Dang, the prices are crazy. $10k for the lowest tier? I expect federal data to be free. Most federal data related to stuff like regulation (e.g. FCC spectrum use coverage maps), maps for federally-managed lands, etc. is free. And, it should be free. It's a core part of accountability and transparency. Historically there was a copying cost (having a human print physical copies on paper or CD/DVD/tape), which is understandable, but now that the incremental cost of each new copy approaches $0, that's what it should cost users of the data.
The USPS is getting really difficult as a vendor for fintechs to work with. They recently changed the pricing around address verification as well (it used to be free).
I really don't understand the justification for withholding operational data like this. We are trying to make the job of the USPS easier with these tools. It's a lot cheaper and faster for everyone involved if I can query the database while the customer is still standing in front of me.
USPS is in a strange boat where a lot of people insist that it must be profitable, much like UPS or Fedex, while failing to realize the value it provides to the rest of the economy by being cheap. Last time I had to physically mail a check I was shocked that it costs 10-20x as much to use UPS or Fedex than it did for the USPS.
>USPS is in a strange boat where a lot of people insist that it must be profitable, much like UPS or Fedex
This isn't really accurate. USPS had a break even (not profitability) legal mandate since 1970-2006. They would adjust prices each year to hit break even. And that's what happened - they roughly broke even each year.
After a 2006 with broad bipartisan support, the UPS was able to make a profit (not just break even) but a price cap was added (to be reviewed every 10 years). They couldn't raise costs beyond inflation. A stamp to send to anywhere in the USA - anywhere (Alaska, Guam, etc.) - currently costs 82 cent. The postal union is against this, usps management is against this, and business mailers are for it. Dont you love all the ads USPS delivers you?
They have not been break even since 2007. They currently lose about $9B/year with $2.7BM being controllable and the rest largely being pensions, retiree health plans, and workers comp charges. Their core business (mail) has been declining for 20 years and no end is in sight. In 2020 they loosened the cap a bit (allowing USPS to raise prices). Business mailers tried to sue them but lost.
So this attitude that they should just lose a bunch of money and be backstopped by the taxpayer, when mail is less critical than ever, is an recent idea and is aligned with business mailer lobbying interests. It seems obvious to me that they should just go back to the break even mandate.
Yes, thats exactly right. The ability to profit was granted and they were kneecapped by the price cap. They haven't profited since the legislation and immediately started incurring large losses. That's all in my comment already.
By no means was I implying that the ability to profit means they did profit. Im just stating what the law was.
In 2006 a law was passed requiring the USPS to fund pensions far in advance, a requirement no other business had, which instantly put them deep in the red. This wasn't fixed until 2022. But it was just this phony accounting that messed them up, more than the other factors you mention.
>In 2006 a law was passed requiring the USPS to fund pensions far in advance, a requirement no other business had, which instantly put them deep in the red. This wasn't fixed until 2022.
No because they are still deep in the red, about $9B a year.
That’s $2B less than the first six days of the Iran war ($11B). We could just cut a check each year from the general fund. It’s a paperwork issue. The fact that we can spend $100B+ on an illegitimate conflict on a whim while costs like this are bikeshedded over boggles my mind.
US Military costs ~$1T/year and the Pentagon hasn’t passed an audit in the last eight years, comparatively, but not a lot of similar complaints of that waste versus pensions for citizens operating and serving a national public service agency.
"Contrary to statements made by some employee groups and other stakeholders, PAEA did not require USPS to prefund 75 years of retiree health benefits" https://www.gao.gov/assets/gao-13-112.pdf
Also the 2022 Postal Service Reform Act eliminated the prefunding and ~$57 billion of overdue payments, but USPS finances were still dogshit afterwards: USPS revenue was $80 billion and expenses were $89 billion in FY2024
I don’t think of their finances as being dogshit because there aren’t revenue positive. It’s a fraction of a sliver of the federal budget to run a constitutionally mandated service.
The (non-navy) military has no constitutional mandate at all, it’s an industrial scale cash incineration device which the postal service can’t hold a candle to, and we don’t blink at it wasting 10x or more of the postal service’s opex.
> So this attitude that they should just lose a bunch of money and be backstopped by the taxpayer, when mail is less critical than ever, is an entirely new idea and is aligned with business mailer lobbying interests.
Not when combined with the idea that spam mail senders should be legislated out of legality.
The problem is that there are categories of spammers willing to pay more than what the poorest regular people can afford. Much more reasonable to make junk mail illegal just like other countries already have.
