Hacker News new | past | comments | ask | show | jobs | submit login
USPS facility in Utah does nothing but decipher handwriting (kutv.com)
252 points by gscott on April 9, 2023 | hide | past | favorite | 229 comments



In the 70s I worked in the mailroom of a college that had 1500 teachers and administrators whose departments I had to learn (for $3.75/hour). We'd get those big plastic boxes full of mail, and just know that Garrido was in Athletics, Loftus was in Psychology, etc. We'd throw the mail in the slots, then rubber band them together, pack into the electric golf cart, and deliver them. I still remember dozens of names and their departments. Also Dominique, one of the receptionists in Social Studies. Sigh.

Anyway, my cousin lives in a small Virginia town, current pop. 38,000 but closer to 5,000 back then. She has a slightly unusual first name. About 25 years ago I guessed it was rare enough that whoever threw mail in the town would remember it. So I sent the mail to Dorothy, Gloucester, VA (not her real name). No street address, no zip code. It got to her with apparently no delay.

I tried the same thing about 5 years ago but included her last name and the zip code for Gloucester. It never got to her, which makes perfect sense.


Did you include your $3.75/hour pay to imply it was a low salary?

$3.75 in 1975 is $20.97 today. A quick search shows starting pay for a mail carrier today is $16/hour. The federal minimum wage is 7.25/hour which would be $1.3/hour in 1975.

You were actually paid very well at $3.75/hour compared to today.


There’s something weird about hourly wages to me. I had a grading job in 1983 that paid $12/hour. Which definitely felt good then, especially because I could grade papers while watching football games on Sunday. I know the math would make that number much higher, ~$36 in “current dollars.” What’s weird is that it didn’t feel that way back then or now. It’s just how I process it in my head I guess. The tuition at the school then I think was $17K/year and now it’s over $80K.


I use my great-grandfather as an example of how stagnant wages are. He was a bowling alley manager, he supported a family of 4 with just his salary. This was the 1950's. When he retired he built a brand new home on a lake.

He lived in a fascinating time of growth and expansion in the USA. He was born on a farm without a car or electricity. And he lived until the early 2000's with a cellular phone in his pocket.


I'd imagine that the $12/hr grading job at the $17k/year university is almost certainly not paying $48/hr now that they charge $80k/year, or $36/hr inflation adjusted.. if the job even exists. It's probably been efficiencied away into being some grad student TAs problem.

A lot of these pre-00s hourly wage/cost stories are kind of perplexing to me. It's interesting in how clearly they show the relative strength of labor in different eras.

I remember my dad telling me he worked in a record score & lived at home to put himself through college in the 70s. Going to college in the 2000s, this was obviously no longer in any way possible.

This was an era I was making $5/hr working retail myself, against gas that was $1.50/gal & eating off the $1 menu for my meal break.. it wasn't very much.

Looking at data, 30 years before when my dad was working he'd have made at least $2.10/hr against gas of $0.35-0.40/gal or so.

Currently the ratio in our home state would be $14/hr against gas of about $3.50/gal.


Just a data point but we hire some undergraduate graders every semester and I think the pay is around $14/hr.


I think this is because inflation in the US has two tiers.

There's the big "essential" stuff- housing, healthcare, education- which have skyrocketed in price and nowadays take a higher percentage of people's salaries. Then there's groceries, consumer goods, electronics, restaurants, which have gone up much slower. So in a way, that $12/hour was good in terms of the amount of non home cooked food you could get (or live music tickets or whatever other things you considered luxuries at the time).


The tiers are not essential vs non-essential.

The tiers are relative to the labor component.

Goods & services that have gotten more efficient over time as we've introduced technology, automation, (what we do on this site) or been offshored.. have gone up slower than inflation.

Everything that has a fixed, domestic labor component has gone up faster than inflation.

There's really no other magic trick to it.

Food is a great example. Something like 40% of the country lived & worked on farms in 1900. That number is now about 1%. That means 40% of the entire countries labor was required to feed 100%, and now it only requires 1%. So a 2.5:1 to a 100:1 ratio change of food labor consumers to providers.

Other areas like education & healthcare, the lack of efficiency is often a selling point. Look at colleges that advertise student:faculty ratios or K-12 districts that talk about classroom sizes. In my experience we've probably actually gotten less efficient as everyones gotten much more precious about the education their kids receive.


