There's likely an awful lot of tricky logic in the queries, stored procedures, etc. Dealing with taxes (very location and situation depending), details of any given customer's plan, equipment charges, etc etc for lines going back for decades... this stuff is hard to migrate correctly. It's not necessarily about qps.
The data tier is one of the lowest tiers in the stack (for bigcorps with centralized DBs), and changes bubble up the stack. Change an API and a few upstreams have to change. Change the core DB and _everything_ has to change.
That makes the schema calcify, and the DB becomes the most stable format of the data. I've worked more than one place where "upgrade the core DB between major versions" was a multi-quarter effort.
I don't know specifically, I'm not the person who posted about that database. But it happens all the time, even in 2026 at good tech companies.
It tends to come from a couple of common places...
* efficiency - triggers, stored procedures and constraints can often speed up throughput in terms of transactions per second by reducing the number of round trips from the application to the db. I know above i mentioned its not about qps necessarily, but a telephone billing db probably includes records of calls and txts and so on, which across millions of people adds up.
* Older architecture patterns didn't put a single service in front of a database for that service. Instead theres a single database for the company, or for a function within the company, and many programs and services interacted with that database. In such an architecture, stored procedures and triggers are really useful for some business logic, to ensure consistency between various systems... Need to update tax rules for one state? Just donit in the db instead of the sales pos system at the stores and the billing system and the phone operator sales system and the website and the gateway for 3rd party sellers, etc. there's a certain logic to it too... This data is fundamental to the business and has lots of rules associated with it, those make sense to place together - the rules change independently of the systems accessing those rules, and over the life of the company those rules tied to laws and accounting may last far far longer than any given pos system or website rewrite.
Instead of ranting about nimbyism, why not go to these neighborhoods where people don't want to live next to datacenters, and offer to trade houses with them?
You get the datacenters you want, and they get the place they want without a datacenter. I know you won't, because you only think the consequences of a datacenter next door are just fine when its other people who have to deal with them.
What about the water treatment plant? Or the power station? Or the new subway line?
Why don't you go live in the woods since building anything where anyone doesn't want it built is so evil?
Maybe because you only think of the consequences of building things when it's for anything other than AI?
In many places these data centers make communities money. It's really not as black and white as you are making it. If I'm getting paid out I also am not sure why you think it's a trade nobody would make. Maybe you live in a very nice house.
I support UBI. No one should be poor in the United States. My goal would be telling someone "why don't you trade places with someone in x" in the United States should never feel like a threat. If I were to have to reside next to a data center to make this happen then so be it. Data centers are not the problem. Income inequality is. I think you have me confused with someone else.
What's your plan to solve income inequality? Don't build anything, crash the economy, and push more kids into student debt so they can take up bullshit knowledge work jobs?
If they make the communities enough money the community will be happier to let them be built. Try increasing that amount and offering it up front, maybe.
I would support that. Companies should do that. Sometimes communities disagree, obviously. Communities that really don't want them often succeed. That's why the administration is throwing such a fit.
I don't support that. The administration is the worst advocate for AI because they are so clearly corrupt. If local people pursue their economic self interest at the end of the day that's their right. This whole nimby thing was a diversion from my point about knowledge workers turning on AI. Perhaps I should have just ignored that comment.
If people want to stop a local data center then fine. I think it's bad national politics. But if it's helping these companies make better deals or find better places to go then good. What I object to is the doomerism about it, and the uncritical conflation of self interest with civic virtue or robust public policy.
> Why don't you go live in the woods since building anything where anyone doesn't want it built is so evil?
Wow, you're really triggered by the idea that people should get a say about what happens in their lives.
You'll note I didn't oppose the datacenters - I offered a solution to getting the datacenters built. If you trade houses with the people who don't want a data center in the neighborhood you're getting the datacenter! It's just that you're so upset about someone else having the ability to affect their own lives that you aren't seeing that you have the same power, and can work towards your own perfect world with power plants and datacenters and sewage treatment plants and subway stops just down the street!
I think the problem with that plan is what I pointed out - you want other people to have all the downsides of those things in their neighborhood while you get the benefit. It's pretty sick to think that them not wanting the downsides is a problem, but you not wanting the downsides also means they have a problem. Fortunately there's a solution! Go form a YIMBY community that allows all the projects that have representatives promising benefits, and prove that the NIMBYs are all wrong. I'm 100% certain it will make you way happier.
> What's your plan to solve income inequality?
Well part of it starts by not giving giant companies with lots of money even more money in the form of tax breaks, extra access to government, and the ability to use eminent domain for their profit. All these thing grow income inequality.
You wouldn’t want to live next to a data center because of income inequality? Or you would want to live by one because you think data centers cut checks to townspeople?
Sure? I would be completely fine with a data center next door assuming it's a normal one and not something like xAI's. There are several not too far from me already.
In general, you're right. They are just buildings that sit there emitting hot air.
