So maybe someone can open a new sandwich shop and accomplish the same thing without screwing everybody else in the process. Not only that, Lewiston probably doesn’t have a glut of data center talent seeking employment —I wouldn’t be surprised to hear that not a single person living in Lewiston when a project like that was approved would be employed there.
The data center jobs are middle class. With progressive taxation, those 30 middle class jobs generate about 100 average people's worth of revenue. Presumably Lewiston has schools which could use the funds.
And presumably everybody in Lewiston already has lunch somewhere, so a new sandwich shop that was successful would put another lunch spot out of business, for a net zero jobs.
Edit: the larger point here is that jobs, particularly skilled jobs, don't grow on trees (much like money). New technologies and new jobs replace old technologies and old jobs. If you put a ban on new technologies and new jobs, you'll just have more unemployed people.
Bingo. Data centers are a net negative wherever they are. Giant, employ far fewer people than a grocery store after they’re built, crank up electricity costs, use tons of water, air pollution if it’s self-powered, noise pollution (it’s really worth watching Benn Jordan’s video on infrasound,) ugly… the only local entities that win are the landowner and the municipality that collects taxes on them. Though I’ve seen some astonishingly misinformed politicians offering big tax incentives for data centers not realizing that they employ so few people. From what I hear, even much of the construction is done by flown-in contractors with experience doing it elsewhere.
The people that own these data centers have only themselves to blame. They’ve been obnoxious, at scale, for so long that damn near everybody knows how much they suck, and they’re losing their ability to railroad locals into eating their turd sandwiches.
Edit: I know it’s gauche to talk about votes here, but this comment trended upward consistently for 45 minutes. In much less than 10 minutes, it collected more than half that amount in downvotes. I’d eat my hat if there wasn’t some kind of organized/automated brigading happening here.
Edit again: Now close to 70% gone. Not exactly surprising given the forum, but pretty depressing nonetheless.
> Data centers are a net negative wherever they are.
They really shouldn't be.
There is a need for them and they aren't inherently damaging. There's no reason they can't be placed under some environmental regulations that cancel all their negatives, at least on some places. And they would still pay taxes.
But no, datacenter owners are using their connections to remove any regulation instead.
Obviously the solution is to tax them instead of ban them so they end up dispersing income to the surrounding areas. The entire point though is that they won't get built where they are taxed, and eventually, through regulatory capture or governance capture, they'll get built without having to compensate for their exteralities.
The cynicism of residents is reasonable. They've have to be highly educated to actually understand the implications of what they're doing and how that revenue can be distributed. America's decline lends itself toward small-town corruption, where patronage is more important than communitarianism, due to large and accelerating net worth inequality, and an economy where outcomes are based on inheritance over labor.
This explains the logic behind an outright ban. You don't have to be vigilant about corruption and the principle-agent problem if the thing is just banned.
>The entire point though is that they won't get built where they are taxed
I dont think this is entirely true. Maybe not the first wave of data centers, but there are a lot of factors that go into the cost calc and its possible that it would still be worth it to build them even if taxed.
He's not saying it's economically unfeasible to build where taxed. He's saying they'll simply build elsewhere where they won't be taxed.
About a decade ago, a bunch of data center companies got fantastic deals with my city (low/no tax). People are pretty upset about it. A few years in there was a report on how many people they employeed. I think combined it was under 10 who lived in the area.
The community is a heck of a lot poorer now because they were convinced to offer incentives for a factory that never came. Once these firms can dangle hope in return for tax treatment or infrastructure, then you have a zero-sum game between townships where the winner — if there is a winner — ends up being the firm first, and the loser — if there is a loser, will be the township first.
Unfortunately it’s a race to the bottom in most of America: If you pass such regulations locally or in your state, the data centers will simply choose to not build in your area of authority (county/state). Unless we were to pass sweeping, nation-wide regulations (which this administration is aggressively against because they believe we are in an AI arms race with China), those regulations/bans just drive the data centers elsewhere.
Maine obviously wouldn't have a problem with that, this law indicates they want them somewhere other than Maine. Environmental regulations that are as good as a ban seem far preferable to an outright ban, IMO. There's a large segment of the population that see outright bans as oppressive but support environmental regulations.