Housing might be neither. There we have artificially constrained supply, driving up prices and price volatility.


How would you drive down cost in housing? Automate building prefabricated homes of a simple common designs. Use smaller lots of land. Build smaller units. Build higher to put more units on the same land. Build multi-family/apartment style to reduce single family home costs (siding/roofing/boiler/yard/etc). Basically all the things government tends to ban and populace tends to frown upon.

Most stats have shown we haven't really made much labor efficiencies on housing construction. No one wants to live in a prefabricated / factory built home, which is what would drive down the labor costs dramatically. That said I'm kind of a weirdo and like to browse the container home websites frequently, lol.

Further, modern homes are bigger (and nicer) than our parents or their parents generation lived in.

Couple that with the land component being basically inflationary due to natural scarcity PLUS government zoning rules restricting lot sizes to be artificially high, banning multi-family construction, etc.

You have a lot (basically all) tipping the needle towards higher cost.


Given that the same house costs vastly different amounts in different places, it seems reasonable that those places drive the costs to be different. Maybe in rural Nebraska the actual cost of building materials and the technology of construction is relevant.

Here in the San Francisco Bay Area the twin forces of NIMBYism and Big Government have made it difficult and uneconomical to build anything. The NIMBYs have enacted extremely restrictive zoning laws almost everywhere that prevents you from entitling projects on the front end, and well-intentioned over-regulation has driven up the cost actually building anything on the back end.

After you spend years in public hearings placating bored and angry old people, you then need to hire more professional services, pay higher impact fees, pay higher permitting fees, etc than practically anywhere else in the country. It'll cost me more to knock down my house here than it would cost to build the same plans somewhere else.


Yes, the land & zoning compliance costs in HCOL areas crowd out any of the actual building materials (and some of) the labor component.

My parents live on a 1 acre lot in a non-HCOL area, which costs 1/2 of what a parking spot in Manhattan sells for. Their 1 acre lot comes with a 2000 sq ft home.


You drive down housing costs by building more housing, full stop. Housing is expensive because the supply is extremely limited, and the supply is limited because it's too hard to actually build housing, because governments and NIMBYs seemingly don't want more housing.


There is tons of cheap, prefab housing available, they're called trailer parks. They're everywhere except where you want to be. The anti-NIMBY argument basically boils down to "If I can't have that house, no one should have it. Let's invoke the unfairness of life and have it torn down, or else build a vertical trailer park next to it."


I think the NIMBY argument is - no you can’t build stuff on your land because I don’t like it.

See it works both ways.

YIMBY argument is largely that government and neighbors shouldn’t have such a front ability to restrict development.


The YIMBY crowd basically boils down to "do whatever you want, I don't care, just don't stop me doing what I want". More basically, "government regulation bad, capitalism good". which happens to be one of the Republican party's main platform points (in theory, but only when it suits their own goals). The YIMBY crowd doesn't want to tear down your house, it wants to let YOU tear down your house and build something more profitable in its place.


Activists recently prevailed on my city's government to rewrite the zoning ordinance, making my block of single family homes eligible to be torn down and turned into 4-plexes. I personally don't have a huge problem with it - from a libertarian point of view - because in theory, that means exactly what you said. I can sell my property for more money or develop it and make more from it. The aspect of it that irritates me is that those weren't the conditions under which I bought the house, and the reasoning behind the change was a foolish sop toward social justice, and clearly comes from a place of sheer hatred born of jealousy toward anyone living in a single family dwelling. I wouldn't have bought the house to live in if I'd known that ten years later it would be surrounded by apartment buildings. And so that plus the castigation of people trying to maintain their space as "NIMBY" leaves me mainly with the attitude that they can have it, I'll take the profit and move somewhere they can't afford and have less leverage to screw up.

Again, trailer parks have been a thing for 70 years now. The innovation here has been rewriting zoning rules without actually rezoning, to bring the trailer park to people who can afford to live somewhere better. The funny thing about class warfare, though, is that people take their lack of class with them and end up making slums wherever they go, and other people manage to make a buck and stay ahead of them.


The only constant in life is change. Or as certain fictional character said, "Nothing is a line. Everything everywhere is always moving forever. Get used to it."

Buying a piece of land doesn't entitle us to control what happens on neighboring pieces of land.

My parents house used to be surrounded by farms and woods, now it's surrounded by homes.

100 years ago 5 story brownstones in NYC were surrounded by other brownstones, then by mid-rises, and now by high-rises.