>while failing to realize the value it provides to the rest of the economy by being cheap
but that "value of being cheap" is provided to the people using the mail (or other delivery services) for that transaction; it does not benefit people who are not involved in the transaction but have to pay the subsidy. The economy operates most efficiently when things cost what they should or as close as possible.
what Microeconomics is about is calculating what is cost efficient and what is not; free mail services are not; less free mail services are not; rationally priced mail services are.
This is how it works in Denmark. A letter costs 23DKK ($3.44) within this small country. A fast letter is 36DKK ($5.39).
I think I've received two letters in the last year. I haven't received any individual junk mail letters.
The letter volume last year was 90M for 6M people, so about 15 each. I assume elderly people and others who have opted out of receiving government and business mail digitally drive that number up.
(Someone still delivers a bundle of leaflets from local businesses all together every month or so, and a local newspaper, but I don't think it's the postal service. Every home gets the same bundle.)
I personally get a massive benefit from USPS being cheap, despite the fact that I send no mail. It provides a massive low-cost physical delivery network across the country, with service to every address. This is a benefit to me because it makes business and personal communication simple and reliable. If I need it personally, it's there. If I don't, I still rely on thousands of parties who do.
Your position is like claiming the interstate system is useless because you don't have a car, ignoring how all those goods your life relies on actually get to you.
Parent poster is exactly the kind of person who argues that using tax dollars for school is wrong because they, personally, do not have kids. Because they can't realize that they indirectly benefit from it.
John Green: "So let me explain why I like to pay taxes for schools, even though I don't personally have a kid in school: It's because I don't like living in a country with a bunch of stupid people."
Congress has mandated that USPS has to be self-funding. If the leadership at USPS fails to raise revenue, they are forced to cut services. You could write a letter to your senator.
If you want them to exist you they can't just ignore all finances. USPS had a break even mandate from 1970-2006. They projected revenue and priced services to break even. That was scrapped in 2006 and they had a price cap put on services which now puts everything into jeopardy. They havent been break even since.
Their core business has been declining for the past 20 years and business mailer lobbyists work very hard to keep the current price caps. This idea that we should let the finances spin out of control and raise taxes is a recent idea.
No to remove the cap on pricing. USPS was break even from 1970-2006 until the 2006 law that forbid them from raising prices. The business mailers love it though.
Companies are already using the roads, police, fire services (where applicable) and many other government services without paying directly for them. But the taxes! you'll scream - yes exactly!
Standard communication (including addressing) is such a basic, critical function of government that it only makes sense to do it at scale and just fund it with taxes.
>Standard communication (including addressing) is such a basic, critical function of government that it only makes sense to do it at scale and just fund it with taxes.
Your approach leads to exploding federal budgets, which are already the bane of the US, Canada, and the EU. When will you learn? Everything is not the exception, everything is the problem.
Speaking for a patchwork of samples that have fascinated me in the U.S.—New developments tend to incur impact fees to offset their immediate effects on the roads, sewers, and sidewalks. Police charge a fee to connect your alarm system to their dispatchers, an hourly wage for an off-duty cop to stand guard at your commercial premise, and a fine if you waste their time with false alarms… in some cases they attempt, at least notionally, to recoup the cost of the investigation from criminals upon conviction (I hear the collections rates are unimpressive…). Heck, some places bill inmates for each night in jail! [0]
Fire services tend to send you a bill for a false alarm, a fee for an inspection and, if you did it on purpose, a criminal charge plus the full cost of the response. And lord forbid you accept EMS services…
All are tax-funded, but all gesture at apportioning costs in relation to use. If they send this much junk mail while it’s quarters-a-pop, how much would they send if it were free…
Should email be also run as a public service, not a business?
Maybe some amount of messaging connectivity, for a narrow set of cases like court summons and other official messages, could be run as a public service at the taxpayers' expense. But likely the bulk of it should be run as a paid service, either regular businesses, or/and non-profits that still are expected to break even.
Not just email but internet as a whole should be run like a public utility, electricity as well. It's US centric to think everything needs to be private.
Perhaps we are entering an era where it should be.
Comparing today to 40 years ago: as far as I can tell, letter mail is virtually always used for the purpose of transacting business these days. Forty years ago, it was used for business and personal purposes. If I wanted to communicate with family on the other side of the country, it was more likely than not done by mail. Even 25 years ago, a long distance call took some thought because it was charged by the minute. These days, at least in Canada, all but the cheapest of mobile plans comes with nation wide long distance and we have the Internet. Both services are available for less than basic phone service back then.