>Environmental regulations that are as good as a ban seem far preferable to an outright ban, IMO. There's a large segment of the population that see outright bans as oppressive but support environmental regulations.
So basically steal legitimacy from real environmentalists by applying their label to something that's not really motivated by environmentalism but can be construed that way?
"They don't actually want what I'm selling so I'm gonna dress it up as something else, they'll never know"
AreWeTheBaddies.jpg
The other problem you're gonna have is that this isn't an original thought. You're at least 20yr late to the party. So, so, so much absolute garbage has sailed under the flag of environmentalism that the public is starting to be more critical (see for example the kerfuffle over wind turbines off Rhode Island) and it's not unforeseeable that eventually the environmentalists are gonna have some sort of purge or reformation or reversion to more traditional environmentalism and serving corporate interests in order to reclaim some lost respect/legitimacy. Trying to sail "obviously not primarily about the environment" stuff under the flag of environmentalism is only gonna hasten that.
Unfortunately enviromental law has become a favorite tool of industry to stop, delay, increase costs to competition. Industry funded pretend enviriomentalist groups bringing up friviolous lawsuits has done much harm to the movement.
And a tool of industry to make easy money. You pay off the right people and your "environmentally friendly" product that's 99% as bad as the other thing while performing 50% as well can become "technically not mandatory but you won't get shit approved without it". Or some competing solution will get nerf'd driving business to you.
In a particular state in the midwest. The regulator has adopted the policy of "no new septics". You have to do a mound system at great cost. No rule, no code, no performance standard, just an unofficial policy of "we don't approve those". I know someone who's got a textbook perfect property for a traditional septic. They don't care. The shit pump people are laughing all the way to the bank.
But people probably wouldn’t have a problem with them building a data center in central Aroostook. Nobody making these regulations wants to simply stop data centers from being built anywhere— they’re trying to stop people from building them where it will really suck to have them, like densely populated Lewiston. I actually left tech to work in manufacturing. I know the value it provides and how much it can negatively impact others. Big companies want to build this shit near population centers because it’s more convenient, profitable, easier to hire people, etc. Tough cookies, I say.
> There is a need for them and they aren't inherently damaging.
One solution: local taxes on the economic value generated by the data center. MNCs love to play accounting games, so a simple formula based on metered GWh multiplied by reported worldwide revenue with a scaling factor a fraction of a percentage. This fund should be ring-fenced be address whatever externalities are introduced by the data center, including electric bill subsidies, infra maintenance, and funding independent oversight.
Perhaps they are simply not taxed enough to benefit the community. If the local municipality is bearing a lot of these hidden costs, then perhaps the taxes need to be higher and directed at efforts that mitigate the worst of the problems. Water management solutions, air pollution management. Are there ways to mitigate the noise pollution? It seems like they should be taxed /more/ to help offset the negatives. There is surely a way to mitigate the problems. For example, can the noise pollution be addressed by forcing more green spaces around them, etc?
Almost anything can be mitigated at some cost - but it has to be determined what those mitigations are, and then demand them.
Many municipalities are unequipped to deal with a "datacenter" because on paper it is the same as an office building (that draws a lot of power), where it should be treated like an industrial site (rail yard, factory).
True. There likely needs to be some sort of templating handled by states. Each data center and location will be different and require assessment. This does drive costs up for the data center, but I don't see another fair way to handle it really.
They get their own unique third category as unlike industrial sites there's no hazardous chemicals and even the noise pollution is substantially different in nature.
The old datacenters are analogous to office buildings that emit some unusual noise and consume large amounts of electricity.
The new ones (ie gigawatt class) consume enough electricity for ~1 million households and at minimum enough water for 100k households (but possibly many times that).
I believe evaporative cooling is the norm (thus my "possibly many times that" remark doesn't apply) however theoretically they could provide hot water as a utility or (as you say) just dump it into the sewer. If located next to a river or the ocean they could conceivably dissipate it that way but I'm not aware of any examples.
It's the sort of externality that could be solved with a well placed megaproject. A related question to my mind is why we're building such expensive strategic assets in the open rather than under a mountain.
Large industrial customers often have to sign up to load-shedding agreements wiith grid operators in times of peak grid load.
Can datacenters load shed when requested?
If not, they should be paying a premium/spot market prices for power if they arent participating in load management.