The vacant lot next to me is now going to be developed into a house. If I wanted it vacant, I should have bought it and carried the RE tax indefinitely.

No one is banning single family homes. We are just trying to reconcile that the main cost lever is density, and in a country with growing population, mostly crowding into a few metropolitan areas.. if you want the next generation to be able to afford a place to live, we can't leave the real estate market ossified.

This is an attempt to change some laws set by the previous generation that restricts your right to develop your land, and freezes the housing market as it is, constraining supply.

If the state was literally banning the construction of single family homes, or taking yours away with eminent domain, you'd have more of a leg to stand on.


> The aspect of it that irritates me is that those weren't the conditions under which I bought the house

There were no conditions that the surrounding properties wouldn't change over time. You might have thought there were, but it was an illusion. If you want to control your neighborhood, buy it for yourself.

> the reasoning behind the change was a foolish sop toward social justice, and clearly comes from a place of sheer hatred born of jealousy toward anyone living in a single family dwelling.

This is ridiculous. The reason behind the change was likely because residents of your city want cheaper housing, and your government is trying to make that happen. Most Americans grew up in single family homes, but that doesn't necessarily mean they want to keep living there themselves, and any animosity towards single family homes is because these people are forced to live there because nothing else is available in their city, and they would rather spend their money on something else rather than a single family home.

Your suggestion that denser housing options like low-rise (plexes), midrise, or even highrise are "slums" is patently false.


Generally the government problem is government subsidizes demand for housing without subsidizing supply. More dollars chasing fixed goods.


Yes I think we agree here.


Restaurants have gotten more efficient over time? I don't think so. Not enough to explain why their price is so different from education or housing.

Class sizes haven't gone down in colleges, admin costs have gone up though.

What's common between them is they're essential services and the government increases their affordability. Student loans, mortgage interest deduction, tax breaks on healthcare spending (HSA, etc).


Yes, restaurants have gotten more efficient. When was the last time you went to a restaurant where you sat down and then someone walked over and took your order?

Okay it still happens but until the 1990s it was pretty much all restaurants except fast food chains. Now, it seems like most new restaurants use the "customer orders at the counter" model to save money on wait staff.


My original point was more on the farming end of things, but yes I'd argue there's clearly been efficiencies in restaurant service.

First, like it or not, chains are a form of efficiency in terms of management/procurement/logistics overhead.

Certainly more fast casual formats than before that reduce the need for front of house staff.

And as always with labor shocks, efficiencies get found. With the recent pandemic shock of restaurant shut downs & slow reopening... a lot of restaurants are back to full pre-pandemic high tables served (or more, in places like NYC where seating has expanded to large outdoors spaces never occupied 2019 or before), but never hired back their full staff.


> The tiers are relative to the labor component.

Im my experience, it seems more like the tiers are relative to the regulatory or barriers to entry.

Starting a hospital or college or building a home has regulatory barriers. And those prices have climbed dramatically the past two decades.


This is called the baumol effect. It's certainly part of the picture.


Indeed. IIRC it was .25 over minimum wage. I worked 20 hrs/week and that paid all my living expenses-apartment, utilities, food, bus pass.


Thanks for this, it’s always good to have up to date information.

What did you think of the rest of the story?


The story made me think of two things. First is my school, they had a huge shelving unit divided in ~4"x4" squares on the wall where all of the employee mail was placed.

Second was a story I read on hacker news fairly recently. There is a Latin American country (I forget which one) where the street addresses are essentially unusable and the mail carriers have to find the final location by using their own knowledge of the area.


I recently discovered that all of the newly built houses in my street in Japan have the exact same address. I thought the delivery guys were just wankers when they asked 'where do you live', but apparently me and all my neighbors share the same address, so delivery people have only the nameplates on the doors to go on.

Don't understand why they didn't create new addresses when they subdivided these plots from 2 to 6 houses.


Addresses in Japan have always been perplexing to me as a visitor. Tokyo has the x-y-z kind of grid system of increasing granularity, but then within the final level of granularity you kind of have to "know", or have a business card with a map on it to find anything. The building number addresses on a given street are more chronological than directional.


In spite of having been there a number of times, it was only GPS that got me to the point where I didn't need to basically give up finding something a few times a trip.


That’s really odd, if nothing else, it’s a problem for emergency services. If somebody has a heart attack, where does the ambulance go?


