We’re building data centers in the U.S. at an unprecedented and increasing rate [1]. Is it so hard to imagine that something which was fine at one scale may begin to cause issues at some massively higher scale?
If I drink one glass of wine per day, it’s probably fine. If I drink three bottles per day, I’m gonna start having problems.
Yeah, but when we account for that scale, and look at resource usage, datacenters aren't topping the charts, or even really getting in the top 5, for anything.
You have people who eat meat at every meal losing their mind over water usage and people who set their AC to 68 panicking over energy usage.
Eating meat fills my belly. Setting my AC to 68 keeps me cool. A data center costs my community money, damages the environment directly and indirectly, and further enriches a few already rich people.
The estimates I've seen place the water consumption of water at such a low level that a whole month subscription consumes roughly as much water as 1-120 grams* of beef.
> Setting my AC to 68 keeps me cool.
What, is 70 not good enough?
Air con, and basically all things connected to the power grid, also "costs [your] community money, damages the environment directly and indirectly, and further enriches a few already rich people."
* yes, one to one hundred and twenty, there's that much uncertainty in the estimates for the AI.
When Amazon gets to run an offgrid power plant for a new data centre that's permitted to release more co2 than any other single power plant in the US[1], you'll pardon me for wondering what the rush is and why it becomes literally everyone else's problem as a result.
Build the data centre, sure, but if it takes you five years to build out the grid and sufficient generating capacity in a non stupid way in order to do that, sorry, but that's by far my preferred route to get there.
Rush may be FOMO. Even as one who likes using AI, I think the economics is an investment bubble and mostly avoid it.
I can't read your link, paywall, but my question regarding Musk's infamous polluting DC, I wonder why it had to be built near so many people. It's not like there's going to be a need for a huge pool of workers to commute to it.
> I wonder why it had to be built near so many people
I can't read the link either, but at a glance, the first one wasn't. It's in clearly industrial land a long way from any houses, with its closest neighbors as far as I can tell being a steel mill, a few existing power plants, and a wastewater plant.
The second one is practically right in the middle of a bunch of neighborhoods, and for that, yeah, I want to know as well. I'm guessing the land was cheap, industrial-zoned, and near a substation?
(Not to excuse what Musk is doing in either case - it's fucked and has wide ranging effects. But I think it's important to clarify the first because I've seen a repeated trend of people acting like a new DC is going to be built in the middle of a peaceful picket-fence neighborhood, and then you look it up, and it's actually being built on top of a former steel mill a mile away from the nearest house. It gets annoying.)
Because they aren't really people unless they could afford a DC of their own. When you view half the world or more as NPC then it becomes trivially easy to ignore the needs of others.
I want to understand how his mind works. Not sure what value this is, now I think about it: I'm unlikely to time a bet against him well enough to win it.
So, for this point, I don't care that the words and actions he emits have no apparent rational long-term coherency, but I do care about the coherency of the system behind them which generates them.
Think of it like me wanting to learn how an LLM works so I don't get surprised by it doing more than a Markov chain.
It can be done even in rich countries, but the point is I think the opposite of your implication: I am claiming that it is hard to avoid causing costs to one's community (and that almost everything causes more impact than one might expect), I am not claiming to have achieved such avoidance.
JTFC why are you arguing minutiae? The analogies aren't the argument.
AC and meat are bad for the environment but they deliver personal benefit. Having a DC in your community has no appreciable personal benefit, only negatives. That's it.
> JTFC why are you arguing minutiae? The analogies aren't the argument.
I'm replying to someone who themselves argued the analogies.
> Having a DC in your community has no appreciable personal benefit, only negatives.
To the extent that statement is true, it applies equally to most primary and secondary economic activity.
Having a farm in your community has no appreciable personal benefit: for scale, the UK famously gets a lot of their sheep meat from New Zealand, as far away as it is physically possible to be.
An abattoir similarly has no substantial local benefit, but worse has much detriment owing to the psychological damage it does to the workers, leading to higher rates of arrest for violent criminal offences: https://journals.sagepub.com/doi/abs/10.1177/108602660933816...
Unless it's at the scale where it becomes a company town, having a factory, a warehouse, etc. these things are all heavily automated or mechanised and employ only a few of the locals as a percentage of the population. Putting anything that isn't service-sector in your community (i.e. schools, hospitals, shops, restaurants, hairdressers, etc.) is going to be of "no appreciable personal benefit" when considered in isolation.
Almost all primary (farming, fishing, mining) and secondary (manufacturing) comes with negatives (pollution, higher wear on the roads and utilities).
Does that water usage take into account all the water that went into training the models, including the electricity it took run the data centers the GPUs are housed in?
> These bad faith "but you exist and do unrelated things" fallacies are a.) bad faith b.) make people have ai people more.
> Bringing uo AC while you defend datacenters making global warming worst is just a cherry on top.of it. Just underlines anti-human fck you all ideology of it all.
You are badly breaking the HN guidelines in a dozen different ways. You should refamiliarize yourself with them:
> Bringing uo AC while you defend datacenters making global warming worst is just a cherry on top.of it. Just underlines anti-human fck you all ideology of it all.
This one is factual on both counts. All you need is to read what CEOs we talk about write and listen to what they say.
> These bad faith "but you exist and do unrelated things" fallacies are a.) bad faith b.) make people have ai people more.
This one was entirely correct assessment of the situation and proper response. It was a fallacy. It used the "but you exist and do unrelated things" argument. And yes, it makes people hate ai culture more.
> This one is factual on both counts. All you need is to read what CEOs we talk about write and listen to what they say.
What they write and say is all over the place. Some of it (IMO, everything in the last few years by Musk and Zuckerberg) is obviously nonsense.
Altman, much as people like to hate him, is presenting himself as wanting governments to figure out the rules by which this should all be restrained. Amodei put his money where his mouth is on attempted morality, and got burned by the Trump administration for his efforts.
> This one was entirely correct assessment of the situation and proper response. It was a fallacy. It used the "but you exist and do unrelated things" argument. And yes, it makes people hate ai culture more.
It's fine to care about the stated negative consequence. I sure do.
You a software developer? If so, where do you put your efforts on optimising code, the bit with 5% of the problem, or the bit with 95%?