Parcels are a different story. I can't guess how people's personal habits have changed, but the business front is very different as far as I can tell. Mail order was the exception in urban areas back then. It would have been different in rural areas, but in my explorations of the country I discovered that some towns were places where people lived while other towns were regional hubs for businesses. There would have been some mail order and a lot of doing without, but I suspect it has seen a shift where proportionally more parcels are commercial in nature. On top of that, the standards for delivery have changed. When I was a kid, it could take weeks to receive something via the postal service (from across the country). Now shipping times are more aligned with couriers.
I realize there are economic benefits to subsidized mail, but I'm not sure that subsidizing businesses should be a priority.
Well, serving it costs money for USPS. And, come on... you're not trying just to make the job of the USPS easier here. This adds value to your product, and you're a for-profit entity.
I don't know why the taxpayer should subsidize fintechs.
Verified addresses (CASS & friends) receive automation discounts. These save taxpayer money by reducing delivery exceptions.
We can help the USPS by using a very cheap API to ensure that data we are entering into bank core systems is accurate.
Yes, the entity is for-profit. The bank is for-profit. Everyone is for-profit. Who gives a shit. This isn't a zero sum game. Everyone can win, including the USPS and the taxpayer.
I give a shit; I want costs and incentives aligned as closely as possible, and that means not shifting the cost of things to the taxpayer in a way that ensures there will both never be competition, and the true costs are not borne by the people who actually incur them.
The people arguing against you are not going to see it your way.
There is a ideology of some people that think that taxes are a magical way of funding anything they feel is a common good. Most of the time these are well intended thoughts of public good. However they NEVER take into the account the unintended consequences of their ideas. It stops with tax more = more public good.
If this was about serving costs they wouldn't have terms like this in their license:
>LICENSEE shall use the Database and Data for the Permitted Use only
and shall not disclose, release, distribute, or deliver the Database or Data, any portion of the
Data or Database, or any other information derived from the Data or Database to any third party
without LICENSOR's prior written consent.
> And, come on... you're not trying just to make the job of the USPS easier here.
Most of this address data is made available in order to enable pre-sorting of mail, which makes the job of the USPS easier. Presorted mail can also have a lower mailing cost, but I wouldn't be surprised if the USPS doesn't pass on all of the savings.
> I don't know why the taxpayer should subsidize fintechs.
So then GPS and the Internet can go away then.
One of the development methods for new technologies in the United States for a long time has been the federal government creating projects that fund long term research, often that create multi-use technologies.
Only one person has to pay for it, then they can share it with the rest of us for free. It's not like the data is trade secret or protected by copyright or patent. It's public data.
> LICENSEE shall use the Database and Data for the Permitted Use only
and shall not disclose, release, distribute, or deliver the Database or Data, any portion of the
Data or Database, or any other information derived from the Data or Database to any third party
without LICENSOR's prior written consent. For the avoidance of doubt, LICENSEE may
incorporate the Data into the products or services listed in Exhibit B (ZCB Order Form), as
allowed by Exhibit D (ZIP Code Boundaries Tiers of Access and Usage), but no Data either in
whole or in part may be transferred or sold by LICENSEE.
Sooo... You can't compel the government to create a document or data as part of FOIA, but...
You can request any already-existing dataset (which this now is). And they're required to charge only reasonable duplication costs and respond in a timely manner... And such FOIA response would not come with the same restrictive license.
Sharing it with everyone is probably going to get you banned from getting quarterly updates, as there's strings attached. Though I do wonder just how they keep track. Are they fingerprinting the data somehow? Does that make the dataset less useful?
Maps have a lengthy history of including "Trap streets" to identify copies.
Changing the value of the least-significant decimal place for a couple of points in a polygon is unlikely to significantly change the shape or reduce the accuracy so significantly that it reduces the value of the data.
>Each Party acknowledges and agrees that a breach or threatened breach
by LICENSEE of any of its obligations under this Agreement would cause the LICENSOR
irreparable harm for which monetary damages would not be an adequate remedy and agrees
that, in the event of such breach or threatened breach, the LICENSOR will be entitled to
equitable relief, including a restraining order, an injunction, specific performance, and any
other relief that may be available from any court, without any requirement to post a bond
or other security, or to prove actual damages or that monetary damages are not an adequate
remedy. Such remedies are not exclusive and are in addition to all other remedies that may
be available at law, in equity, or otherwise
I legit tried to get one of those new. They're practically indestructible and have shit tons of storage. They would be awesome get arounds on the farm. If they were AWD or 4wd, I would give my left leg for it.
Those old postal vehicles were made from the GM parts bin with a custom shell built to spec, though most of them were from the S-10 platform. They are not indestructible, the USPS started scrambling to buy COTS vehicles while the NGDV was in development.