The streets aren't numbered anyway, so once you have a block (ban) you still need to loop round it to find the building by number (go)

Bad title but- https://geoawesomeness.com/how-the-crazy-japanese-addressing...

I can't find an image, but I was lost in Japan and had the address I needed, and a shopkeeper took out an ultra local map where each /shop/ was about the size of my thumb, to find where the address actually was.


> There is a Latin American country (I forget which one) where the street addresses are essentially unusable and the mail carriers have to find the final location by using their own knowledge of the area.

There may be more than one, but Costa Rica is definitely addressed this way. A hotel I stayed in was 70 meters behind the church of Saint somebody, there was a great restaurant who was “near the old tree” (and everyone knew which one!). It was impressive.


>the mail carriers have to find the final location by using their own knowledge of the area

I imagine it's pretty uncommon these days especially after all the E911 rationalization of the past 2-3 decades. But when I was growing up in a hardly back of beyond Philadelphia suburb we had a rural delivery route which was just a street name.

I imagine most everyone at a university in the US these days still has a physical inbox of some sort. But working as a not-officially-remote worker at a company, you basically can't get a work package to me unless we make special arrangements. (I basically give you my home address on the rare times I need something physically delivered.)


> There is a Latin American country (I forget which one) where the street addresses are essentially unusable and the mail carriers have to find the final location by using their own knowledge of the area.

Nicaragua and Costa Rica use a landmark based addressing system, where "From the park, 1 Block South" could be valid address.


> There is a Latin American country (I forget which one) where the street addresses are essentially unusable and the mail carriers have to find the final location by using their own knowledge of the area.

This is definitely the case in India and I used to look for a postman when lost.


Was it the Costa Rica story?


Yes! That is the story.


When I was an undergraduate mail to "tzs, 1-59, 91126" would have probably got to me.

91126 was the zip code for Caltech undergraduate housing, which would have been sufficient to get the letter to the Caltech post office. The Caltech post office would have recognized 1-59 as the building code for Ricketts House and put it in the bundle of mail for Ricketts. The carrier there would have put it in the "T" box in the mailbox array in Ricketts lounge.


Lived in a house that had its own postal code in the country it was in. Tried sending a letter to the postal code, It didn’t make it :(


In the US that generally works. I write my return address that way (for the few letters I send any more)


LAME AF


Those dedicated zip codes are increasingly rare but def a hidden superpower you wielded appropriately


I send a fair amount of mail to addresses very near where I drop the mail-- as in, in the same ZIP, if not almost the same mail route (but not at the actual post office). I'm constantly puzzled that a) it takes a few days to reach the addressee no matter how close you drop the letter to the destination address and b) when I check, local mail is being postmarked at a regional center fifteen miles away....


I have previously worked for USPS and can help clear up what is happening. The mail carrier and their assigned station (Post Office) should be thought of only as the ingress/egress of the postal system. It would be wrong to say that no sorting happens there, but the type of sorting that doesn't happen there is why your mail is being postmarked at a plant (regional center) fifteen miles away.

I'll use the location that I worked at as an example. In Wilmington, NC, once your carrier returns to the station, they drop your letter into a collection bin. This collection bin is driven to Fayetteville, NC, where it is then processed via OCR. After being OCR'd it has a machine-readable bar-code, and will be sorted for delivery to a local station. The machinery involved in the OCR and sorting process is larger than any of the local stations I worked at. Additionally, almost all of the mail pieces that I delivered were intercity, not intracity. Your use case is the exception, not the rule.

There are many exceptions and corner cases to what I described, but that's what the process looks like for most mail pieces. Locating the scanning and sorting machinery at a hub rather than each individual spoke makes sense. UPS, where I currently work, has some minor differences, but the hub and spoke model is largely the same.


At some point in the past, the balance of intracity vs intercity mail would have been different.

Now, utility bills come from some distant office, as do letters from banks, and we don't send invitations very often.

X years ago, bills were sent by closer companies, the bank was local and it was polite to invite someone to dinner by letter.

With that, it would have made sense to separate local and other mail as soon at the postman brought it to the post office.


Both of your observations are due to constant relentless cost optimization within the postal system. Local post offices, for the most part, don't postmark mail at all. Everything gets batched to the regional processing center where it goes through the big automated sorting machines.

I seem to remember a "local only" box in the town where I grew up but that hasn't been a thing for decades now.