Likewise: If you care at all about the stated negative consequence, you do not stick your fingers in your ears and moan about the people telling you the origin of 95% of those negative consequence, just because everyone's favourite target of hate this year is 5% of the total problem.
Meat in particular is way, *way* worse than most people think it is. I presume this is due a mixture of "but I like the taste" and PETA having about the same level of political nous as Liz Truss.
For most of the environmental issues under discussion, the total impact of meat production by itself has a *much* greater impact on both local and global environment than even what happens if all the currently planned data centres get built, though of course "how much" varies by both specific location and specific measure. Greenhouse gasses? Meat is 20x worse; Electricity? Meat's worse by a factor of 2; Fresh water? Meat's worse by 4000x in aggregate, with specific local stresses going both ways, wars have even started over fresh water supplies for agriculture; Biodiversity, eutrophication, deforestation, for these things DCs are obviously negligible, not so for meat; Mining/material extraction is hard to compare, given how much mineral fertiliser we go through, but the extraction of mined fertiliser has basically killed off some ecosystems; Local air pollution? Specifics matter, though planet-wide meat's still worse.
And then there's all the other things people do, and demand the right to do.
We (humanity worldwide) need to reduce greenhouse gas emissions by 99.9%.
Of the two, use of air con in the US is a bigger source than current data centres, and in the middle of the range of extimates for how many new data centres will actually get built over the next few years.
It would of course be much better if both AC and DC (in both senses) were powered by renewables, but y'all elected someone who thinks they're a woke Chinese scam or something, possibly because one got built near a golf course in one of his foreign properties, possibly as a favour to the people who gave him that new aircraft for his "library".
It is based on the mean lifetime of a CO2 pulse being order-of a thousand years, which is itself approximation of what IPCC says; actual decay rate behavious is not even as simple as exponential decay.
At 1000 years exactly, steady-state requires 99.9% reduction in the long term.
Good news is that in the short term, the higher partial pressure in the air means it leaves faster at first, so things stop getting worse much sooner than that.
> It is based on the mean lifetime of a CO2 pulse being order-of a thousand years, which is itself approximation of what IPCC says; actual decay rate behavious is not even as simple as exponential decay.
> At 1000 years exactly, steady-state requires 99.9% reduction in the long term.
So, that's ignoring carbon capture by means natural or artificial, and therefore is a totally irrelevant number, both to this conversation, and to most others.
> So, that's ignoring carbon capture by means natural or artificial, and therefore is a totally irrelevant number, both to this conversation, and to most others.
No, it inherently incorporates all natural processes that capture carbon.
Artificial, sure, if you want to point at them I believe current effots might reduce the required reductions from 99.9% to 99.8%. However, I wrote "emissions", and the overwhelming majority of CCUS systems are reditecting what would otherwise be emissions directly to capture, i.e. what I wrote would include that as reducing "emissions".
> No, it inherently incorporates all natural processes that capture carbon.
It literally cannot, because it's not parameterized on the amount of additional plants being planted.
> Artificial, sure, if you want to point at them I believe current effots might reduce the required reductions from 99.9% to 99.8%.
Goalpost moving (i.e. fixing a particular value in hypothetical future assumptions that was not fixed or agreed-upon previously in order to support an otherwise-false point), which solidly sets the original claim as false.
> However, I wrote "emissions", and the overwhelming majority of CCUS systems are reditecting what would otherwise be emissions directly to capture, i.e. what I wrote would include that as reducing "emissions".
Technically compatible with your statement but not how the vast majority of individuals would interpret it as you originally wrote it.
To be clear: I'm not arguing that we don't need to substantially reduce greenhouse gas emissions or that carbon capture is necessary. I'm arguing that last point, that "we need to reduce emissions by 99.9%" is extremely misleading and much more harmful than hurtful for those ends.
> It literally cannot, because it's not parameterized on the amount of additional plants being planted.
Primary production of the planet, i.e. total carbon fixation, is broadly not influenced in this manner.
The things which do influence it, are included in the function.
> Goalpost moving (i.e. fixing a particular value in hypothetical future assumptions that was not fixed or agreed-upon previously in order to support an otherwise-false point), which solidly sets the original claim as false.
My point is that it is tiny in scale even at best.
Furthermore, I claim it is unlikely to get meaningfully better than it already is, given the economics at present appear to be more "how do we game government objectives to get subsidies" rather than directly performing useful work that does not require subsidy. Perhaps I am too cynical, but typically if there's a good economic reason for doing something anyway, it's not in a list of big showy projects that might happen at some unspecified time in the future, it gets implemented by people who are happy to never get noticed.
> Technically compatible with your statement but not how the vast majority of individuals would interpret it as you originally wrote it.
Can you prove that?
> To be clear: I'm not arguing that we don't need to substantially reduce greenhouse gas emissions or that carbon capture is necessary. I'm arguing that last point, that "we need to reduce emissions by 99.9%" is extremely misleading and much more harmful than hurtful for those ends.
I do not accept your argument.
Especially "much more harmful than [presumably auto-corrupt and you meant helpful]".
*Net* emissions must go down approximately that scale. The common phrasing is "net zero" rather than "net 0.1%". However, your assertion that me missing the word "net" from "We (humanity worldwide) need to reduce greenhouse gas emissions by 99.9%" seems pedantic rather than helpful.
> Primary production of the planet, i.e. total carbon fixation, is broadly not influenced in this manner.
Plants account for roughly half of global carbon fixation. "Broadly not influenced" is straight-up false. Your follow-on arguments are consequently false as a result.
> My point is that it is tiny in scale even at best.
You are posting on a website that is hosted in a data center right now. You use data centers at your job to send email, share documents, run software. When you go home and watch netflix, that video is streamed to you from a data center. All of the internet, your phone, and modern digital entertainment and life is run through data centers. My guess is that you don't seriously want to turn all of that off because 'reasons'.
Not at all. The post being responded to completely failed to acknowledge any value brought by data centers to the general populous. That's disingenuous. To force the point: run me through the scenario where we turn off all the data centers. Consider how much more water we will have, how much less expensive electricity will be, and what other problems might arise.
But that value is subjective. If my tap water is brown as a result of a DC, I may not agree with your positives. If my neighborhood is humming all night because of a DC, again, I might not have the ability to see what you see as positive.
So, my take is that maybe the pro-DC side can do better at seeing multiple perspectives instead of acting like anything negatively said about DCs is a psychosis.