The one I toured (decades ago) and dual redundant multi-megawatt diesel generators (big boys, always ask for a tour when they’ll be testing them, it’s fun to hear them start).
Local utility would ask them to shunt to the generators now and then during potential rolling blackout situations.
The city making money off of it doesn’t make the impact smaller. You can’t tax away the air pollution coming from a gas turbine running in a populated area.
That was my point. It doesn't all have to be taxes. It can also be agreed upon mitigation maintenance. Better filtration on gas turbines, etc. Green spaces to mitigate sound impact. I don't know, I am just wondering if there is a model that can be designed that makes a data center "balance" within its local environment instead of getting the opposite, tax incentives. Right now I agree, they get to socialize the costs and reap the benefits of building data centers to a large extent.
That all sounds nice in theory, but does the Lewiston municipal government have the resources and expertise to determine what countermeasures would be effective? Would it be left up to the company paying for the mitigations to decide what’s reasonable? I think we know how that would turn out. Even in heavily regulated states, industrial pollution still heavily impacts people in the vicinity. They usually accept it because so many of them work there. This place was estimated to employ 30 people. We don’t even know if problems like infrasound are reasonably avoidable or mitigated, and it’s not like they can make more water. Additionally, the way the industry has conducted itself over the past decade has been abhorrent. There’s no reason to believe they wouldn’t try to circumvent every last shred of mitigation knowing the city has comparatively minuscule resources to fight it.
If we put them anywhere — and I’m not convinced we really need all of the data centers we have, let alone all the ones we’re building — they should not be in the middle of densely populated areas like Lewiston.
youre starting a good conversation but as per typical internet fashion you are being critiqued as though your direction of thought is being presented as some sort of final solution.
i completely agree that we should be looking into modelling this in terms of what is possible to mitigate its impact and what does that look like with current technology and costs, and where would we need to develop new tech, and what would be the critical values to hit to consider mitigation a success
The fact that they need to use gas turbines at all is a tragic condemnation of how the US can’t build shit at all. We should be consuming more (green) energy to make our lives better, and rushing toward diminishing returns on energy consumption. Instead, we have this unholy alliance of (usually right wing) NIMBYs and (usually left wing) degrowthers that make it much more convenient to use a gas turbine than build renewable energy somewhere windy/sunny and plumb it in with some transmission lines. Renewable energy is way past the tipping point of being cheaper, the gas turbines are just there due to regulatory burden at all levels.
Yeah, but unfortunately, here we are, and there are the companies that want to build these things in completely inappropriate areas because it’s more convenient.
noise pollution (it’s really worth watching Benn Jordan’s video on infrasound,)
Noise from data centers is a real issue, but Benn's measurements and analysis are not great (speeding up the sample rate to demonstrate frequency effects is just wrong, among other issues).
It was a bit misleading in terms of the audibility of infrasonic noise, but I think he did a good job of highlighting some of the effects of infrasonic noise on QoL/health with the study towards the end. IIRC, he also recorded some regular human-range noise that I would personally find disruptive to have to live with (though this was a fair bit closer to the data center than the claimed range of infrasonic noise's effects)
Doesn't this also apply to new housing? Strain on services per job created is probably even higher. The benefits are for someone currently not living here, just like data centers used for remote users. And if cheaper housing is available obnoxious poor people might move in. I think there should be a moratorium. Not in my backyard!
I’m not sure who you’d expect to sway equating data centers to east coast urban housing during a giant, sustained housing crisis, but your obviously disingenuous argument is completely ridiculous.
At least my comment is not completely content free. How is it different?
First, the gp comment says the data center is not good for existing residents - obviously true for housing too, which you didn't refute.
Then it assumes statewide ban should be based on personal preference of local residents. That is just a definition of nimbyism. While in reality I am a YIMBY and the end was part sarcasm, I would genuinely prefer living next to a data center, rather than next to non immigrant poor in the US. I grew up lower middle class or poor by US standards, and also live in Seattle, lots of experience. So I say along with data centers we have a statewide ban on anyone who is lifetime net consumer of tax money anywhere near my backyard.
> Though I’ve seen some astonishingly misinformed politicians offering big tax incentives for data centers
My national government is currently giving massive tax breaks for one of these. It's going to be, after all, "the biggest foreign investment in the country ever"...