A friend lived in a smaller town where mail was processed in the nearest larger city (in a neighboring state). But, if you needed something sent to (most of) the rest of the state, it was faster to have it processed in a different city in the other direction. They had a specific drop box for that, which cut delivery from two days to one. I, living in the second city, was stuck with having my mail routed out of state to get to him.


FWIW, early 2000s where I lived had a local only chute. When I was <10 I always wanted to put the letters in the box.


Knowing that our mail passes through the regional processing center before returning to the local area, we expect about three days if we mail a neighbor.

Post offices around us have two mail slots when you walk in to drop mail in the box: local only, and everywhere else. Presumably they have (once had?) separate localized processing for those in the “local” bin. Because of post office closures and reduced hours, I think everything is just passed off to the regional facility for the last few years.

Another interesting discovery: I live about 3mi from my parents. They’re over the line in another county. My post office services them, but when I still lived in their house, all our local government business (schools, taxes and whatnot) was in the county seat about 15mi further away from the serving post office, closer to the next regional facility further south. If we ever dropped mail, say from mom to her mother, in that town, it took at least an extra day because it had to hit two regionals before getting to grandmother’s local.


What’s with the ellipsis,”….”? You’re using the completely wrong system to deliver a message at a distance it’s not meant to optimize. Sure maybe that would have been routine in the 1920s when almost all business was local, but those days are gone.


Ah, don't you need to know the use case before you condemn the solution? Inefficient yes, but still the best solution for this edge case. The ellipses were a lightly informed nod to the explanations offered by other commenters.


I’m not condemning the solution. I’m condemning the implied critique of a system that you admit worked even in your edge case.


I was not critiquing the USPS. The parent comment observed that the USPS no longer can complete delivery on a partial address. I was suggesting an explanation, i.e., sorting no longer is done by people with the local context necessary to complete such a delivery. That was confirmed by other comments, and indeed suggested by the article.


A funny story went around the internet last year about a guy in Northern Ireland who received a letter that only had his first (common in Ireland) name, the town he grew up in, and a rambling description of his life story: https://mobile.twitter.com/weefeargal/status/147906907614423...



And the HN discussion of that video: https://news.ycombinator.com/item?id=32388279


This is a pretty good example of the things required to keep a service running reliably that likely would not be done by a commercial company with no delivery guarantee offered. I can't imagine FedEx, UPS et al going to these lengths.


UPS and FedEx would have similar facilities if they didn’t require you to mail through their stores or partners, or with a preprinted label.

It’s the guaranteed pickup of any stamped paper from any mailbox that is the the marvel of USPS, when you think about it.


This is a pretty good example of a thing that is not required to keep the service running reliably (akin to sending an IP packet with only a partial IP) and that governments, with no incentive to properly allocate Earth’s scarce resources, would typically keep running aimlessly.

Then they tell us we consume too much resources.


Alternate view: this is a form of ECC, which is widely used in computing.


USPS is a government service but it is self funded, the domain system has evolved a pure unregulated capitalist way of dealing with this dilemma.

Purchasers of domains are simply expected to buy all misspelled and closely matching addresses or a opportunistic advertiser or worse will set-up shop.


If a private company contracted this job, not only would they earn money from the USPS, but they could provide the service to other companies. Lots of public and private entities need this service.

But even better - they'd build up an institutional set of training data for machine learning and could build the best handwriting recognizer on the planet.


> not only would they earn money from the USPS

You mean siphon money from USPS

> they'd build up an institutional set of training data for machine learning and could build the best handwriting recognizer on the planet.

You mean, they'd use private people's data to build up a closed dataset they'd sell to other companies for money.


I agree that the privatisation of the USPS would be a very bad thing. However the language you’re using, esp, in the second point, is optimised to be alarmist at the expense of being accurate. You aren’t doing anybody any favours by being intellectually dishonest.

Is it a closed dataset, or are they selling it to other companies? Your language evokes privacy concerns. I’m sure you know that, and I’m sure that that’s intentional. Using someone’s handwriting as training data for OCR, as part of a training dataset significantly large enough to be used as training data in the first place, does not pose a material additional privacy concern.


> Using someone’s handwriting as training data for OCR, as part of a training dataset significantly large enough to be used as training data in the first place, does not pose a material additional privacy concern.

