AnthonyMouse

When you're outside you're not inside the walls of a data center.

for example adult content that has huge chargeback risk or straight up blocked by other processors

This has the default inverted.

The best place for the traditional payments system is when you're making a large purchase from a questionable seller, and then you're willing to pay extra for the assurance that you can reverse the transaction.

If you're making a small purchase or buying from someone you regularly do business with, why would you want the cost of other customers' chargeback fraud or use of stolen cards to be added to the price you pay?

Ultrasound is extremely non-penetrating and wouldn't make it through the walls of a building. Infrasound is the opposite and would go through most anything, and in principle can be produced by anything that moves, but the thing to care about at those frequencies is the amplitude, i.e. the sound volume. It's a meaningful concern from things like wind farms that have massive slow-turning turbines, but there is nothing in a data center that moves that kind of volume. On top of that, having multiple independent small sources in the same place lessens the effect, because multiple sources that emit at the same frequency but out of phase with each other cancel each other out.

Undoubtedly if you get some measurement equipment and stand next to a data center you could register some infrasound, but the same would be true if you stand next to an ordinary office building or in an empty field, because infrasound carries for hundreds of miles and there are much larger sources than data centers.

"There is a sound but you can't hear it trust me bro" is solidly into YouTube quacks territory.

The whole thing is a motte and bailey game. The largest of cloud companies might build a data center which is 1GW. There won't be 4000 of those and it won't be next to your house. A huge number of commercial office buildings already have a 100kW data center in them, and have for decades, without anyone having any reason to know or care. For those 4000 is certainly an underestimate.

The average person around them knows quite a bit about the noise level, and how far it carries.

Do they? How much noise do you imagine comes from a data center that isn't operating a gas turbine?

I think this is because the demand is so extreme for datacenters now that companies are building them in a manner that disrupts the environment around them.

This would make sense if the ones being targeted are the ones that do this, but that isn't the case. Moreover, those would be simple to address, e.g. pass an ordinance that prohibits long-term operation of fossil fuel generators, defined as operating them for more than 40 days out of each 700. Or impose a reasonable decibel limit as measured at the property line.

Meanwhile people are proposing "data center moratoriums" and so on, even if the data center would be built in a non-residential area with a closed-loop cooling system and generate its power from a completely silent and non-polluting solar+battery system. Which makes it pretty obvious that the motivation is something else.

As far as I can work out, the cost is competitive as long as you're using the capacity every day, especially when it lets you avoid paying transmission costs. The thing that doesn't work is to have enough battery storage to handle extended periods of low generation, because for that you need dramatically more storage capacity while only using the surplus on rare occasions. But a data center could use the grid or backup generators on those rare occasions while still using batteries overnight every night.

Complex societies exist and work.

They certainly exist. Whether they work is rather the question.

Everyone who has a choice makes the choice, dominantly, to stay in them.

Which people actually have the choice? If a group of people want to stake out a piece of land somewhere -- even if they pay for it -- and then try to operate some kind of self-contained society there without being subject to an existing government's laws or taxes, what happens to them?

There isn't a lot of land on earth which no existing government claims is its jurisdiction.

Imagine a system with competition but no taxation. You lose public services.

You lose tax revenue. That isn't the same thing.

Suppose nobody is maintaining the road in front of your house and there is a huge pothole, or the road isn't paved to begin with. You and a few of your neighbors, with nobody forcing you to, agree to split the cost of paying to fix it so that people can get to your house. Maybe you even just pay for it yourself because the thing is right in front of your driveway. Attempting to charge a toll or something is pointless because there isn't enough traffic to justify the administrative costs and you just want the pothole gone. Does this have a different set of benefits and trade offs? Sure. Are there still various roads that are open to the public? Yes.

And then you have to ask whether having a third of your neighbors not chip in to hire the paving company costs you more than having the government pay 600% more to have it done as a result of various corruption and administrative overhead.

Same for competition without private employment–you're in a totalitarian state with a monopsony on labour.

A canonical example of the absence of competition.

The benefits of co-operation mean the real trade-off is 1 million split five ways versus 100 to me. This was almost untrue in the age of conquest. It became barely true with industrialisation. It's massively true in the information age.

The problem here is that it's true of specific things, not specific epochs. If all the government did was collect 5% in taxes from everyone and use the money to prosecute murders and maintain bridges then the result would be a huge net positive. Meanwhile in reality the government takes billions of dollars from ordinary people and gives it to the likes of Lockheed, Oracle and Microsoft.

For the amount of money the US government pays Microsoft for Office subscriptions and the like, it could pay to have an office suite developed and released into the public domain many times over. Instead it uses the incumbent, in turn requiring others to do so in order to have formats compatible with the what the government uses. Who benefits from this other than Microsoft?

Orthogonal concept.

It's the sort of thing people generally mean when they say that someone acting in their own interest can be in your interest, and is the thing which is happening in the example from the thread.

Concepts like taxation; deterrence through corporal punishment, jailing and fines; paying salary for labour; hell, religion–these are all about aligning individual self interests with collective goals.

