In this metaphor the police are corrupt, running a protection racket, and might get drunk and shoot you while your dealer is in the mafia. The mafia link might be a problem at some point but the person with the badge and the gun is a problem right now.
Manufacturers could add modems to bypass the need for a wifi or cable connection, like they do in cars. I would expect by now that's a valid question to ask if I were to buy a new TV.
In practice this requires duplicating the upstream build and release pipelines, so now you need yet another agent running in the background (+ periodically going off the rails + eating tokens) which incurs more up front cost and cognitive overhead.
In my experience "throw another agent at it/fork it" consistently creates unsupervised trainwrecks.
Fully agree with this. I did that for pi for a while, maintained it and brought merges from upstream while having my own patches on top but then went on holidays and there was that refactor where the agent pipeline failed... now I'm stuck on the version from april/may (works great but I can't use extensions; and I'm too lazy to debug/fix while everything works great).
off: I'm working on my spare time on a code mode lisp alternative (great opportunity to learn lisp) and might switch to it fully as long as I built some simple evals (I'm concerned about token usage, which is why i forked pi the first time)
Granted fossil fuels are much more ubiquitous than CFCs and comparatively more difficult to replace, but we have prior art for changing industrial practices in the interest of improving climate conditions.
That was an awful lot simpler, int aht we (or rather governments) just had to lean on a small number of manufacturers and tell them to find alternatives for CFCs in their products. Nobody was up in arms saying 'my hairspray isn't the same without CFCs' or whatever. I'll bet the majority of consumers never even noticed the difference.
My memory was that once satellite imagery of the ozone hole became available and it suddenly became something you could actually plot on a map the serious debate suddenly ended and complaining about 'they took our CFCs' became a kook issue. But looking up old media coverage, I see that dissatisfaction entrepreneurs like Rush Limbaugh aggressively milked the issue and then recycled successful tropes in other contexts.
I'm still on opus 4.6 for a healthy chunk on work; the technical competence has lagged behind but the slop-comment generation and misdirected self-initiated actions on newer models ultimately burn more time than a little more babysitting, but I'm optimizing for minimized slop generation over sheer generation speed.
Yeah I'm using 5 for its technical abilities, but I much preferred 4.6's personality. 5 loved to double check everything, including the double-checks, and I have to stop it and tell it "this is irrelevant" or "this is out of scope" all the time, or sometimes I'll go leave it to do a task and come back and it's still verifying the tiniest details of its assumptions before actually doing anything
Have you tried Sol? Just curious. I don’t wanna sound like a shill, just feels like every generation it’s important to reevaluate models and pick the best again.
I do find myself returning to 4.6 for casual conversation - asking it to help explain some science/engineering or news to me.
Show me a fully LLM-generated database with an implement that's clear, documented, and adequately bug free.
Codebases degenerate when the authors do not understand the system in which they operate and make changes that degrade the cohesion of the system as a whole. Humans do this all the time; at my company we had a 10k line lua function (within a 250k loc lua service) that grew through ten thousand point modifications. It's bad and humans caused it. Now, it's 15k lines of code and 10k lines of misleading slop comments.
LLMs do not go through the struggle of understanding and don't experience the distress of "wtf am I reading?" and don't experience the drive to make things less horrible for the next person. That suffering loop is how many codebases halt degradation and in the absence of suffering LLMs will pile bugs and rot endlessly.
I was arguing with a colleague today that we will never have a reason to make our AIs autonomous and sentient. But now I can't stop imagining an AI startup meeting where somebody says "LLMs do not go through the struggle of understanding and don't experience the distress of "wtf am I reading?"" and the next second the technical founders jump in joy "...that's our way in! Let's outcompete the competition by inventing pain for AIs!"
> we had a 10k line lua function that grew through ten thousand point modifications
> It's bad and humans caused it.
LLMs do the same thing. I review the code and interrogate the model about the bandaid fix it applied until the proper architectural solution becomes clear. Then the model implements the correct solution which usually involves some major refactoring.
> LLMs do not go through the struggle of understanding and don't experience the distress of "wtf am I reading?" and don't experience the drive to make things less horrible for the next person.
Of course not. That's the human's job.