USPS literally processes people's private addresses. I'm amazed how you can say "Does not pose privacy concern" with a straight face when the entirety of the past decade has shown us that if there's private data, companies will abuse it, sell it, leak it etc.


does not pose a material additional privacy concern.

But it does. It so very much does, because it always does. Always. Every time.

Data gets anonymized, then other firms de-anonymize. Co-mingled anonymized, too.

And once the MBAs see it, it will be pried away, and sold, and processed, and snarfed, with hooks during training too.

Because it always happens. Always. Every time. It's like candy to a child, like food to a starving man.

At this point, we are insane to think data stored or even observed, is not being monetized against us.

So the parent's post is not alarmist, because it always, always happens.


I worked ~20 years ago doing this exact job at Royal Mail in UK. The software that they were using was not their own, but being created on contract by Lockheed Martin, who did not write it themselves but subcontracted out to another company. I'm pretty sure that the basics of the software will have been sold again and again to different companies. The end result of the contract was that Royal Mail did not own the software and when the development contract was up, there were several stupid things that could not/would not be fixed without serious money being spent on a new contract.

The issues that we (as keyers) had with the software in development were virtually impossible to address, given that there were several corporate layers between the contacts. The one time that a load of developers came to our data centre to observe keyers in action, one of our quick-thinking lower managers managed to snag a contact detail and for about a month there was an easy option to get small niggles fixed.

oh, btw my tph was 2-3000 so I don't know why the USPS workers are so slow. Actually thats just boasting, I probably do - it is likely that the software is a whole lot better than it was, which means that the quality of the images that it cannot handle has gone right down and takes longer for a human to process as a result.


I did the same job back in the mid-90's at the facility discussed in the article when it first opened. The piece rate was probably higher, though the measure we cared about was keystrokes per hour so I don't know if I ever saw a real statistic for this.

The work-load almost certainly has changed with technological improvement. Back then, my recollection is that we'd see more machine type than handwritten addresses, but only slightly. Also we were only filling in the gaps depending on how well the original OCR worked. Sometimes we'd have to key in the important portions of the whole address (house number, part of a street name, postal code)... sometimes just the postal code, and sometimes only a portion of the postal code. Some of these data entry modes were faster than others; the extended "zip code" data entry only required two numeric keystrokes whereas the full data entry mode required much more time.


I worked on the LM side of this team ~8 years ago, and it sounds like they learned some lessons since you had to deal with them! The ML team was still in house and working on all sorts of projects, and there was a sizeable software team working on the software around this.

One of my most valuable lessons working there as a green engineer - you can figure out more in an hour sitting down with the end user then months of meetings before then.


What does the data entry look like for this job?

The article says, "Our keyers key about a thousand images an hour, so they're spending about 2 seconds on an image".

Are keyers typing whole addresses in that quickly?


No, at least I wasn't. My kph was about 10-12k giving 5-6 keystrokes per task if I recall.

At least in the UK, if we had a postcode then that was enough. Not even required to type the whole thing. The UK postcode is split into two parts, the first part (outcode) designates the sorting office to send to and for newly posted mail on the afternoon shift then thats the main thing they wanted immediately; the image has been taken and the mail item has been tagged and will be in a van for some hours. If no postcode then usually the postal town is enough to get it on its way. When I was there, the second part (incode) was not done but I think these days that is processed by the night shift before the posties came in at 5am to take it for delivery.

Looking at the pictures in the article, they only have one address on the screen at a time. We had two; the one you are keying and the next one stacked above. It doesn't take long to learn to pipeline, so you are looking at the top image all the time but typing the details from the bottom one which was previously up top. There was an indicator to the side saying what it wanted and you generally only had to type that or various rejects.

also, 70 hours a week of that is pretty insane. Our normal was 40 and sometimes (eg xmas) up to 20 hours overtime was available. We worked 60 mins for each 10 minute eye break, and one 30 min meal break during the shift.


USPS already has extremely good handwriting recognition tech. If a human is reading it, it's going to be a very difficult and unique case that the machine couldn't figure out.


yes, but does AI can deliver mail to Dorothy, Gloucester, VA (see one of root comments)


Notice that the Utah facility handles the entire United States, so unless "Dorothy" is very famous for living in Gloucester ("Trump, Florida" might work) that's not going to work even with humans in the loop at this stage.

It'd get delivered, if it does, because rather than drop mail they'll send it to Gloucester and the delivery people in Gloucester might be able to fix that, so nothing to do with this facility.