The S-10 itself is a really good farm truck though, so I'd keep your eyes on the market for one. Or maybe the equivalent Ford one.
> If they were AWD or 4wd, I would give my left leg for it.
I assume you're talking about the LLV?
I had thought the post office bought a small number of 4x4 LLVs, but I don't see anything about it on wikipedia, so probably not. That said, maybe you could bodge something together from S-10 4x4 parts? (wikipedia says the rear axle is from the 4x4 s-10 blazer and the front axle is from the 2x4 blazer ... it might not be easy, but it's probably not impossible to grab a 4x4 front axle and stuff it in... )
I've read that the post office doesn't sell their old ones, but they weren't the only buyer and some of the other buyers do sell them.
Wikipedia does say the NGDV comes in all wheel drive. Although yeah, probably hard to buy a new one.
I have heard it claimed that they are not DOT compliant which prevents them from being sold to the public, so USPS crushes them all at the end of their life.
That said, once in a great while a non-USPS LLV shows up on GovDeals. Last one I saw was local to me a few years ago and it sold for over 10 grand.
There's still the overhead cost of collecting this data and the operational cost of delivering it. Many government functions charge money because it's unsustainable if they don't.
For example in Seattle people can request body cam footage arbitrarily. But the faces and identifying info of witnesses and victims need to be redacted which takes time and money. Someone started bulk requesting almost all body cam footage, and that placed a big cost on the police department. Now they charge $17 per body cam request.
But in this case, collecting the data is already a part of USPS core business so there isn't much overhead. And there's no individual processing needed per request like in your body cam example.
This is also a congress-made problem as they privatized the USPS and created this pointless exercise of making them turn a profit, while being regulated like no other company (because that's us admitting we need mail).
The corporate agenda of the past several decades has been to destroy the Constitutional government (which is the only thing bound to our narrowly-construed Constitutional rights) in favor of ceding authority to unaccountable corporations.
I worked somewhere that purchased ZIP code polygons from somewhere at one point, pretty sure it was way way cheaper than this. Idk if there was a difference in quality.
No, it's a more fundamental problem: Zip codes are not "areas" and have no boundary.
Instead, what they are, are collections of mail delivery points. The "full" 5+4 zip code represents one mail delivery route that a postman follows to deliver mail. And the "fuller" 5+4+2 zip code has the last two digits of the street number.
So couple issues with trying to make polygons from zip codes:
1. There is no guarantee that a given zip code has a finite area. There are several zip codes that exist at a single street address. So... is that a point? Is it the footprint of the building where the mail is delivered? Maybe a hexagon with some minimal side length? Examples: Several IRS return addresses, and some large hospitals.
2. There is no guarantee that a given zip code has a contiguous polygon representation. It depends on the algorithm you use to generate polygons from address points.
3. There is no guarantee that a given zip code has a fixed geospatial representation. Example: the armed forces zip codes for delivering mail to overseas personnel on deployment. The AFPO zips represent "wherever the troops happen to be today". How do you represent that as a polygon?
This is why the USPO has long resisted publishing polygons. It doesn't make sense at the conceptual level, so there are a bunch of weird edge cases you wouldn't expect if you naively treat them as polygons.
ALSO: The census does publish zip polygons. Some of those have zero people living in them (like the hospitals and IRS locations), so you can get weird div-by-0 effects if you aren't careful.
Oh, I meant the PostGIS "polygons" which can have holes in them. So #1 I believe they just had a small polygon, #2 was no problem, #3 wasn't supported but whatever. The polygons did overlap too.
1. A given land parcel may not have mail delivery (think: rural land, even with people living on it), so... what is its zip code? Does it even have a zip code? Remember, a zip code is defined as a set of mail delivery points.
2. A given land parcel may have multiple mail delivery points, each of which may have a different zip code (think: a large mall or commercial building with multiple tenants, where it may be more convenient for the USPS to deliver from two or more mail routes). Does that land parcel have more than one zip code or do you arbitrarily pick one? If it has more than one, does the whole parcel have both codes (so, overlapping polygons), or do you try to subdivide it somehow?
[Users of zip code polygons often want to preserve the capability to look up the zip code from a given lat/lng point. So they will yell at you if you choose to have overlapping polygons or two zip codes for one polygon.]
3. What about AFPO addresses that represent the current location of a group of people who move around all the time? An aircraft carrier out in the Pacific is not near any land parcels.