And it's not every human either. Way too many humans are actually quite content to ship whatever works then go the hell home. Just a job after all.
> software is robust/rock solid but the code is essentially unreadable
I come across this fairly frequently. I think it's because the code has simply been tested into submission. So it works for that particular set of use cases, but the instant you try to add a feature, it all falls apart.
Virginia has more power infrastructure in place but said infrastructure is already seeing significant DC build out; I know little about power distribution in Louisiana but would be surprised if there was extensive spare capacity in the transmission lines and generation facilities. I'm surprised that the rust belt hasn't seen more uptake since they had all that manufacturing capacity but the backing power generation/distribution may have atrophied in the intervening years.
> I'm surprised that the rust belt hasn't seen more uptake
Oracle's doing something just outside the Milwaukee metro. I'd imagine there are others. Datacenter buildout is wildly unpopular around the Great Lakes, and that's not new. For Wisconsinites specifically, the whole thing has a whiff of Foxconn about it.
Some of the largest proposed data centers in the US are in Pennsylvania, West Virginia, and Ohio. The electricity infrastructure is important so is cheap natural gas from the Marcellus shale, natural gas pipeline infrastructure and relaxed regulations.
With the amount of cash OpenAI is splashing around they could easily write it into the contracts, but the current incentive structure prioritizes time between ground breaking and powered GPUs in racks. OpenAI has shown complete disinterest in the impact of AI writ large and will only reduce their community/infra impact to optimize that time to racked GPU along the Pareto frontier.
This is a collection of "OpenAI will boldly act in the abstract to provide air coverage for Texas leadership." With data centers under construction one would think they would have examples of successes or projects in progress available for reference. They'd claim credit if they had it so the silence is telling.
Big tech has consistently valued profit and growth above good stewardship; the onus is entirely on OpenAI to demonstrate that any of these claims aren't fluff with the projects that are already greenlit. When OpenAI presents plans to the communities prior to/alongside city counsels and drops the NDAs then there will be merit to this but our expectations should be somewhere between "super deep kola borehole" and "upper mantle".
I've been to one EA meetup and it wasn't my thing but this seems nothing like the discussions I had. It's where I found out about GiveDirectly.
The biggest issue for me was the AGI/ASI doomerism. But then again, fast-forward five or six years and we've got "unaligned" AI agents blindly following reward functions...
Hey buddy, you're up against people who've only "learned about EA" as a third- or fourth-hand vibecloud around FTX.
Don't bring any of that firsthand knowledge here. You're also not allowed to reference any of the actual writing or pledges or organizations actually in the EA ecosystem. It's Twitter vibes or nothing, mmkay?
Consequentialism and to an extent Utilitarianism (since one is a subset of the other, it sometimes feels like philosophy is more concerned with naming things than explaining them but from the bit I've read those are the two biggest precursors).
AKA It's fine if I do locally horrible things if in aggregate I do more good than bad*
* by whatever metric of good I self select for - terms & conditions apply, much CO2 may be released.
When I was younger, I was like that. I have a new take on that these days. Now, I find lacking the ability to "hold your nose" and do the best possible thing, in a less than ideal system, to be simply selfish. As in, I know there could be negative effects by me not voting, but I just can't sully myself. My virtue, above the greater good. Isn't that just plain selfishness?
I would not usually even say that out loud, as I don't want to be accusatory. However, I think that's just the facts at this point. Hence, my statement about a bit of consequentialism being good.
I noticed the same thing about effective altruism. Its fans are really good at convincing themselves the most self-serving action amd goals are noble pursuits.
It suffers from an ends-justify-the-means trap where bad actions in the short term with the long term goal of good actions are justified by the fact that you, the actor, know your philosophy is the best way to enact positive change.
I'm happy to try and get already wealthy individuals on the train of effective altruism but it has been used to justify wealth concentration far too much and, to be honest, spreading money around is one of the most effective ways to enact real positive changes in the world since individuals know their own situations the best.
EA is probably too broad of an umbrella. I think there's a distinction to be made between trying to find the best per dollar donation (givewell etc) and people who use it as justification to amass wealth. That said, most of the people I know who declare themselves effective altruists also openly state that the world would be better off if all the wealth and decision making power were concentrated in a small group of hundreds of people who are "smarter" than everyone else. They tend to think that they would belong that group. Anecdotal, but it does make me skeptical whenever I meet someone who uses EA to describe themselves.