>If my tap water is brown as a result of a DC, I may not agree with your positives. If my neighborhood is humming all night because of a DC
Those things are both a result of a failure of government, not the existence of data centres. If there are externalised costs the company wants to put on people then your government should extract them from the company as a condition of permission.
Data centres are factories that turn electricity into heat and money. They're basically the nicest factories possible. There is no fundamental reason they need to harm people, any issues data centres cause should be easy to address as a fraction of the already massive sums involved.
So if I build a heavily protested foundry that spews carcinogenic lead smoke into a residential neighborhood, it’s not my fault and the operation of my foundry is unimpeachable. It’s the government’s fault for not stopping me from doing it. Since the people affected are responsible for compelling government to limit business, businesses are not responsible for anything. Privatize the gains, then socialize the losses, then blame the government for not stepping in, and ignore the company actively undermining the public’s leverage in the process. Ok, got it. That makes complete sense. It’s decidedly not nihilistic justification for doing obviously unethical things in pursuit of profit.
The third paragraph of my comment already addressed this. A data centre is not a foundry.
In a metal foundry some of the physical externality is intrinsic to doing the process. It's a continuous industrial chemical operation emitting smoke, metal fumes, VOCs, dust, sound etc. You can capture 99% of pollution, but you're still melting, casting, shaking out and finishing tonnes of hot metal and sand. Complete physical isolation is expensive and imperfect.
With data centres, most local externalities are engineering choices. With enough money and sensible siting, you could plausibly build one whose effect on neighbouring residents is extremely close to nil. The biggest remaining externalities occur upstream in the electricity/water/grid system - and those are unusually amenable to being charged directly to the operator.
If a local data centre is impacting you negatively it's a failure of your government because it didn't need to be that way. It's not an intrinsic part of data centre operation. They turn electricity into heat and money, a comparatively simple factory to have completely deal with its own externalities.
People aren’t complaining about the idea of data centers, abstractly. They’re complaining about the actual way they’re being built. To ignore the reality of what living next to GPU data centers is like, and the amount of resources they actually demonstrably use, and instead paint this as a theoretical argument is completely disingenuous.
One tylenol is good at reducing my fever. 20 tylenol will likely damage my liver and might even kill me.
I'm not saying tylenol is bad. I'm not saying it's good. I'm just pointing out that your position is equivalent to opposing a person saying "lets no shove 20 tylenol down someone's throat" with the argument "but one has reduced the fever".
Unlikely. I'm not sure where HN is hosted, but I think it's a bare metal server in a colo, not a data center.
The childish behavior is disagreeing with how datacenters are being deployed and reducing that to "more is bad", and then stretching the position to "data centers bad". I'm begging the collective here to consider what they are really asking for, because our society, from the government down, is built on the cloud. Finding a healthy way to provide compute is important, and "data center bad" doesn't get us there. We need the solar/wind equivalent for hosting. The problem is data centers have efficiencies at scale that we can't achieve with more distributed solutions, but I'm not a specialist in that particular analysis.
How much if that is true and how much of it is carefully constructed propaganda? Implanted and reinforced beliefs to make workers hustle to ensure the server fields produce enough crop this quarter.
Most people on the planet are not tech people. Their world is not built on cloud at all but ensuring there is sufficient food and stuff to satisfy their biological needs.
It's asocial terminally online tech workers who think the cloud is akin to strong and weak nuclear forces and gravity; so essential they are fundamental forces of nature themselves
Hopefully you can get the obvious bias of your perspective; species of 8 billion is not propped up by a few million script kiddies
I think you're vastly under-estimating the extent to which power grids, water systems, and global supply chains and manufacturing rely heavily on datacenters.
The Trump administration pulled a bunch of ocean sensors, whose data was aggregated in the cloud, in an attempt to save money. Except, those sensors were not only providing research data on changes in ocean currents, but helping fisheries make predictions about yield. The administration backtracked because of this.
There is a similar story with weather forecasting, and farmers being able to manage their crops.
Let's not even get started on the economic impact worldwide of datacenters were not available to process stock and bond trades.
New technologies seem optional, but once society passes some tipping point in their adoption, they no longer are. AI isn't there yet, but cloud computing is.
I think I've given you far more than you're due in this conversation, so I'll stop here.
In short, I think you're being dismissive, particularly when you try to frame my position as "society is propped up by script kiddies". Seems like a deeply inaccurate way of summarizing my points.
> I think you're vastly under-estimating the extent to which power grids, water systems, and global supply chains and manufacturing rely heavily on datacenters.
These things were with us and worked better before datacenters even existed. Datacenters are a mere convenience which is now used to force monopolization of compute while subjecting society to extremely risky, but unnecessary dependencies and centralization.
You say we're already there? A good reason to start undoing it ASAP.
That's not the proposal, and is utter fantasy. I'm engaging a hypothetical here -- none of what's being discussed is even close to reality, in the sense that it won't be implemented. There will be new datacenters, some communities will successfully resist, but many won't.
What you propose is similar to proposing that we'll solve climate change by moving back to horse-and-buggy. Theoretically possible, practically impossible.
> The childish behavior is disagreeing with how datacenters are being deployed and reducing that to "more is bad"
Why is that childish? Bubble data-center economics is bad, we have the history to prove it, this isn't the first bubble. Do you have real arguments against it besides calling it "childish" which is itself childish?
> I'm begging the collective here to consider what they are really asking for, because our society, from the government down, is built on the cloud.
You're begging "the collective" to stop worrying and learn to love the bubble... or else? OK, or else what? If everything is already "built on the cloud" there's already enough cloud to go around, built at a normal pace, without bubbles and without incurring huge debts, deficits, circular financing and what not. See, the how can be done better.
> Finding a healthy way to provide compute is important, and "data center bad" doesn't get us there.
That's a strawman, which you sneaked in here: 'reducing that to "more is bad" and then ... to "data centers bad"'
The issue is, and always has been - the bubble way of doing it as bad - slow down and do it properly. Either way the public isn't going to get the benefits of the build-out, at least try to reduce the negative externalities to an acceptable level.
But do you really think we have more datacenters now, or even an equal number, to what we'll have in the future?
It's like power consumption: it's going to scale, so we need to control how it scales. I don't know where the conversation goes when folks just talk about how useless it is and how compute is bad.