I was in case #2 for a while, and tbh it should be treated as the building's fault for being this way, cause ambiguous ZIP codes aren't really supported. The IRS refused to send my refund, mail kept getting lost, and various things involving credit kept flagging me for fraud. Also it looked like a PO box to some systems but wasn't. My solution was to move, or I could've gotten an actual PO box or other mailing address.
So 1 means the polygons don't cover the entirety of the US which is fine, 2 means some polygons overlap which is fine too, 3 means some zipcodes are excluded from scheme, as one might not be legally allowed to know where a particular unit is.
"Many postcodes for particular lat/lon" should be part of the polygon delivery schema, i.e. you get a list all the time.
Ok, but you have to think about how people actualy use the polygons you create. You might be assuming these are minor edge cases or that I have enumerated all of the relevant edge cases (I have not) and that we can sweep this in a corner and forget about it.
> 2 means some polygons overlap which is fine too
For some use-cases, sure, but not all. One common use-case for zip code polygons is "I have a lat/lng, just give me the zip code, damnit!" Your users might expect that a lat/lng point can have no zip code, but most won't expect that it can have multiple (because no address has multiple zip codes).
At a minimum that leaves you scrambling to explain "well, but it only happens in a few edge cases where a land parcel has multiple mail delivery locations from different postal carriers, and it's super-rare, pinky swear".
> 3 means some zipcodes are excluded from scheme, as one might not be legally allowed to know where a particular unit is.
Another very common use-case for zip code polygons is "I have a zip code, just tell me where it is, damnit!" Maybe they want to estimate shipping distance without hitting the shipper's API.
Part of the problem here is that the AFPO zips do have a physical location for this purpose, which is the facility where mail and parcels are delivered from the USPS to the AFPO. And your users will probably expect you to know that's what they mean. So when you try to explain that "actually an aircraft carrier is constantly moving and we're not allowed to know where it is", they will think you are being pedantic and condescending, and look for someone else to sell them polygons.
But let's say the user actually wants to forecast sales of the new ThingyWidget 3000 by estimating the average weather by customer zip code (by getting the polygon for each zip and the doing a spatial join to weather data). When they do that for the AFPO drop-off point, they're going to get a completely wrong estimate of the weather because the people receiving mail at the AFPO zip are not physically present at that place. So when you explain that "actually that is the physical location of the AFPO facility" they will also think you are being pedantic and condescending, and look for someone else to sell them polygons.
Notice how you're damned if you do and damned if you don't?
Any choice you make for polygonalizing zip codes will be wrong for some use case, and it will be spectacularly wrong for certain edge cases in very intuitive ways that certain users deeply care about and will blame you for.
Which brings me back to the point: You can make a set of polygons for zips which is "good enough" for many use-cases, but ultimately you cannot represent zips as polygons in a way that is suitable for all use cases because, once again, zip codes are not polygons!
Well the alternative to "here's a service that converts a zipcode to a polygon, and coordinates to a list of polygons, and here's the fine print of limitations and how they are communicated" is declaring that postcodes are arcane entities and their connection with the mortal plane is only an illusion, and no such service is possible.
If I was developing something that required dealing with US GIS I'd very much prefer a service that does the above, and handle the edge cases, than have no such service at all.
Like most large data sets, ZIP codes are complicated. They are defined by delivery routes, so sometimes they overlap when that makes sense for delivery purposes. Some post offices with a lot of P.O. boxes can have their own ZIP code and that's typically represented as a point rather than a polygon. Large naval vessels have their own ZIP+4, and the base 5-digit ZIP code is a single point that is the land-based processing facility that the vessel is associated with. The same for some other types of military facilities and government buildings (e.g., the Pentagon and White House). I'm sure there are more special cases. It would be interesting to know how accurately the official data being offered here handles all this.
I came across all this when building maps for sales territories defined by ZIP codes. I needed to have a simple, clean partitioning of the US for this use case, and ZIP codes are not that. So I took all ZIP codes as points and generated a Voronoi diagram to yield clean, non-overlapping polygons that covered every point in the country. That was plenty good for the purposes of doing visualizations of the US split into tens of territories.
It took a bit of scrolling to finally find someone who understands that in most countries postalcodes can be rendered as areas — if you have the points of each address, using them to generate a Voronoi diagram.
USPS is in really rough shape, budget-wise, to the degree that they needed an emergency cash infusion earlier this year (I highly recommend reading up on it, it's a bad situation). Sadly, it's not that surprising to see that they're charging commercial rates for this service.
That’s not an accident. Congress made it a requirement that the USPS pre fund their pension plans, decades in advance. The only government institute required to do such a thing.
It is great for governments to run responsible budgets, but this was such a unique constraint, it seems designed to strangle the USPS.