- Datacenters don't use tons of water compared to just about any other business.
- It's not really fair to blame datacenters for using the municipality's electricity and then blame them again for being loud and generating electricity instead of using the municipality's electricity. Any given datacenter is only doing one of these.
- Presumably, the municipality's tax dollars benefit the people who live in the municipality in the form of better services
This thread is obviously being brigaded. Wiped out like 30 upvotes on one comment in no time. If you need to silence people talking about your company or favorite toy, maybe you should re-examine your life choices. Pathetic. Is there some polymarket nonsense going on here?
People who are insecure about their expertise in many subjects, (especially within art and the humanities, in my experience,) will unfailingly use any opportunity to point out when they know something someone else doesn’t. However, if someone posted “check out my super rad FP framework,” but it was actually OO with some bolted-on FP ideas, even if it was neat in its own right, people here would be totally justified in saying “hey… about that whole FP thing...”
Even if they look similar, there’s a big difference between sour grapes ego boosting, and people with subject matter expertise pointing out common misconceptions. A major problem with Engineer’s Disease is mistaking the former for the latter based on maybe having read a few blog posts and falling down a research rabbit hole once two years prior.
It’s not that straightforward. Art directors and designers get paid to visually communicate things the business wants to communicate— anything from brand vibes, to directing people to click on a “buy me” button, to the state of an interface. Most designers in tech companies aren’t even the ones that design things like branding — that’s done by specialists in extremely well-compensated studios, and corporate designers are stuck following their guidelines. Taste is nearly irrelevant to an interface designer, for example.
Don’t feel bad — most people honestly have no idea what designers really do, and I think developers’ proximity to them instills false confidence in their understanding of it. (I worked in both fields for years, and am also a graphic artist.)
A great example of this understanding gap is people thinking interior designers choose color schemes, sconces, and throw pillows. That is an interior decorator—- someone who’s professionally tasteful. An interior designer has an architecture license and designs the inside of buildings, often including everything from where fire exits are, where utility lines flow, what floors everything is on, and even things like wayfinding signage.
Even the most visual practical design is often more about visual hierarchy, semiotics, using gestalt to convey informational relationships, and conceptual thinking than taste.
> The author also needs to improve their grammar. The occasional capitalisation is diabolical. Either do, or dont, and at least if you dont its obvious youre a child.
* Either do, or don’t, and at least if you don’t, it’s obvious you’re a child.
There was a paper recently that demonstrated that you can input different human languages and the middle layers of the model end up operating on the same probabilistic vectors. It's just the encoding/decoding layers that appear to do the language management.
So the conclusion was that these middle layers have their own language and it's converting the text into this language and this decoding it. It explains why sometime the models switch to chinese when they have a lot of chinese language inputs, etc.
Pretty obvious when you think that neural networks operate with numbers and very complex formulas (by combining several simple formulas with various weights). You can map a lot of things to number (words, colors, music notes,…) but that does not means the NN is going to provide useful results.
Everything is obvious if you ignore enough of the details/problem space. I’ll read the paper rather than rely on my own thought experiments and assumptions.
Oh, Jesus Christ. I learned to write at a college with a strict style guide that taught us how to use different types of punctuation to juxtapose two ideas in one sentence. In fact, they did/do a bunch of LLM work so if anyone ever used student data to train models, I’m probably part of the reason they do that.
You sound like you’re trying to sound impressive. Like I said, I’ll read the paper.
It is text prediction. But to predict text, other things follow that need to be calculated. If you can step back just a minute, i can provide a very simple but adjacent idea that might help to intuit the complexity of “ text prediction “ .
I have a list of numbers, 0 to9, and the + , = operators. I will train my model on this dataset, except the model won’t get the list, they will get a bunch of addition problems. A lot. But every addition problem possible inside that space will not be represented, not by a long shot, and neither will every number. but still, the model will be able to solve any math problem you can form with those symbols.
It’s just predicting symbols, but to do so it had to internalize the concepts.
This gives the impression that it is doing something more than pattern matching. I think this kind of communication where some human attribute is used to name some concept in the LLM domain is causing a lot of damage, and ends up inadvertently blowing up the hype for the AI marketing...