And the practical implementations of all of those things are severely flawed to the point of questioning whether most of them are even net positive.

Taxes are supposed to benefit the public, and be paid with some fairness. In practice they go disproportionately to cronies or buying votes from affluent retirees, the tax code is so full of carve outs for special interests that it looks like swiss cheese and various political incentives cause it to impose severe benefits cliffs on lower middle income people that create poverty traps that benefit no one.

The criminal justice system on paper operates based on the rule of law, but the laws are so complex, overlapping and sparsely enforced that it really operates on whether a prosecutor is inclined to charge you with something. The results are mass incarceration and a system that enables a corrupt incumbent to use the threat of prosecution to extract favors.

The principal-agent problem inherent in hiring someone is well-known and is dramatically exacerbated by large organizational hierarchies that put long chains of inaccessible authority between the customer and the person ultimately doing the work.

Religion seems like a long debate but I don't think it would be controversial to assert that there have been issues there.

I'd argue competition is more an optimiser on these primitives. Not a primitive per se.

Try to imagine any of the others operating without it. You have to pay taxes but have no alternatives on which jurisdiction to live in or who decides how much tax you pay or how the money is spent, what happens? You want to be hired or use the money you earn to buy something but there is only one employer and only one supplier of goods and services, what happens?

These models do have the cyber offensive capabilities claimed.

People use this argument against every new technology. We need to license these new printing presses or subversive elements will use them to publish seditious literature. We need to ban strong encryption or the government won't have invisible warrantless access to everyone's private messages, think of the children. 3D printers can be used to make gun parts -- as can a variety of ordinary tools people commonly have at home, but never mind that bit.

The sad truth is that a lot of people are not going to believe it until something happens and people die. Successfully preventing that from happening will be seen as evidence that the prevention wasn’t needed.

A 12 oz bottle of water is too dangerous a technology for ordinary people to have on an airplane. Four 3 oz bottles and an empty 12 oz bottle to pour them into after passing through security is totally fine though, naturally. And we need to keep this up forever, or don't you remember 9/11?

The issue here is not that it's impossible for 12 oz of unknown liquid to damage an airplane.

Complex society is a potent counterargument to this hypothesis.

Complex society is the demonstration of that hypothesis. Misaligned incentives are widespread and corruption and inefficiency are the result.

Systems that rely on good people to work are fundamentally flawed. Instead, the game has to be about aligning self interets in favour of the collective.

But now you're making a different argument.

"The enemy of my enemy is my friend" works by random chance. When Evil Corp pays off Candidate A and Pollution Inc pays off Candidate B and then it's Candidate B who gets in and retaliates against Evil Corp for backing the wrong horse, you're getting a good result by chance rather than by design. All it would have taken was for Candidate A to make a better prediction about whether they need to bend the knee to Pollution Inc too in order to win and the same system produces something even worse.

How to actually get their incentives to align is an extremely unsolved problem. The best method we know if is to subject them to competition, e.g. break up concentrated markets and place strong limits on what lawmaking can happen centrally, leaving everything possible to state and local governments while allowing people free choice in where they live, so that no one is forced to stay in the jurisdictions that make the worst choices. But the forces of corruption want the exact opposite of that, and have been gaining ground.

Anything using 1GW of power is going to be "enormously expensive". The all-in cost of a 1GW data center is on the order of $50 billion. A billion dollars in storage batteries is ~2% of the cost, which is completely reasonable.

~1600 GWh of lithium batteries are produced each year and hundreds of GWh/year of production capacity have been added in each of the last few years, so ~12 GWh isn't going to upend the market.

The two obvious problems being that a) you're advertising that so that people in the countries you actually intend to do business in can find it, and b) the same IP address may be hosting thousands of independent sites any subset of which could be intended to be accessed by people in the other jurisdiction, and publishing the route is done for them.

People do care about income streams for their descendants or charitable organizations etc. So it’s not completely ineffective.

It is once you make it to the other side of the marginal utility of a dollar.

Suppose you write a popular work in your 20s. You make plenty from it, and then people keep buying it. As long as they do, why spend your time creating new works when you can drink whiskey or play video games or live in a cabana on a beach? As long as the copyright doesn't expire, you have all the money you need and thereby value having more time over having more money. When it doesn't expire for 70 years after your death, so do your kids.

Whereas if it would expire after 14 years and you want to provide for your kids (never mind that they should maybe make a contribution of their own) then to do it you would have to keep making new works.

This is like saying that having a trademark in the US is advertising in France because you're paying the USPTO to publish your trademark registration.

If your scenario makes it impossible for someone to do ordinary business in country A but not in country B then it's fundamentally unworkable.

Also notice that the FBI has no jurisdiction over foreign domains. It happens that the .com (in addition to .us) registry is in the US, but e.g. the .ru registry is in Russia and not under the jurisdiction of US courts.

If grid storage batteries are economical in the general case then why wouldn't they be doubly so in a place where the inverters are a sunk cost and you're saving the transmission costs because the load is at the same site as the batteries?

The competition isn't the cost of generators, it's the cost of grid power, or the cost of operating gas generators every day.