"Contrary to statements made by some employee groups and other stakeholders, PAEA did not require USPS to prefund 75 years of retiree health benefits" https://www.gao.gov/assets/gao-13-112.pdf
Also the 2022 Postal Service Reform Act eliminated the prefunding and ~$57 billion of overdue payments, but USPS finances were still dogshit afterwards: USPS revenue was $80 billion and expenses were $89 billion in FY2024
Nah, they're not trying to strangle the USPS. They're trying to fatten it up for privatization.
If you fully fund the pensions before privatization, you can simultaneously claim that privatization is necessary (look! they're running out of money!) and radically reduce costs once you do privatize (all the pensions are pre-paid, so you can stop contributing to the fund and extract profits for a few decades before you have problems).
This zip code subscription is part of the same plan: Get people used to paying for data from the USPS so it has an additional income stream after privatization.
PAEA 2007 required funding of accrued liabilities for the USPS, same as PPA 2006. However, other governments in the US are not required to fund any liabilities for post employment benefits, at all, as they are excluded from both ERISA 1974 and PPA 2006.
Page 55 discusses setting money aside for retiree healthcare benefits:
>Not later than June 30, 2007, and by June 30 of each
succeeding year, the Office shall compute the net present value of the future payments required under section 8906(g)(2)(A) and attributable to the service of Postal Service employees during the most recently ended fiscal year.
"Net Present Value" is accrued, i.e. benefits owed to current and past employees.
It makes sense to require an organization to set aside money it promised to pay someone...otherwise organizations would just go around paying people today with future taxpayers' money (which is what governments literally do).
Page 53 discusses the same for defined benefit pensions. All of the present values are of accrued benefits, not projected benefits.
However, I do suspect this was a clever way to handicap the USPS specifically, since it did not apply to any other governments such as state/county/city.
> The Trump Task Force acknowledged that without the costs imposed by the Post-Retirement Health Care Mandate, USPS would today be profitable on an operating basis.
I don’t understand why recent Congresses haven’t repealed that requirement. It seems well known to be a bad thing that hamstrings USPS. There have been multiple congresses dominated by Democrats since that original law.
Due to Baumol's Cost Disease and email, it doesn't make sense to be paying labor to drop off junk mail every single weekday. Cutting this by 2/days a week in urban areas and by 3 days/week in rural would save incredible amounts.
The U.S. Postal Service doesn’t make financial sense because it is fundamentally not a financial idea. Like other aspects of Constitutional U.S. governance, its purpose is to stitch a people together into a nation.
Your argument rests on a pretty narrow definition of "financial sense".
Like when you buy food you like instead of the cheapest calories, this can make financial sense, because you value the preferred food more than the additional cost.
At around $20 a household projected loss for 2027, you wouldn't have to try very hard to do better than dismissing discussion of the value it does or doesn't bring.
Yeah, we're in weird times where things that should be non-partisan have become very partisan. The postal service has become a lightning rod for the most extreme elements; privatization and anti-democracy activists want to kill it, including the guy who's supposed to be running it. Since they can't easily kill it, they just make it suck in every way they can.
Check out the National Technical Information Service, which publishes some federal data sets like the SSA Death Master File (https://dmf.ntis.gov). They charge $1,531 per seat per year to do online queries or $2,930 + fees to download. They are exempt from FOIA, but the products are still public domain.
Half of their links are dead but they've updated their website to say:
> We help agencies leverage new and innovative capabilities, including super intelligence, machine learning, and robotics.
Well just wait a day or so and the data will be free. It’s not like these polygons are constantly changing (or that the impact of slightly stale data is of any consequence). The speed is at best mail-scale, and someone will surely leak it to an open data source.
In Europe that kind of geodata is typically free and maintained by a governmental agency. I haven't checked for every country, but the three countries I needed that for had that on websites. In fact I didn't even had to go to websites it was a standard map layer in QGIS (great software btw.)
It always kind of weirds me out just how much of the inside of The Pentagon is dedicated shopping space. I mean, I get the full food court and maybe the barbershop. But the full-sized CVS? The candy store? The local chain specialty bundt-cake store with the punny name? The souvenir shop in the central area, away from the souvenir shop where the public tours go. It feels like it's trying to say "America's defense rests on exporting commerce" without knowing it's actually trying to say that.
In reality, USPS is starved of resources, and like with all other open source things, some billionaire would make millions out of it without contributing anything back.
USPS is run as a financially self-sustaining business wholly owned by the federal government. (So, not for profit, but also not funded by the government.)