Take two steps back: do you really think we're at peak compute? If not, shouldn't we focus on how to scale in a sustainable way? The rhetoric doesn't foster that inquiry. That's my issue here.
Every single critique has said "slow down to do it right". That's similar to the DEI fallback of "Do the work". It's not that the goal is wrong, it's that it takes no effort to demand something vague like that, because it requires no thought. What forwards the discussion when implementation matters is the specifics: what do we want folks to do differently. If gas turbines in data centers are causing noise pollution that affects quality-of-life (which is absolutely appears they do - it's really bad), that's not a call to ban the idea, but rather the implementation. I see that distinction too often being dismissed in these discussions.
> It's like power consumption: it's going to scale, so we need to control how it scales.
I don't think your premise is obvious at all. Developed countries have stagnant or shrinking populations, and there is a general trend toward greater power efficiency over time. So why should total power consumption be expected to keep increasing?
Extremely resource-hungry GPU data centers have a much deeper impact on the surrounding area. The only data centers power hungry enough to install local turbines for power before the LLM craze were GPU-based crypto miners, and living next to one of those also sucks pretty hard. One person I saw interviewed described the sound of the gas turbines like having a clogged shop vac in your house 24/7. If you look into it even briefly, you’ll see that you can’t honestly equate their impact to standard CPU server farms.
Racked servers pre-GPU used about 1kW each at high load. 5-10kW for a rack (20 to 40 servers) was common, much more than this would be a special thing.
AI use has driven this to 20, 50, 100kW per rack or more. Supposedly 1000kW for the hyperscalers.
Extrapolate to the whole building and you have a lot of waste heat that has to go somewhere.
Instead of being similar to light industry, the hyperscale AI datacentre is using as much energy as a large steelworks.
What a weird line to draw. Data centres do a lot more than what you report. E.g. they might host your electronic health records. They might host census data, or weather data, or your kid's school learning management system.
In my experience, there's not a ton of overlap between people who say this and people who use AI for everything they do... but most people who say this do use social media, cloud storage, on-demand video streaming, and plenty of other things that rely on data centers :)
Not for nothing that YouTube alone is responsible for a double-digit amount of all internet traffic.
Consider that there may be large populations doing both, but the overlap might be less than the internet jockey concludes via finger-in-the-wind hypocrisy check.
The interesting thing is: they don't. Most people not on HN could give two shits about an AI generated overview on Google search, in fact if you asked most people of age they'd rather have a Google Search that worked vs what we have now.
Just because the bubble here uses AI doesn't mean it's even remotely valuable for the 80%. But they've done a great job convincing a specific group of people this.
No one was objecting when pornhub was building out data centers. Anything they like to consume, no matter how frivolous, is accepted as morally better on a metaphysical level.
> further enriches a few already rich people.
Meat eaters damaging the Earth is okay because it's "equal"? 300 million people deciding they're entitled to consume anything they want in ever-increasing quantities is a problem because... there's 300 million of them!
Listen, I am no meat advocate. I just see a very clear link between the resources required to make meat and the benefit to the person eating the meat. Where data centers cause local and global harm, while not benefitting the locale or the globe. Which is why people don’t like them.
And it’s not like non-meat is fine - the corn industry uses a blinding amount of water, as do almonds, etc.
We have a finite budget for water usage, and i’d allocate that to non-meat food > virtually anything else > meat > data centers > golf courses.
> No one was objecting when pornhub was building out data centers
I’m not entirely sure of the scale of pornhub’s operations, but i feel like one large ai data center uses more resources than all of pornhub’s data center usage around the globe. But maybe i’m wrong there, and the hub was just flying under the radar.
> And it’s not like non-meat is fine - the corn industry uses a blinding amount of water, as do almonds, etc.
Switching from meat to non-meat unequivocally lowers your GHG and water footprint and there's really no room for quibbling on this unless you are switching to a diet of exclusively almonds
What is? The opposition is to datacenters being built to meet imagined future demand for LLMs. Are LLMs directly and indirectly increasing my quality of life? Please explain how, because so far it seems to me like they’re decreasing my quality of life.
Data centers generate export revenue. For a country that has struggled to export as much as it imports, that seems very valuable to its national economic health, and sabotaging one of the few major export industries, and one that's growing rapidly at that, seems counterproductive.
Data centers also generate local tax revenue. In fact, across industries, they have the best tax revenue to tax resource burden ratio, using very little government services while paying significant property tax.
And the environmental damage, whatever that is, is negligible relative to other industries. Their only output is some noise pollution nearby and whatever emissions are produced in the power plants that give them electricity. All industries require electricity in proportion to how much economic value they produce. So unless you want degrowth and no industry, you're going to have energy consumption.
In fact, rising energy consumption is a sign of a growing economy. That's why China has seen its energy consumption grow far faster than any other major economy while its economy has also grown far faster.
They do use water but their water usage is mininscule compared to agribusiness, or even fast food restaurants and golf courses. And the potable water usage can be brought down. They just use the water for cooling. So they could use treated wastewater or recycle the water they use. Newer data centers use these more water efficient cooling systems. So water usage is not a major long-term concern. Not one that justifies crippling this key emerging industry.
maybe i'm working on software for supply chain routing. maybe i'm making a commodity trading bot that more accurately prices wheat futures. AI helps with both of those things.
> maybe i'm working on software for supply chain routing.
We have supply chain routing software already. Humans wrote it, humans can improve it. Getting ai in the loop won’t change the realities of the supply chain, except maybe they’ll hire less programmers.
> maybe i'm making a commodity trading bot that more accurately prices wheat futures
How does that benefit me? Further financializing food production seems like a bad thing.
Not to mention all of what you’ve described could be run on a $1000 consumer GPU in a PC somewhere.
This is what I mean about not understanding the second order transitive labor that goes into fulfilling people's wants.
Making supply chain routing software cheaper, improving it for less money → all of these things directly make the products in grocery stores cheaper.
"Further financializing food production seems like a bad thing." →
Why assume rather than approaching with curiosity? Commodity futures are, again, critical to global supply chains and food markets. Farmers in eastern Europe actually look at futures markets to determine when and if to plant their harvest, cargo ships that transport your food buy futures in oil & gas so that they can have stable fuel costs to provide you with your food, etc. etc. Speculating on the future gives rise to the corrections necessary for providing things that people want in the future.