I think what's causing a lot of damage is not attributing more of human attributes (though carefully). It's not the LLM marketing you have to worry about - that's just noise. All marketing is malicious lies and abusive bullshit, AI marketing is no different.
Care about engineering - designing and securing systems. There, the refusal to anthropomorphise LLMs is doing a lot of damage and wasted efforts, with good chunk of the industry believing in "lethal trifecta" as if it were the holy Trinity, and convinced it's something that can be solved without losing all that makes LLMs useful in the first place. A little bit of anthropomorphising LLMs, squinting your eyes and seeing them as little people on a chip, will immediately tell you these "bugs" and "vulnerabilities" are just inseparable facets of the features we care about, fundamental to general-purpose tools, and they can be mitigated and worked around (at a cost), but not solved, not any more you can solve "social engineering" or better code your employees so they're impervious to coercion or bribery, or being prompt-injected by a phone call from their loved one.
Except I actually mean to infer the concept of adding things from examples. LLMs are amply capable of applying concepts to data that matches patterns not ever expressed in the training data. It’s called inference for a reason.
Anthropomorphic descriptions are the most expressive because of the fact that LLMs based on human cultural output mimic human behaviours, intrinsically. Other terminology is not nearly as expressive when describing LLM output.
Pattern matching is the same as saying text prediction. While being technically truthy, it fails to convey the external effect. Anthropomorphic terms, while being less truthy overall, do manage to effectively convey the external effect. It does unfortunately imply an internal cause that does not follow, but the externalities are what matter in most non-philosophical contexts.
>do manage to effectively convey the external effect
But the problem is that this does not inform about the failure mode. So if I am understanding correctly, you are saying that the behavior of LLM, when it works, is like it has internalized the concepts.
But then it does not inform that it can also say stuff that completely contradicts what it said before, there by also contradicting the notion of having "internalized" the concept.
You never met a person that isn’t always right or one that makes up shit to sound smart? Because that’s the pattern you are describing that is being matched.
If you look at the failure modes, they very closely resemble the failure modes of humans in equivalent situations. I'd say that, in practice, anthropomorphic view is actually the most informative we have about failure modes.
after you go from from millions of params to billions+ models start to get weird (depending on training) just look at any number of interpretability research papers. Anthropic has some good ones.
I heard that they might have fixed the problem, but I initially dropped it when they stopped respecting quotes, even in verbatim mode. Like, if I’m looking for an obscure product number, I don’t want a bunch of shit with a few digits off if there are no actual hits. I want no hits if the settings and query demand it.
In my experience, most garden variety security problems stem from a) the developer not understanding the implications of something (maybe because they’re new, or operating outside of their usual domain,) or b) the developer not paying close enough attention to realize they did something they know is stupid. We’re only human.
Vibe coding obviously doesn’t make something insecure, per se, but saying it doesn’t reduce the attention paid to any given line of code, or encourage less knowledgeable people to write code, seems pretty dubious to me.
The Claude Code team is clearly competent and professional, yet they accidentally published the proprietary source code for one of the world’s hottest products. That’s like a Bank manager walking away with the keys in the door and alarm disarmed. When’s the last time you heard of a human team of developers doing that?
Again, I’m not saying that vibe coding necessarily creates unsafe code, but I don’t see how anyone could say vibe coding was devoid of security implications. I think this is an organizational/logistical problem that we’ll figure out at some point, but in think it’s going to be more of a C buffer overflow ‘figured out’ that never really goes away.
Very reasonable take, I agree 100%. But I don't you're putting any responsibility with users of the such very vibe coded apps. OpenClaw was primarily marketed towards devs and people in touch with IT. They should know better.
Sure. I reckon blaming the system for the intentional actions of a few is a great way to avoid individual accountability. Conversely, blaming many individuals for fundamental systemic or leadership problems is a great way to avoid accountability for leaders and systemic beneficiaries. It’s not rational to exclude either.
I’m also not sure that the distinction of dev makes much of a difference in this space because chatbot marketing works pretty damn hard to imply everybody is a prompt away from being a developer. How are those people going to know that they aren’t even qualified to make any given technical decision, let alone evaluate the output of a confident chatbot that’s magically writing programs for them?
[…]the data center would have employed only about 30 workers, the city estimated.
reply