On Tom Scott's video, you can see the software is provided by a commercial company "Eagle", and Siemens is mentioned too.


I can't help feeling a little sad.

Handwriting was such an integral part of a person. You'd almost be able to identify a close friend just by looking at their handwriting and you'd see it frequently enough on letters and things.

As someone who practices Western calligraphy and who carries a fountain pen, i can't help but feel a little sad reading articles like that prove that the written hand is dying out.


>> Handwriting was such an integral part of a person.

That's a noble thought and all but it doesn't always work like you say. I bet my school teachers would recognise my handwriting from a mile away, yet even I have trouble reading what I used to write back then. And I wrote all the time. The only time I didn't write was when I was reading (I have this weird taboo about writing on books. I'm fine with drawing stick-figure animations on the corners though; you should see my school editions of the Iliad and the Odyssey, truly epic stick figure swordfights). And just to be clear, I never had any trouble with reading, nor with grammar or syntax. It was just my handwriting that was horrible.

My handwriting was what you'd call cacography, the opposite of calligraphy. It was like a chicken dipped its claws in ink and went digging for worms on the page. Very recognisable, but really not very good handwriting at all. I think it's because I was always in such a hurry to take it all down that I didn't stop to think about reading it back later.

To be honest, computer keyboards saved my writing.


> It was like a chicken dipped its claws in ink and went digging for worms on the page.

Hilariously, my handwriting was frequently described by teachers as "как курица лапой" which is just about the equivalent of your description.

Despite my terrible handwriting though, I still enjoy writing by hand. I've found I rarely ever end up going back to much of what I wrote down, but the whole process of turning thought into word into writing is very therapeutic and helpful for remembering and processing things.


It might be a Slavic thing, in Polish bad handwriting is called "piszesz jak kura pazurem" which means "you write like a hen(would) with its claw".


Where I am from, it's called "chicken scratch".


In France it's "pattes de mouche", meaning the letters look like a fly's legs.


Here in Brazil we say "parece titica de galinha" (looks like chicken shit). Seem like a universal thing, comparing bad hand writing with chickens...


>> Hilariously, my handwriting was frequently described by teachers as "как курица лапой" which is just about the equivalent of your description.

My turn of phrase wasn't entirely original, I'm Greek and we call handwriting like mine "chicken scrawlings". So basically what I wrote :)


I think it was the same for me.

Writing fast and nice is impossible, back then and now.

I learned to write nice and slow, it is completely impractical, can't do it for long texts, but it makes me feel better after so many years of only ugly handwriting.


> I have this weird taboo about writing on books.

I completely understand this, I used to be unable to write in my books, but forcing myself to start writing in my books was completely worth it. Now it feels natural. Its extremely valuable in non-fiction works to be able to pull open to a chapter and instantly find everything I found noteworthy.

Writing in books also allows for a good notetaking workflow: read a chapter or section while highlighting interest parts and writing symbols/questions/notes when needed. Then re-read the chapter and take notes, focusing on what you highlighted. Move questions and comments to the notes, and perhaps research the answer to those questions if necessary.

Finally I would also suggest writing in books in order to instill a proper relationship with property. Books are tools for conveying knowledge, writing in a book simply continues to fulfill that purpose. They aren't idols that ought to be put on a bookshelf to look at, they ought to be read and referenced. I think being willing to write in books helps create a healthier understanding of ownership: we own property; the property doesn't own us.


Apropos your point on writing in books, here is Mortimer Adler (of how to read a book fame) exhorting you to go ahead and scribble in your books

https://stevenson.ucsc.edu/academics/stevenson-college-core-...


Tell us more about yourself.


As someone who was ridiculed by peers and teachers and having had grades be dropped whole letters for poor legibility, I don't share the feeling. I tried years practicing and I am skilled in other dexterous tasks, but handwriting is not one of them. I still use a fountain pen though.


When personal computers hit the scene it was like manna from heaven. It has always been a huge physical challenge for me to write. I have to expend an extraordinary amount of energy to keep my arm and wrist steady when writing. The result is barely legible at best. Torture.

Later when programs came out they let you draw perfect squares, circles, and lines, I felt we had finally ushered in a New Age. I’ll be happy if I never have to pick up a pen again.