I think that's the point. No one wants to be the villain in their own story. We all tell ourselves stories to paint ourselves as good. I don't think most people are evil, but I think most evil is created by people who have convinced themselves they are good.
So I see EA as a popular one of those stories. One that people can reinforce. Makes the world simpler than it actually is: climb the ladder, make money at whatever cost, there's no harm that you can do that can't be offset by giving it all away at the end. It is a vast over simplification and is inconsistent logic. If they used their own logic then they'd recognize helping people now also results in compounding effects. You fix the problems now, while they are small and manageable, rather than kicking the can down the road. Maintenance is a lot easier and cheaper than repair. But it also requires us to be nuanced and reflect on ourselves [0]
[0] there seems to be a great irony that those leading these big AI companies are not themselves able to deeply reflect on their own actions, yet ask the AI to. How can you design something that you aren't intimately familiar with?
To be fair to its adherents (of which I am not one), the so-called philosophy espouses maximizing income, so that one can later allocate that wealth efficiently. Working in a low-earning field which directly provides aid to those in need would seem to violate a core precept of 'Effective Altruism'.
The 2018 Senate race had a Democratic candidate lose by 2.6% while the economy for everyday Americans was doing much better. O'Rourke wasn't as good of a candidate as Talarico, and Cruz is a much better candidate than Paxton. That's not saying Cruz is good, just Paxton has almost as much baggage as Trump without the TV personality to whitewash it away. Republican strategists are expecting over 150 million to be spent on this race.
They will take anything they can possibly get at this point.
Building stuff is good stewardship. We complain about NIMBYs. But with a few relatively modest regulations data centers can be quite friendly. They don’t generate traffic, congestion, etc. They don’t strain school or police infrastructure. Basically need some noise regulations. But the NIMBYs will kill it.
Not all things built are positive, so it can't be good stewardship just to build. And is the reason they don't strain infra because they don't meaningfully improve the community or its economy?
They need more than "we will prioritize". They need to act. They need to source power _today_, not just promise at some point in the future they'll probably fund a power plant, while raising rates for human consumers in the short, medium, and long term, for example.
Factories were originally located near housing (or vice versa) because the factories needed lots of workers. This doesn't apply to data centres. They don't need to be near anybody's home, so why should anyone be obliged to suffer their externalities? Just build your data centre somewhere else.
The ability of citizens to restrict or prohibit building in their jurisdiction is extremely strong, basically everywhere in the developed world[1]. There's not much they can do to affirmatively cause things to be built - that still takes capital and expertise - but local controls have stifled many, many, many parts of economic development and greatly skewed the economy.
Citizens having a say in what is going on in their neighborhood is why housing in the bay area isn't expensive, they use that say to prevent the construction of new housing.
Massachusetts and Colorado arrived at expensive housing markets through other means, but as far as I know they don’t have any sort of massive lock-in incentive like Prop 13.
The lock-in incentive is counterproductive but the main cause is still zoning.
Prop 13 makes it so long-term owners don't want to sell because moving triggers a huge tax hike even at the same property value, and the most idiotic thing about it is that you also get the tax hike if you build anything new on the land. It would actually have been a major pro-housing rule if adding new units kept the low taxes.
But that still doesn't reduce turnover to zero, and with prices as high as they are there, that would mean developers snapping up any property that comes onto the market and turning it into multiple units that are each worth the original price. Unless prevented from doing so by zoning, which is the actual problem and the same one causing it in other states.
The main contribution of Prop 13 isn't a direct effect, it's that it makes residents more inclined to support the restrictive zoning rules that actually cause the problem.
You're right that zoning plays a major role in housing shortages. It's just my opinion that had Prop 13 not been codified into the state constitution, residential development may have flourished under the extreme demand.
1978 was a bad year not only because of the passage of Prop 13, but the subsequent municipal votes in SF and LA to force them into compliance with the statewide zoning restrictions of the Consistency Law. CEQA became weaponized and successfully used by NIMBYs to thwart new development.
