalize it and place the blame on easy villains: Corporations, capitalism, managers, coworkers.
There were a lot of situations like this where someone would get a new job, take fewer hours, or even some people requested demotions and took a pay cut. Then when nothing changed about their well-being, they’d shift to blaming something new at work. I talked to people who could seemingly find ways to “burnout” in every situation they’ve encountered from high school through job hopping. It’s really hard to get someone to accept that maybe the problem is running deeper than their environment, and maybe something they’re doing (unintentionally, I’m not trying to place blame) is actually leading to this outcome over and over again.
I agree that work coaching is not the solution. It presumes the diagnosis and anchors blame to work, which starts obscuring possibilities outside of that pre-chosen explanation. Regular old therapy from a trained therapist is probably a better place to start. Might be cheaper, too, especially if insurance is involved.
Hate to sound like a Marxist but... this is what happens when capital has this much power concentrated in so few people. Eventually the system eats itself.
If you're in the right elite at the right time though, you can make a lot of money while everything falls apart underneath
Miraculously everyone else suddenly learns how to drive and doesn't try anything stupid
reply