A data center allows the website you're whining on to function. If they're really as inconsequential to your life as you're implying, maybe you should boycott them by not using any websites or apps that rely on a data center to function.
AC at 68F is obnoxious and wasteful. At least data centres enable me to communicate with distant colleagues, watch enriching movies, and use very strong AI tools to accelerate my research.
It's actually significantly more environmentally friendly than heating to above 68F in the winter in most cases, since most AC is done with heat pumps but most heating is oil/gas (100% fossil fuel consumption) or resistive electric heaters (dramatically less efficient than heat pumps). On top of that AC use aligns strongly with solar generation output while heating load does the opposite, which means that even when heating is done with electricity it comes disproportionately from gas turbines at natural gas peaker plants.
Data centers also enable the NSA to store the metadata and contents of every phone call you make and police to (without a warrant) to track anyone who's walked in front of your doorbell or many street corners in your city. They also store all the location data your phone collects for Google or Apple, nearly your entire internet history collected surreptitiously because many sites automatically use your GMail or Facebook logins, and your entire purchase history collected via Visa or MasterCard.
And even better: all of that data is aggregated together and stored in its entirety several times over across multiple data centers because everyone shares your data with everyone else, such that if any single of more than a dozen data centers is ever compromised, it all gets leaked to whomever compromised it.
My building is leaky. The association won't fix it. How about this compromise, you set yours to 74 to compensate for my 68. Fair enough, or will you say I should move?
and if we’re gonna start worrying about how much a building raises the nearby temperatures, let’s also start looking at whether or not buildings are using dark or light colored roofing
People have their priorities, and putting the data centers into their mix isn't what they asked for. To say "resource abusers think this is a resource abuse" isn't really a counter argument.
When you account for post build runtime use it never looks like a big deal
Account for the runtime of the extra mining, production of materials... the scale blows up exponentially when externalities of all those extra screws and zip ties are appropriately accounted. An issue with Americans in particular as capitalism propaganda completely ignores externalities like kids in textile sweatshops not given a choice of college or the mines
3c global increase by 2050 is locked in. Meaning another 10-15 degrees local warming.
If Exxon could predict the min-max of the exact trend in the 1970s.
Physical reality doesn't give a shit about our rhetorical ability to blame a corporation or rich person; everyone consuming this shit is part of the math of the problem.
To be completely honest, normal people being able to set their AC to a sane temp does sound very valuable, and probably totally worth slowing down the slop machines a little bit.
what if rather than enumerating the list of allowed/disallowed economic activities based on some moral criterion, we just tax the rich and redistribute it. i kinda thought this is what we spent the entirety of the 20th c. litigating but maybe i'm mistaken
It's incredibly disingenuous to wonder why people would be concerned about projects that:
* Take a bunch of their tax money with no benefit to them
* lower their property value
* Make electricity and water more expensive (both of which make other things they use more expensive also)
* can cause other environmental impacts that raise inconvenience and expenses both directly and indirectly
so some AI douchebags can have a shiny datacenter to enable their bosses' stated desire to replace them with the very same AI might be opposed to the current AI drive datacenter boom.
Whatever actual benefits AI has, or tech in general, the way it is being approached right now is a disgusting and blatant attempt to steal as much as possible from everyone. People don't like that, and the like even less how the people doing it pretend they don't understand the dislike while building bunkers to protect themselves from the anger.
They aren't tax positive. The deal might "theoretically" be tax positive in 30 years if things go exactly according to some plan. In the short term they are usually associated with large tax breaks for some number of years, accompanied a guarantee to occupy the data-center for less than the payback period (so the tax breaks can be extended later with threats to move on so those tax paybacks never happen). They also come with the locality footing the bill for a lot of infrastructure build-out.
That's just a bad use of my tax dollars. Let the datacenter people pay for themselves at the very least - my taxes don't need to go subsidize some pile of shit AI fanatic. If you like AI, pay for your AI, don't take it from me. It has nothing to do with me being pro- or anti-industry, however if an already profitable industry can't exist without my tax subsidizing it just so some rich guy can be a little bit richer... well maybe it doesn't need to exist.
I have no idea how any of this is conservatism... I'm opposed to a specific set of changes, not all changes. Perhaps you are just trying to throw shade because I think you should pay for your own toys?
Oh, it's absolutely easy to imagine. That is what should give us pause. We are fantastic at imagining things. We are less fantastic at being reasonable.
It's not so hard to imagine that something which was fine at the old scale may cause issues at the new. At the same time, it's also not so hard to imagine that the new scale still isn't particularly large and that there are about a million more important things to worry about.
The question isn't whether it's possible for something to become problematic past a certain scale. The question is whether data centers are at that point, or will be anytime soon.
The actual facts and figures that are presented in opposition tend to be stripped of the larger context, which makes me skeptical. The graph you linked looks wild, but in a country that spends two trillion dollars per year on all construction, is the jump from $1 billion/month to $4 billion/month very important?
I feel like this has to be reiterated every single AI discussion. Anytime someone points out something it enables, someone has to go “well that existed before” as if the scale is somehow 1) irrelevant and 2) not orders of magnitude larger.
If you gave me one night, a pot of coffee, and few scruples, I could probably dump hundreds if not thousands of fake posts on the internet. Hundreds would easily break through and who knows, a handful might even go viral on Twitter. Yeah I could do that 5 years ago, but not nearly at the same scale. Maybe a few dozen?
Back to DCs: it’s like saying “well we had coal fire plants years ago.” Ok…well there are more now generating way more power and spewing way more pollution. It’s not the same situation anymore. This DC build out is massive.
I think both sides of the debate are guilty of being too vague about it, probably because being more precise is pretty difficult. "Data centers already existed." "The scale is vastly larger than before." OK, this exchange contains zero new information. "Already existed" doesn't mean it's fine now, and "vastly larger" doesn't mean it's bad now.
This DC buildout is massive? Yep. So massive that it merits substantial public opposition? Citation needed on that part. And any citation that compares current or projected DC activity with past DC activity is missing the point.
Two concrete examples: the above graph showing a multi-fold increase in construction expenditures is useless, because it gives no indication of the absolute significance of the current level. It tells me very little. For a better example, here's a graph of RAM prices over the past year and a half: https://whereismyram.com/us/price-chart. Here, we can see that data center buildout is big enough to massively affect the global market for an important commodity, and we can translate that into some sort of understanding of how this affects ordinary people.