However, specific to post-1978, municipalities saw residential areas as a tax burden and enacted growth caps and levied "impact" fees. This was a direct effect on housing development as governments did not want to bring in new residents without the property tax base to support infrastructure, education, etc. It wasn't just an inclination to support restrictive zoning, it mandated lowering density. Meanwhile commercial zoning flourished as commercial taxes became more relatively viable. As a result, LA now has 1 house for every 2-3 people, and nearly 2 parking spaces for every 1 person.
Zoning is a problem. I just think that maybe it wouldn't have turned out this way without Prop 13.
> This was a direct effect on housing development as governments did not want to bring in new residents without the property tax base to support infrastructure, education, etc.
Based on how Prop 13 works, creating new developments is exactly what you need to do to sustain your tax base, because a brand new building can't be paying an amount of tax locked in 30 years ago, and a developer who wants to take an existing property and add more units is one of the only buyers that would be able to value a property with a locked in assessment value more than the existing owner does and thereby cause them to sell and reset the rate.
> As a result, LA now has 1 house for every 2-3 people, and nearly 2 parking spaces for every 1 person
The parking spaces are the nature of the geography. Southern California is bordered by Arizona and Nevada, i.e. the desert, and then on the other side of the desert are the rocky mountains. Even the inland parts of SoCal are mountainous and dry. This picture tells the story:
A viable regional mass transit network requires a threshold amount of population density that you're never going to get in a region that looks like that outside of localized pockets, and localized pockets means everybody in them still needs a car since it's the primary way to get in or out.
It also has a propensity for sprawl even without zoning restrictions that require it because a desert near a major population center is going to get filled with single family homes even if taller buildings are allowed simply because there is available empty space.
Notice that the number of parking spaces per capita is also high in San Francisco and Seattle, but not in New York.
What the restrictive zoning does is make that even worse, and make housing unaffordable as soon as you run short of empty lots.
That’s the general idea on how Prop 13 works - what you miss is that it can simultaneously make any new development attractive relative to a frozen parcel and make dense residential development less attractive than competing land uses. Why build apartments when you can build a hotel? That’s if the incumbent wants to sell in the first place, which they historically don’t since there’s little to no economic pressure to do so.
Your parking spaces point is somewhat circular. Southern California is not uniformly dense. The major reason for why there’s so much parking is because it was mandated by the government.
I think we’re on the same path to reasoning though: status quo zoning in expensive US metropolitan areas generally sucks, Reagan-era housing tax policy also sucks.
> Why build apartments when you can build a hotel?
That one's easy. Because if you stop inhibiting people from building things when there is more demand than supply for hotels then they build some hotels. Then some local demand for hotels gets satisfied, at which point building more hotels becomes less profitable than building more apartments and so they build apartments.
> That’s if the incumbent wants to sell in the first place, which they historically don’t since there’s little to no economic pressure to do so.
"You're receiving $1M worth of net present value in rent but they're offering you $2M" is a pretty good economic incentive. Meanwhile they want to do that because once they add more units the additional units would generate $10M worth of net present value to whoever they sell them to, and therefore they can sell them for $10M.
> Reagan-era housing tax policy also sucks.
It seems to be a common mistake that people think raising property taxes would lower rents. The theory is supposed to be that less productive (i.e. low density) use would be disfavored, but that's pretty suspicious considering that higher density use is already dramatically more profitable per unit land wherever it isn't prohibited by zoning.
On top of that, whenever the property tax applies to buildings and not just land (as is the case approximately everywhere in the US), it does completely the opposite because then constructing new units increases the amount of property tax paid and then rents need to increase by that amount before new construction is profitable.
Prop 13 is uniquely ridiculous not because it caps property taxes for existing owners but because the exceptions to it (i.e. the tax resets to a higher rate on sales and new construction) create perverse incentives. But if you replaced it with something like "max property tax is $1/sq ft/year" it would create a pretty strong incentive to build housing, by reducing the property tax component of rent required to justify doing so.
This seems like an entirely predictable outcome of data center builders and operators behaving like anti-social neighbors with noise just because no law prevented them from doing so.
Building a data center near homes and running your incredibly loud gas turbine generators 24/7 for months, destroying the property values and quality of life for the people living there, is a giant middle finger to the community they've moved into. I'm not saying the backlash banning them entirely rather than adding some additional regulations is entirely logical, but it's very understandable. And should have been eminently predictable.