However these AI ones are being built with all expenses spared. That's why they come with on-site electricity generation that is loud enough to keep the whole town awake at night and pollutes enough to give them all permanent asthma.
If you don't like it there, trade with someone who is in a town with a datacenter and is unhappy that the datacenter is there! Or maybe convince more YIMBY types to move into your city and get the datacenter back.
I understand you're disappointed that people had the ability to choose things in their lives and didn't choose the thing you wanted. You can still do plenty to get the datacenters right across the street from you - sometimes in this world you have to work for the outcomes you want. IDK if it's fair or not, but it is how it seems to work out.
Nobody is building datacenters directly across a house. This kind of ridiculous exaggeration doesn't help. And I'd happily live in a town or place near one.
You linked to a data center that was a bitcoin mining operation and is explicitly having noise abatement being applied before being brought back online as an AI one?
That doesn’t support any argument against new builds nor does it support the notion that they are inherently noisy.
Also, in reply to your further up NIMBY nonsense suggesting I move.
I don’t think you understand how people operate. I like my small city and am upset idiots spread fear and harmed its prospects that we will pay for with continued anemic local employment, poorly budgeted infrastructure, etc.
I want to see good things here. I’m not just going to give up on it and go to some city that has its shit together on the business side.
Datacenters have been around for decades and the Dulles Tech Corridor, the ones around Portland, the SIX in Seattle, etc all have shown they can operate just fine.
> I understand you're disappointed that people had the ability to choose things in their lives and didn't choose the thing you wanted
No, I’m disappointed they did it based on ignorance and propaganda. There was no informed decision making going on. This is no different than people who voted against gay marriage because they thought their god would seek retribution through calamities.
> You linked to a data center that was a bitcoin mining operation
This is a good point. Was it a data center across the street from somebody’s house when it was mining Bitcoin? Yes. Will it still be a data center across the street from somebody’s house when it reopens as an ai data center? Also yes
Research pilot programs are always much more expensive than the comercialized version will be. The money is used to work out kinks, do a bunch of engineering, discover and account for unexpected issues, and so on. A lot of the work done on there is reusable to make the scaled version cheaper. Further, once the design is settled, economies of scale factor in and you get a much cheaper unit price.
According to my searches, steel production is something like 6-8 MWh/ton to extract, refine, and shape. A solar panel lasts 25 years at up to 80% effiency. This thing produces 1.7MW at peak. so call it 3 MWh/day average to account for a lot of adverse conditions over the lifetime. Thats 27000 MWh... so unless we are looking at more than 3000 tons of steel (we aren't) the steel production is accounted for. A panel takes something like 1.9 MWh equivalent in energy.... and again we aren't looking at 13K solar panels in this pilot.
So yes, in fact this does generate far more energy than goes into its production.
I seriously doubt that you can chain even 1/4 mile of panels together without destroying things. Thats already puting thousands of volts and hundreds or thousands of amps through the silicon. If you are making it all parallel you still need wiring between panels that can handle that. The cabling doesn't go away, it just moves.
Wouldn't there still be supports in a feild of solar panels? Are you sure that those supports + the supports for the shade material are going to be less material than the supports for this?
Most shade material wears out pretty quickly. Will their replacement result in more expense, more waste, etc than just putting the solar panels?
Shade material is generally pretty heavy, is it really going to need significantly less robust support? Weight aside, how much of the load those supports are rated for is due to the actual weight of the panels, and how much is for forces from things like wind?
"Thats already puting thousands of volts and hundreds or thousands of amps through the silicon."
In a typical string of solar panels design, you'll get tons of volts but not a lot in amps - current cell maximums top out at ~11A and the connective MC4 wiring can't handle too much more current than that, so what you end up with is like a 1,000V 10A string on one MPPT connection into the inverter.
A solar panel is 4 feet or so on the long side. Theres 250+ of them in a string 1/4 mile long.... at 48 V/panel, you get to 12KV. even if it's topped out at 10A thats still 120KW... you need a hefty cable to carry that panel to panel. Which is the core of the point I was making.
Amps determine how hefty a cable needs to be, not volts. Volts mostly determine how thick the insulation needs to be.
14 gauge wire is basically all you need to carry 10A safely for an extended period of time regardless the voltage. It doesn't matter that you are carrying 120KW.
The proof of this is in EV charge cables. Those bad boys can carry up to 350kW. Yet the cables are often thinner than you might expect. How do they do this? It's by using high voltages (around 900V) which cuts back the amps to around 300->400.
Tesla's chargers peak (or used to) around 600V which has required them to have much beefier cables to handle the high current.
I get that. I'm pretty skeptical that 120 KW going through the last panel doesn't cause damage, and in the "multiple miles chained together" (to quote op) many MW going through the panel at the end of the chain doesn't run into some sort of issue in the real world. Even small percentage losses turn into a lot of heat or other issues at that scale.
If it doesn't why not just make the whole 100 mile stretch of canals discussed a single very long daisy chain of panels, and still address the concern of the OP in terms of extra material for conductors?