I’m not saying anything about opposition. I’m just saying I’m tired of “well people made unwanted porn before” or whatever as if the tools haven’t radially changed.
Your comment sure makes it sound like you think there's a problem with the new scale of things. And I'm not saying that's incorrect, just that it needs to be backed up with something more than "it's way bigger than it was."
Completely agreed that the "this existed before" argument is tiring. But so is the "it's a lot bigger now" argument. Neither one actually says anything interesting. They're both statements of opinion (this is fine, versus this is bad) hiding inside obvious statements of fact.
I just want something concrete to judge. The data center debate rages but almost never in terms that help me figure out what to think about it. RAM prices going through the roof is something I can easily contextualize and shows the impact on the broader world. Data center construction spending going up a few hundred percent isn't and doesn't, and when I look up the numbers to contextualize it then it doesn't really look all that significant.
“It’s a lot bigger now” just means we have to deal with realities of scale, that’s a true statement so I’m not sure why that’s different from my scale argument. The whole point is “yes these 2 things are the same, except scale, which can make it very different.” 1 bee sting is one thing, 1000 is another. 1 piece of candy is one thing, 1000 is another.
You are assigning additional value to my comments - I get it but I’m trying to say clearly that I’m not saying those things. All I am talking about is acting as if scale is not a part of the conversation. Anything beyond that I have yet to weigh in on. i’m happy to elaborate if you want me to on any particular point, I am not being deliberately opaque here. I’m just not entirely sure what you are looking for.
My main point is that this goes both ways: there are those who act as if scale is not a part of the conversation, and those who act as if scale is the only part of the conversation, and neither one is sensible.
It seems to me that we have far more people who think it is not a part of the conversation than people who think it is the only part of the conversation that matters. The latter is far more simplistic and unreasonable.
Honestly we’re getting off track IMO. I think we’re just running in circles here. Appreciate the conversation nonetheless, have a good rest of your week
Complete tangent, but my impression is that one glass of wine per day is "fine", but you're still statistically better off in a non-trivial way by going down to zero. Not counting the benefits of course.... :)
To continue to take us off topic, there is actually pretty limited evidence that this is true.
It is true that wine doesn't do anyone any good (unless you count it making people more social or something). But the effects of one glass a night is indistinguishable from zero. There are definitely statistically significant morbidity effects at 3.
I've been shocked reading public health proclamations that "anything above zero" is bad, and then reading the actual underlying paper they refer to, which clearly states exactly the opposite finding. I think it is an extension of the public policy belief that the public cannot be trusted with moderation, and so therefore must be communicated to in absolutes, and nobody will read the underlying paper anyway.
Most modern research I have seen says there is no safe levels of alcohol where you will get no negative effects on your body. 7 glasses of wine per week isn’t that low. Is it worse than sometimes eating crap food? Maybe not. But it isn’t without risks.
i don't agree, i think you are just describing a measurement problem, we have a clear mechanistic understanding of why 'anything above zero' would be bad, and are able to measure this in sufficiently large populations for larger drinkers. the effect size is smaller and thus unmeasurable above the noise floor for one glass a night, but the mechanistic story is clear.
Simple mechanistic explanations in biology are often wrong. For example, oxidative stress does not have a linear no-threshold harm.
For alcohol, almost every big study has shown either no discernible effects going from zero to 1 drink per day, or actual slight benefits going from 0 to 1 that then reverse when increasing beyond that. There is not strong consensus which of these is true, but there's definitely no real evidence that 1 drink per day is harmful as measured by all-cause mortality.
See GBD 2020 in the Lancet (which showed a slight J-curve), Wood in the Lancet 2018 (optimally low risk at 100g ethanol per week, or ~1 drink per day), Zhao in Jama (no discernible negative effects at 1 drink per day, no benefits either, no J-curve), and the latest National Acadamies review showing lowest all-cause mortality at about 1 drink per day (so, a J-curve).
Alcoholism has tons of negative effects. But these seem to start at around 2 drinks per day and rise rapidly from there.
isn't one glass a night already lack of moderation? by drinking a glass a night, or even weekly such as every friday night you're creating habit and dependency.
It works very differently for different people, I think. Many people metabolize the alcohol more slowly, and get less of a "hit" from drinking, which makes them less susceptible to addiction. Others, like me, metabolize it very fast and that makes it quite a bit more addictive. Risk factors like ADHD increase this.
All that said, no one I've ever met said a glass of wine on Friday nights isn't "drinking in moderation". I think the medical establishment is quite well aligned on this. One source is the CDC, which defines moderate drinking to be roughly an order of magnitude more alcohol consumption than "one drink every Friday night."
> Many people metabolize the alcohol more slowly, and get less of a "hit" from drinking, which makes them less susceptible to addiction.
This is the opposite of how it works, ethanol is the active ingredient for all of the fun parts of drinking, the metabolite, acetaldehyde, is the cause of all of the unfun parts (nausea, headache, flushing).
A significant percentage of East Asians have one or both of a superactive alcohol dehydrogenase (which converts ethanol to acetaldehyde) and a defective aldehyde dehydrogenase (which converts acetaldehyde to acetic acid), and they have significantly lower rates of alcoholism.
My doctor indicated it was more of a risk for me because my liver worked quickly, but your mention of acetyldehyde I know to be correct. I need to better understand this, thanks for the correction.
> Many people metabolize the alcohol more slowly, and get less of a "hit" from drinking, which makes them less susceptible to addiction. Others, like me, metabolize it very fast and that makes it quite a bit more addictive.
I never heard this theory before, but on the surface, it sounds reasonable. Is there any published literature about it?
One thing to keep in mind when discussing "drinking in moderation": I think people will surprised to learn that each country has their own (often, very different) idea of what it means. For example, healthy alcohol consumption levels are persistently higher from UK public health sources, compared to the United States (mostly CDC). My sense (that I cannot prove): If your median alcohol consumption is higher than global average, your national health agency will have higher values for "drinking in moderation". I would love to see a comparison of wealthy (primarily) Judeo-Christian countries and their recommended limits.
There's something to what I wrote, but I'm realizing I don't understand it as well as I thought.