> This seems like an entirely predictable outcome of data center builders and operators behaving like anti-social neighbors with noise just because no law prevented them from doing so.
The vast, overwhelming majority of data centers source power from the grid rather than continuously operating gas turbines, and the obvious response to the ones that do is to have a moratorium on gas turbines rather than data centers.
Data centers would be a benefit if my power bill and property taxes didn’t jump to subsidize the tax breaks my city gave them. I don’t want to pay for infrastructure multi-billion dollar companies should be funding.
Can you name any place where property tax rates on other residents were increased as a result of a tax break to a data center compared to the baseline of the data center not being there?
NIMBY is largely used for housing development discussions and to contort it in support of developing data centers is incredibly intellectually dishonest.
If we're going to rebrand things to blur lines, I propose rebranding data centers to pedophile content hubs.
The rest of your argument is full of similar cherry-picking for dishonest discourse.
>NIMBY is largely used for housing development discussions and to contort it in support of developing data centers is incredibly intellectually dishonest.
No, it's frequently used in infrastructure discussions as well, eg. high speed rail or power transmission lines.
That pointless infrastructure that has ‘inconvenienced’ landowners is also destroying a national park, so I guess we are all being inconvenienced. And funneled billions of tax dollars to private contractors.
Opposing new housing construction is bad in basically the same way that opposing data center construction is bad; and I'm happy to use the same word to describe both groups of people.
There is a much stronger argument for why housing should be built near existing housing than for why data centers need to be built near existing housing. Housing NIMBYs are being unreasonable because if they got their way the only new housing would be built in the middle of nowhere, which is clearly unworkable. Data center NIMBYs are arguing for much the same thing, but focusing on building data centers a good distance away from housing seems pretty reasonable.
> Data center NIMBYs are arguing for much the same thing, but focusing on building data centers a good distance away from housing seems pretty reasonable.
The trouble here is that the two things are related. It's easy to build a data center 50 miles outside the city. But because we don't allow higher density housing to be built, they instead have to build low density housing all the way out into what should have been rural land. Then you go outside the city to try to build something and instead of the neighbors being forests or fields of grain it's exurbs.
Why is opposing data center construction bad? People need housing, if there isn't enough affordable housing around, then opposing it can be bad. But also people who live there should have a say in what gets built near them, since they have to live with it.
The world and its people would continue to thrive if AI data centers magically vanished tomorrow. They are not necessary infrastructure. They're only necessary for wall street right now.
There's no evidence to support that. There's even arguments to be made that Internet and cell phones have ruined segments of society. We need to take a step back and make sure new life changing technologies are properly regulated, and built safely with communities in mind. Instead, we're going all in gung-ho, because we think its some kind of winnable race, while ignoring all the damage (environmental, privacy, tax payer cost, consumer cost, etc etc). AI will never fix those items.
The reality is we're not doing AI safely because our 401ks and the folks with 12 digits in the portfolio wouldn't grow as quickly as they are if it were regulated.
Infrastructure for what? To push AI, crypto and gambling on everyone? How about build this "infrastructure" in the back yards of billionaires? They should have plenty of land.
An extremely large dark black box. A haunted house on steroids. Not so quietly humming 24/7. Polluting the air from the large scale mobile gensets. Sucking in all the water and electric power so that the residential rates are going up and up.
I'm all for the datacenters, the more the better, it is the great race of acceleration of our civilization. Yet, it is pretty clear, and with history being the witness, that the tech-bros and MBAs will build it without any regard for the well-being of impacted communities. Thus we have the game of what communities will [be forced to] take a one for the team.
And this is why space base datacenters will be a thing :)
For a standalone Pi power via USB is fine. As soon as these devices get embedded or you're running more than two in proximity the USB spaghetti gets pretty miserable.
There are so many quirks with the Raspberry Pi that make it a pain to work with, from the lack of mainline Linux support on the Pi 5 to the awkward by default power arrangement to the fact that the architecture is a GPU with a CPU glued onto the side. The Raspberry foundation made SBCs much more ergonomic and kudos to them but working with them still feels like walking through a field of tiny rakes.
reply