For a long time it was in this weird double bind because the federal government mandated generous health insurance for USPS employees, but did not allocate funds for it, because USPS was supposed to provide its own funds to cover all expenses. It looks like new legislation (PSRA) was finally passed in 2022 to move that burden from USPS and onto the OPM.
It wasn't intended to be profitable. It was intended to provide a public service in the national interest. Any funds raised through postage were to hopefully offset the considerable costs, but the considerable costs were in and of themselves seen as a necessary expenditure for a critical responsibility of the federal government, no more or less so than the considerable costs associated with the raising of armies and navies for the defense of the nation.
Anyway it was just a silly joke about how your postal system is state owned but (sometimes) expected to make a lot of money, rather than run at a loss like a normal public service.
Same. Everybody builds one, and then eventually realizes it makes the model stupid and confused. The right solution is pretty much always developer documentation in the repo, in roughly the same shape you'd make developer docs for humans that you want to be able to get up to speed about your codebase really quickly.
This has the benefit of being something you can easily read over every now and then to make sure the model hasn't baked in some random comment you made about a specific circumstance a month ago and made a weird religion out of it.
As usual, I was right about the market but way too early. I sold all of my AI-exposed stocks last year, as I knew the investment going into it couldn't be recouped in a reasonable time. If I'd held longer, I'd be up about 2x on those shares even if I'd only sold in a panic yesterday.
But, it also means I can probably reinvest in some of those stocks now that a correction is in progress. There are some excellent tech companies with strong leadership and strong growth (e.g. ARM) that have been too expensive for a long time; still too expensive, but getting closer to reasonable.
It's about timing and direction. Timing doesn't matter if it's not a well-run business.
Most of my holdings with AI exposure were and are well-run businesses; they were when I bought them, when I sold them, and throughout the AI-inspired rise. The problem was I didn't believe there was sufficient market opportunity to justify the stratospheric rise, and I think the market has now begun to catch up to that thinking. The market has begun to agree with the premise I based my decision to sell on.
I suspect the decline will continue and I will be able to buy back into at least some of the stocks I sold at a price better than what I sold them for.
So, yes, my post was to say I am frequently wrong on timing, even if the premise I based a decision on turns out to be true. I don't understand people, but I understand tech.
Yes, well, remember the dot-com boom/bust cycle? Lots of people lost a lot of money then but not all of those stocks ended up becoming worthless. Some held on to what they had and made a fortune once the internet-hype was over and the internet-era began. The same could happen with some - but certainly not all - of these 'AI-exposed' stocks. Once the 'AI' hype is over the 'AI era' will probably begin, especially if the dearth of RAM and matrix processing capacity is solved somehow.
That was the first time I was right but early. I sold out well before the peak to start a business with the money. I made money, roughly tripling my investment, but lots of folks made more money (and lots of folks lost a lot of money). In this case, I sold at a profit, but I could have made a bigger profit by being more slow to respond to my understanding of the market.
With all respect, you were not right. Your stocks are still above where they were when you sold them. To say that the stocks you held would at some point experience a large one day drop is a certainty.
I sold at a sizable profit, more than 3x for ARM. But, not as sizable as it could have been. And, I'm not buying back in until prices are reasonable again...which I expect to happen. I don't think we're done falling.
The scale of the market fuckery is incredible. We are trying to install an MR scanner. Chillers are becoming hard to get and on long back order. The supplier says that the units we want are not really the right units for data centres but due to shortages of the units data centres really want, they are picking up alternatives.
It just shows the utter hollowing out of the US industrial base to me.
Everything has been cut to the bone so any blip in demand cannot - at any price - be effectively responded to. Even if you were exceptionally good at being able to double internal capacity, the supply chain all the way from raw materials to skilled labor pool your vendors may need just isn’t there.
This latest datacenter bubble is just the last and most acute symptom of this problem I’ve noted in my lifetime. Hyper-financialization does that to an economy.
The best anyone can do is outbid peter to rob paul.
Consider that semiconductors are the worst case situation for a supply chain planner:
- most capital-intensive manufacturing process ever invented
- lead time for expansion is several years
- demand that is highly uncertain on a timescale of months
- technology that goes rapidly obsolete
Unless one of these factors changes, it essentially guarantees that semiconductors will have the highest price volatility of any manufactured goods. As they have since the 1980s.
This sub-thread wasn't about semiconductors though. It was for chilling plant. Semiconductors I do agree are a somewhat special case - but also one the US ceded it's capability in and largely offshored.
But it's way more than just chillers. It's everything from structural steel, to electrical transformers, copper cabling, optical transceivers, electric generation, skilled workforce in the trades, etc.