As for what moderate is: I can absolutely believe it'cultural. I was intending to post from a US/EU perspective, as that's what I'm most familiar with, but your explanation makes sense.
According to the best longitudinal study from the WHO There is no safe level of alcohol consumption (1). Every molecule of alcohol is measurably bad for you physically
People want to conflate their personal desires and perceived social benefits with health benefits because mental health and physical health have overlaps
However they need to be separate to actually make a claim about alcohol as there are millions of people in functional communities that reject alcohol and have non of the negative effects
> According to the best longitudinal study from the WHO There is no safe level of alcohol consumption
This would be a misquote, but you didn't use quotes. The WHO said that they "can't identify" a safe level of alcohol consumption and that there are no studies that would indicate a safe level of alcohol consumption.
> The new WHO statement clarifies: currently available evidence cannot indicate the existence of a threshold at which the carcinogenic effects of alcohol “switch on” and start to manifest in the human body.
I could see why you would say that from this:
> “We cannot talk about a so-called safe level of alcohol use. It doesn’t matter how much you drink – the risk to the drinker’s health starts from the first drop of any alcoholic beverage. The only thing that we can say for sure is that the more you drink, the more harmful it is – or, in other words, the less you drink, the safer it is,” explains Dr Carina Ferreira-Borges,
But that seems to be a mixture of some language difficulty and some intentional hyperbole.
>The WHO said that they "can't identify" a safe level of alcohol consumption and that there are no studies that would indicate a safe level of alcohol consumption.
How do you not then clonclude that there IS a safe amount such that you would then consume it.
What about this study would make anyone who is not an alcohol drinker become one?
The only answer here that is actually logical is that
you accept that there are risks to drinking alcohol, and that those risks are less important to you than whatever social personal benefit that you personally get out of it
That's not a study from the WHO, it's a public statement they published in Lancet. It's an opinion piece.
Moreover, the WHO never changed their stance when later big studies showed that the "lowest risk" level is not actually at zero drinks.
The timeline was this:
- In 2018, a study was published in the Lancet that was the Global Burden of Disease analysis. They claimed the modeled minimum risk exposure was zero drinks. But the magnitude effects were extremely small: moving from zero to one drink a day pushed the incidence rate of alcohol-associated health effects from 914 per 100,000 people to 918 per 100,000 people. This was the basis of the "no safe level."
- Four years later, an updated Global Burden of Disease analysis, also published in the Lancet, actually reversed course on this. After controlling for more factors, they found risk was actually slightly J-shaped, and for 40+ year olds the theoretical minimum risk was about 0.5 drinks per day. The WHO ignored this updated study when they wrote their opinion piece in 2023.
- Another big study with ~600k participants published in 2018 in the Lancet shows lowest total risk consumption levels at 1 drink per day, with mortality rising above that level. The WHO also ignored this.
- The JAMA study in 2023, with >4 million participants, showed no benefit to low-volume drinking, and also no increase in harms from low volume (1 drink daily) drinking. Again, the WHO ignored it.
Several other studies have shown either no discernible risk change between zero and 1 drink per day, or lowest possible risk at 1 drink per day.
But the WHO and much of the public health policy community ignores almost all of the actual credible science on this because it does not fit with their ideology. It's a huge problem with public health policy: the actual scientists and statisticians are doing great work, but if it's not what the policy people want to hear, they discard it and keep repeating their mantra.
One thing I never see pushback about: Northern Virginia has the highest density of data centers in the United States. Google tells me there are over 600 data centers in Virginia (that is nuts!). This build-out has been happening for more than 25 years. Until very recently, I have not heard any pushback from people in that area. What am I missing?
Yeah man, I'm sure the largest industrial mfg country in the world is desperate to buy time to outbuild the decaying husk of a mfg country that can't even build the transformers needed to power these data centers.
I assume that data center builders can get cheaper land and/or lower taxes and/or easier regulations elsewhere. Also, the US is gigantic, so you want to have data centers geographically dispersed to reduce ping time for customers.
A nuclear power station puts out many times the waste heat of the sites mentioned in the article. We've apparently mostly made peace with the local impacts of those.
Thermal efficiency is in the 30s, so roughly 2x the electrical output as waste heat. Looks like waste heat from nuclear is still quite a bit more than from data centers.
Data centers are growing quickly and that's not including the waste heat from generating the energy the consume, just taking a quick look at the relative direct emission.
> A nuclear power station puts out many times the waste heat of the sites mentioned in the article. We've apparently mostly made peace with the local impacts of those.
Although to be fair, when was the last time we actually managed to build a new nuclear plant?
Hmm, I hadn't considered additions to existing plants versus greenfield plants, fair point. Presumably those have less local resistance and permitting issues considering it only took fifteen years from application to generating power. Guess there's probably a market in selling existing datacenters to the hyperscalers to expand them!
So we got a lot or bike paths and roads instead of the old railway tracks, many of them built during railway mania boom cycles. I wonder what will we use these shoddy windowless boxes for ?
I likely align with you on most political and social issues since you seem to be in despair. However, this question is seriously misguided. This is by far the best time to have been a human in all of history.
You say AI is making our creative pursuits feel meaningless?
How many creative pursuits do you think the average person had throughout history when close to 50% of people died before age 5, fewer than 10% of people could read, women and minorities had close to zero rights, the average worker was impoverished and farmed sunup to sundown to barely eat and avoid famine, there was little treatment for diseases (infectious or otherwise), common people could be conscripted for war at a moment’s notice…I could go on.
You think the average person in history spent their time painting and writing? These are luxuries 99% of people who ever lived could have never dreamed of.
> How many creative pursuits do you think the average person had throughout history when close to 50% of people died before age 5, fewer than 10% of people could read, women and minorities had close to zero rights, the average worker was impoverished and farmed sunup to sundown to barely eat and avoid famine, there was little treatment for diseases (infectious or otherwise), common people could be conscripted for war at a moment’s notice…I could go on.
That's an industrial age work ethic. Before industrialization made everything cheap, you couldn't just go to the big store and buy stuff, and as a peasant you definitely didn't have enough income to hire a craftsman to make everything you needed.
People have historically spent a lot more time than we do now making things they need themselves, or things for other people to trade for their craft. Shoes, clothes, basic tools, food, and so on.
This is myopic “Silicon Valley” rhetoric and lacks understanding of art or why it’s appreciated.