And so ultimately, despite my details being wrong about currents, the notion of stringing miles of panels together is probably not a good basis for rejecting the panels over the canals.
In reality, you shouldn't string a lot of panels together for other reasons. An obstructed panel doesn't just not produce energy. It acts as a resistor, _wasting_ energy that flows through it.
You would have short strings of panels and small string inverters built every ~100 meters along the channel. Inverters will also connect to the high-voltage bus cable, likely buried in a trench along the channel. And most channels in California are not gravity-fed, so they already need pump stations along the way and have power distribution systems for them.
"Theres 250+ of them in a string 1/4 mile long.... at 48 V/panel, you get to 12KV. "
And they're never installed in config chains that high voltage - cell spacing between cells is typically very low and only capable of withstanding ~5kV before internal arcing occurs.
Also, your typical 72-cell solar panel is only 32V.
The obvious answer is a cheap inverter ever few meters - this is typical practice with solar panels - and then a cheap transformer scattered around to get the voltage to whatever the line is. This is the common cheap practice for solar installs everywhere.
Microinverters where there is an inverter on every panel are also common - they are more expensive, but reduce some labor.
Either way the inverter is connected to a network and then get remote monitoring of each section and in turn can plan maintenance as needed.
I wouldn't. I was responding to the GP comment talking about how utterly awful and wasteful this design is because you need some extra wire instead of just making "miles long chains" of solar panels. My 1/4 mile example was just to point out that a tiny fraction of those miles is itself insane.
To analogize coding to sculpting - you're in the "toddler with playdough" phase. At some point you might want to make something other people will find pleasing and maybe even worth putting on display, the rules of composition, the details of materials, and the techniques for not wasting expensive materials suddenly become important.
If you're just writing code to fuck around or automate a small part of your life, whatever. But if you're making a big system or wanting other people to use your product, these things about how to make good software become more relevant.
Or we could look at the utterly stupid things, like maybe stop having a failed war in Iran, and just not pay for those? Thats a very big chunk of budget. As a bonus it would bring gas prices down, reduce future spending in VA and other combat veteran costs, and just possible help make the case for reducing the size and expense of our military (while still maintaining the 2 largest air forces in the world).
What makes you think the poster you responded to agrees with the Iran war expense? He simply commented on how difficult it is to cut anything from the Federal budget. He's right. Remember when all the lunatics were out keying Teslas a few short months ago?
No, its not a strawman. Cutting tiny things from the budget to save money when there are big obvious wastes right out in front is fucking stupid. Complaining about how hard it is to cut tiny things from the budget when there are giant obvious wastes out in front is equally as stupid.
If you want to seriously affect the budget, go for the things that burn large piles of money first.
What is "implementing".... I'd define it as "figuring out the workflow, and storing it in a way that allows reuse". This could be a program written in assembly, or rust, or even python. Or it could be a shell script that calls curl. Or it could just be a set of tokens in the current session. Outside of the computer it could even be a set of processes people do, or a mechanical device.
If it's just a set of tokens in the current session, well then next session it has to figure out the workflow, and then store it in a way to use in the next session.
Nonsense, curl is a tool for executing a single http request.
Many apis require several requests to get things done. (one to auth, one or more to fecth resource ids, one or more to modify resources, etc).
That would be a series of curl calls with logic applied to the output of each call to curl. An api client just does those things in a single function call. The steps are the same, but in one case the AI has to figure out each curl call and implement the logic, rather than just call the function.
By your logic 'ls' is a file manager, 'grep' is a search engine, and 'echo $X >> /proc/sys/$Y' is a settings manager.
It makes sense though. It's the same logic that allows you to say promting an AI with "do a simple thing for me" makes you a programmer, and prompting an AI with "what is an api" allows makes you knowledgable about computers.
Instead of doing that boring thing the unthinking do - I challenge you to ask why 15M seems like a lot.
Is it a lot in comparrison to the costs to the taxpayers for all the emergency crews responding to animal caused accidents over the lifetime of the crossing, not to mention dealing with carcasses of animals, etc?
How much of any single person's tax money went into it? How long until those few cents get paid off in terms of saved time and gas by not dealing with the traffic backup from those accidents? How much do those few cents amortize to your savings in reduced insurance since thats less potential accident you have to bet against... how much savings does that reduction confer on your expected direct costs in cas you had been in a the accident?
Does 15M still seem like a lot?
Maybe you think it could be built cheaper - how? Show your work.
All these things sure are an interesting conversation to have. Much more interesting than "hur dur, 15M is a big number" anyway.
That seems well within reason, an article said about 1 deer collision prevented per day in the spring and fall high season. Having driven through the area and seen dashcam footage of collisions and near collisions, that $7k makes sense to me.
I've known several people that have hit deer on the highways here, and I'm actually surprised if anything that the number is that low.
So my gut feeling is that it actually has slightly better returns than expected from those numbers.
reply