Overall this boom has been relatively tiny as far as US industrial booms historically have been, and we cannot ramp up production even if there was the political will to do so.
It's not just datacenters. Look at the military industrial supply chain these days. We can't even scale up WWII era artillery shell production in any reasonable timeframe. Much less modern weapons systems.
This is why Musk rushed consolidating xAI with Spacex and rushed to an IPO. And, why it was so important to him that Nasdaq and Russell rewrite their rules so he could dip into everyone's retirement accounts as quickly as possible. I don't think he's particularly bright, but he has an incredible cunning about how markets behave and how to manipulate those markets to his ends.
The writing was already on the wall about a lot of things WRT the AI market: 1. there's a limit to how much money will be thrown at it before investors expect a return, and that limit is fast approaching, 2. it's turning out "intelligence" is going to be a commodity and it's not much of a moat, 3. cornering the market on memory and GPUs (even if they sit in a warehouse for months or years before going online, as long as it drives up the price for competitors) only works if your competitors don't figure out how to do it more efficiently and on alternative platforms.
Spacex also soaked up a lot of the retail appetite for AI, even though it's an also-ran that barely qualifies, in terms of enterprise and government usage (and for good reasons, both technical and otherwise). Every other AI IPO is going to be facing a more pessimistic market, especially if we continue to be in a tough economic environment.
In short: They waited too long, now they're in a world where they've burned so much money people expect them to deliver something that no one in the AI industry probably can deliver: a monopoly-like hold on a huge percentage of the market, and to invent a bunch of new markets that aren't currently buying AI. They have to have a dominant position for this to work because they have to be able to set their price higher than is standard now, without everyone fleeing to other cheaper providers. And, when the market for memory and GPUs finally gets back to normal and supply catches up with demand, I'd wager a lot of companies, and even individuals, are going to get real serious about open models and self-hosting (unless prices remain competitive for the hosted models).
> 3. cornering the market on memory and GPUs (even if they sit in a warehouse for months or years before going online, as long as it drives up the price for competitors) only works if your competitors don't figure out how to do it more efficiently and on alternative platforms.
My understanding is that the GPU market is largely sold out for the next few years?
What exactly is the threat here? Like, which specific competitors, what efficiency improvements, what alternate platforms?
I didn't think I needed to spell it out for the HN crowd, since folks talk about the Chinese models every day. But, OK...
At least three Chinese companies have bragged about their latest models being trained and/or served on Chinese-made GPUs. Literally within the past 24 hours, LongCat spent a large portion of their release announcement for 2.5 bragging about how efficient their training process is and that it runs on domestic (Chinese) GPUs: "LongCat-2.5 uses long-sequence training during both mid-training and SFT. With less than 80 GB of memory per domestic AI accelerator, the training system prioritizes memory efficiency and aligns its parallelism with the supernode architecture."
DeepSeek has repeatedly blown the curve in terms of how efficiently a model can be trained and served, Xiaomi and Z.ai aren't far behind, and have also discussed training and serving on home-grown hardware. I'm sure you can search Arxiv for all of the papers from these providers (DeepSeek, in particular, who document seemingly everything, even when they're doing something ahead of the US providers) as well as I can.
You've made the mistake of thinking Anthropic just wants to sell everybody a subscription to their models, so everyone can make their little spreadsheets or whatever.
That's not what Anthropic wants. They plan to be The Company. The only one. The one that does everything and makes every decision.
Sure, there are times in a familiar codebase that it would take Claude or any model longer to get its bearings and chew on all the random nonsense it wants to think about even for small changes than it takes me to open vim and make a one line change.
And, sometimes, I know models will write it in a way I don't like, over-engineering is common.
But, I mostly use models for more than few lines of code or code that isn't in a familiar project.
Bonsai 2-bit quantization is notably dumber than Qwen 27B at eight bits or even four bits. I doubt you'll get this level of quality even with five of them.
Dumber yes, but not by all that much. If the multiple agent thing really improves the quality by as much as is described here, the quality loss should be nearly imperceptible.
Not sure if I agree. Remember the error rates stack up every turn of the conversation. If a 2bit model is 10% more wrong than the fp16 version, over 10 turns that adds up
For years, Red Hat employed more kernel developers than anyone and contributed more. These days they're "only" the fourth largest contributor, after Intel, Google, and Meta, but they are still among the largest corporate contributors.
If that's a lean kernel org, I don't know what a fat one would look like.
My hearsay, from friends and colleagues that work(ed) there, is that the shipped product is not really where you see these "heavy hitters".. a lot of the maintenance happens offshore.
reply