The general problem is that we've made construction suck for a while now and people are suddenly paying attention to it but only in the case of data centers.

You don't get caught paying bribes if government officials aren't accepting bribes. The problem is the governments have been operating in a mode where blocking everything is the default and paying bribes is the expected way for things to get approved.

The whole problem is difficulty getting grid connections on Earth. If you can just hook up to the grid then why not just use it for everything.

Because you already need inverters and batteries for backup power and then using solar + batteries for ordinary operations and the grid for backup is cheaper than the other way around.

Which is also why you may not even need the grid connection. Suppose you have diesel generators to use in case you have some kind of a fault in the solar system, the same as was traditionally done in case of a failure of a grid connection, and then use those for the one week every two years when solar generation is too low.

The Texas grid that already generates more TWh from renewables than any other state?

If only data centers had some kind of backup systems to prevent power interruptions, like inverters with batteries attached to them. They could call it an Uninterruptible Power Supply and then charge the batteries when other power is available and run on them when it isn't.

Have you considered the implications of what you're saying?

A whistleblower goes to a therapist, stressed out over their pending decision to reveal official misconduct. They've been investigating ways to post something on the internet that can't be immediately taken down by the corrupt government officials they want to expose. They express their discomfort, in confidence, to their therapist, about using something they've discovered is also used for CSAM.

You think it's a good thing for the therapist to be required to report this? Should they report that the patient admitted to viewing CSAM with no context so the whistleblower gets investigated and arrested, or should they provide the context -- that the patient is about to expose the corruption of the government receiving the report?

For that matter, consider what it does when someone is actually a pedophile. They find out that if they try to seek therapy to address their perverse attraction to kids, the therapist isn't allowed to keep their confidence and they'll be arrested, so instead of seeking professional help, they keep abusing kids. Is that the result we wanted? There is a reason doctor-patient confidentiality was a thing.

If there was a teapot at a proposed orbit then there would be various falsifiable ways to prove it one way or the other. Once you do the experiment, you have the evidence. Before you do the experiment, you don't.

If there is a cancer risk then good evidence either way would require randomized controlled trials, which for long-term exposure are very expensive and take a long time, so then nobody does them and nobody really knows the answer. This is not the same thing as knowing it to be false.

It is if you want your food cooked in under five minutes. The question is, what happens at a lower frequency but a longer period of time?

People are looking for any excuse to complain about data centers right now. Building transmission lines is bad because you're taking peoples land; not building transmission lines is bad because there isn't enough existing capacity. The first one is obviously preferable but when you're looking for an excuse it works out either way.

What the complainers (i.e. the media companies) actually want is for the AI companies to stop letting people use AI to answer questions they used to read media articles to answer.

A wind turbine which is ~30 miles from shore can't be seen from the coast because of the curvature of the earth. Do you actually think you'll get more opposition to a wind farm nobody can see from shore with a 30 mile undersea power cable that correspondingly doesn't require any use of eminent domain, than needing to seize a line of land 1500 miles long for a transmission line?

The current admin doesn't have a long enough shelf life to be making any long-term plans based on their proclivities.

Commodity market profitability is determined by marginal cost of production. LLMs have marginal cost; traditional software does not.

This is the story for Nvidia/AMD or cloud providers rather than OpenAI.

With increasing inference as % of total compute, if labs create efficient models -- which they can, because they can create highly optimized models amortized over very high inference loads -- they can be low cost producers, and be competitive at $/task rates

It seems like there would be problems with this on both ends.

For general purpose models, everybody is trying to make them efficient, so you can't win just by being slightly more efficient. You would have to be so much more efficient that you can charge high margins while still capturing the majority of the market so that the high margins get multiplied by the majority of users and the users you leave on the table aren't funding open competitors. Meanwhile everyone else is also trying to improve efficiency, so one misstep and you're behind.

Example of where this can be a problem: You spend a preposterous amount of money to create an efficient model, then someone else publishes a paper with a new technique that gets a similar but incompatible efficiency improvement out of a model that costs a lot less to create. You have now spent an enormous amount of money in exchange for no competitive advantage.

And from the other end, one of the best ways to get efficiency is through specialization. A general purpose model can generate code or summarize a meeting transcript, but a special purpose model can do it as well or better with far fewer parameters and resources. But then you don't have a situation where one huge AI company has The Most Efficient Model, you instead have dozens of specialized models produced by independent sources that are each the best in a given niche. Any proportion of which could have open weights, or have an arbitrarily small advantage over the ones that are.

Moreover, these problems combine: Both the computing hardware vendors and the AI companies want the margin on doing inference, but the more of it one of them gets, the less the other does. If the AI companies were actually getting huge margins then it would be in the interests of Nvidia, AMD, Apple, Intel et al to fund efficient open weight models in the same way they fund Linux. Commoditize your complement. And those models don't even have to be better, as long as they're good enough that the closed models can't charge a significant premium and the margin shifts back to paying for hardware.

Data centers are a fixed load with on-site backup generators that can be used to shed load when the grid is stressed. Grids actually like that sort of thing. It lets them use more cheap renewables.