We have video cameras and movies, yet stage actors on Broadway are celebrated like movie stars. People still read books. We have cameras and photoshop, yet a huge market still exists for oil on canvas. We have synthesizers, drum kits, and all the tools we need to make music on a computer, yet live concerts with guitars and drums sell millions of tickets annually. We have the technology to deliver payloads on another continent with millimeter precision, yet every few years, the whole world gathers to watch people who have trained a lifetime to run, jump, and throw stones.
“The old way is dead” doesn’t hold up in the arts like it does in the sciences. Technology moves on, but humans are the same. You can inspire and awe us today using the same tools and methods that were used hundreds or thousands of years ago.
Every art form you've mentioned still had to be created by a human. The only difference was how. Was it oil on canvas or stylus on Wacom? Was it film or digital? Is the synthesizer analog, digital or a VST plugin, and does it really matter as long as someone is still playing the keyboard with his own two hands?
AI has removed the human element entirely. This is a paradigm shift, and there's more merit in bemoaning using AI to wholesale generate a "photo" than there ever was in a film photographer bemoaning digital.
Adopting something like HTMX is primarily a UX choice, and the language you use to serve it is almost irrelevant to UX. JavaScript is a nice backend language. It’s fast, has a massive ecosystem, and many people know it well.
Sure. I personally think SPAs/React have better DX than HTMX. IMO, the most compelling reasons to use HTMX instead (for building dynamic/interactive web applications) are to reduce the amount of JavaScript sent to the client and make pages become interactive faster. These are UX concerns: how much the user has to download (waiting) and how long before the page becomes interactive (waiting).
With HTMX, you get these benefits regardless of what language you use on the server. These benefits could be reasons the parent commenter chose HTMX, and JavaScript is a perfectly good language to serve HTMX applications for the reasons I mentioned.
> the most compelling reasons to use HTMX instead (for building dynamic/interactive web applications) are to reduce the amount of JavaScript sent to the client and make pages become interactive faster
Next.js and TanStack Start don't send any JavaScript to the client if you decide to server-render everything. The upside is that you can opt-in into client-side components, and have all the interactivity benefits of React. I understand using HTMX if you are building a backend in Go/Python/..., but if you are already using Node, I see no reason not to use TanStack Start instead.
Maintenance burden, supply chain attacks and the constant dependency upgrade treadmill. With a very minimal footprint app, you can avoid all that to a large extent.
There are no concrete numbers for Valorant that I know of, but the “Head of Esports Partnerships and Business Development for North America & Oceania Riot Games” Matthew Archambault was quoted saying the Valorant player base is 30-40% women [1]. That seems plausible to me based on my own experience playing Valorant.
Do keep in mind: they also develop, maintain, support, and market Unreal Engine, which is possibly the most advanced and innovative game engine out there. It’s used by tons of other studios (including AAA) and is used in other industries like space flight, automotive, etc.
I think online/digital gambling is worse because it follows you everywhere. I don’t like any form of gambling, but at least with casinos there’s some escape in not physically being there. It’s also harder to enforce age requirements online.
They all have apps these days, and just like a local bookmaker might "accidentally" remove your name from their legally required self-ban list it's very common that a "bug" in your phone app means you can keep gambling after saying you want to stop.
"Mistakes" in the controlling party's favour are extremely common in such industries. Fluke 100-1 sport betting win? Oops we forget to fill out that mandatory anti-fraud paperwork, bet is off. Lost that 3-2 bet that the favourite would place in a horse race but actually you didn't show proper ID? Sorry that's your problem, we're keeping the money
Alex Pretti escalated the situation? Have you seen the video? Did you see how the ICE agent shoved that woman? That was not “law enforcement activity.” It was assault. Alex Pretti had every right to be there, and he tried to help a woman who was being physically and illegally assaulted.
> I haven't heard of ICE hurting any actual immigrants in custody
Re: Minnesota - Everyone out there escalated the situation. Detaining people could be a very boring process, and I haven't seen any footage of any of the detained people starting any fights. Seems like it's mostly leftist white "heroes" running around.
I fully agree that it's terrible that those people have been shot to death by ICE officers, pretty much regardless of anything. But what I'm asking is what they're even doing out there? Why do it?
It seems more like these protests are not about the specific individuals being detained, but a cool group activity to engage in now, so what makes all illegal immigrants, who respect our country so little that their first interaction with it is to sneak in, automatically deserving of being shielded from the consequences of their own actions?
Re: ACLU - it's interesting that it seems like all those who are claiming harm are Mexicans who were given the choice to go back home to Mexico but refused. (If their claims are true, those officers should be punished and fired, but also can you think of any good reasons for detainees to lie about this if they're desperate to be allowed to stay?) Should the US admit all 130 million Mexicans as refugees, just on the grounds that Mexico is allegedly so bad? Or only the ones who break the law and come here?
I'll support accountability for ICE officers who do bad things all day, but it seems like the leftist goal posts are "Don't have immigration officers exist at all, and don't let local law enforcement even consider helping get people deported" and that's why people think Dems want open borders.
I wouldn’t put them in the same category. Bootstrap has a much stronger opinion about how you write and structure CSS, while Tailwind is a framework for building a style system and writing arbitrary CSS that conforms to it. In practice, Tailwind is actually much closer to writing vanilla CSS. It does come with a widely used default (and very good, IMO) style system, but you ultimately apply arbitrary CSS rules just like vanilla.
Tailwind is certainly divisive. It’s better suited to component-based frameworks than templates or plain HTML, so a person’s background likely contributes to how they perceive it. Personally, I’ve worked a lot with vanilla CSS, opinionated libraries like Bootstrap, and Tailwind, and opinionated libraries are the only one I have no desire to use again. I generally want Tailwind in component-based projects and vanilla CSS in template-based projects.
At the end of the day, pick what you like and use it. They all get the job done.
No, it's more like saying "In the context of Java, addition refers to the concept of adding two numbers" which yeah, I guess is technically true but the concept of adding was not invented in Java nor works differently in Java, it's a general math thing. Writing it that way implies that somehow addition works different in Java, which it doesn't.
If I drink one glass of wine per day, it’s probably fine. If I drink three bottles per day, I’m gonna start having problems.
[1] https://ourworldindata.org/grapher/monthly-spending-data-cen...