My handwriting still looks the same as it did in fourth grade. I was blessed with a forward looking teacher and school system that recognized my interests early and didn’t penalize me for bad penmanship. When my parents asked about my handwriting they said “why worry? He’ll be typing everything when he grows up - we all will!” And they were right!


Different strokes for different folks, I guess. And I mean that almost literally. ‘Carrying around a fountain pen’ is definitely a bit of…fetish signalling. Dont get me wrong. I totally understand feeling sad about something that you’re into sinking further into obscurity. I’m just a little weary because it’s such Gen-X-and-above fodder to treat it as a sign of society in decline.


because it’s such Gen-X-and-above fodder to treat it as a sign of society in decline

I think it is more of a "reliance on tech", not of any gen, but of "where society is".

The point of cursive, was writing fast and legibly. You can write far faster than typing, for example, on a smartphone keyboard.

There is no correct or incorrect thing here, however, losing cursive means we are trapped in a world where everything we write, everything we do, is tracked.

This is really more about optimal use of a medium (paper), and is equivalent to being able to type on a qwerty keyboard, which is equally learned and weird.

So for me, it is about the loss of non-electronic means.

And that may be OK, but we may also find it is not some day.

EG, we will surely want efficient paper use, if we all end up in some controlled state. You'd be nuts to use any computing device to communicate then.


>> The point of cursive, was writing fast and legibly. You can write far faster than typing, for example, on a smartphone keyboard.

Smartphone, maybe, but I average 120 wpm on a computer keyboard and I can't get anywhere near that on a phone. By hand, if I write as fast as I think, I have trouble reading it later, myself. And if I try to write legibly --let alone trying to make pretty-looking letters-- my fingers start to hurt after a while, because I have to make such an effort.

I don't think I'm doing anything wrong. Sometimes I even find some of the shapes I make as I write by hand aesthetically pleasing to look at. They're just not easy to read.

I can't read cursive either.


I can't read cursive either

Aren't you making my point here? You can't read it, and I presume write it, but you're saying printing can't keep up legibly?

Cursive fixes that speed problem.


You know, I think I'm confusing "cursive" with "calligraphic". My handwriting is "cursive" in the sense that letters are connected. But it's still very hard to read, both for me and everyone else who's ever tried.

What I was trying to say was that if I write by hand as fast and freely as I write on a keyboard, my handwriting is illegible. If I try to control it and make it more legible, it slows me down very much, and my hand hurts. And btw, it's still not pretty.

So basically I confused "cursive" with "neat, tidy, calligraphic handwriting" when even my sloppy, ugly, unreadable handwriting is at least somewhat cursive. Apologies- I'm not a native speaker of English.


If you have trouble with cramped fingers, try writing and steering from your elbow. Keep the pen in the same position in your hand without clenching and only move your elbow.


I should try that. When next I can find a good place to write. One more advantage of writing on a computer keyboard, you don't need a proper desk. Mine's always on my lap, even at work where I have a desk :)


I can type about 15-20 wpm on a computer keyboard. I don’t think I’m doing anything wrong either, and yet I can help but suspect you have learned and practiced something I have not.

Maybe something similar has happened with your writing.


I grew up in the cursive age and I'm mostly incapable of writing well by hand. No decades of practice ever budged that needle.

Past that, I'm glad to see cursive go away; it is only legible from the people who are capable of doing it well.

I've been doing genealogy for some years now and I can't count the number of unreadable documents that would have been saved, had they been written in block. For each one, I am reminded that cursive begins with curse.


people can still write by hand without cursive, this seems a little hyperbolic


people can still write by hand without cursive, this seems a little hyperbolic

At 1/10th the speed, with hands that tire much quicker.


when has that ever mattered? I grew up well before personal computing and even then we were using typewriters and later word processors… even in college I printed my notes and kept up fine


Tom Scott did a fairly decent video on this a few months back:

https://www.youtube.com/watch?v=XxCha4Kez9c


I'm always surprised at the strangely boring topics he manages to turn into content.

Was even more surprised when I went to the UK, turned on the TV, and he was on University Challenge.


Celebrity/old people university challenge.

He was on only connect back in the day


Christmas University Challenge is the title of the show, it's a variant of University Challenge designed to be shown over Christmas (but obviously filmed earlier in the year) where instead of current students of the named Universities (who will in most cases be undergraduates and thus relatively young adults) it's past students asking similar questions. They're not so much necessarily "celebrities" as they are "notable" in a Wikipedia sense.


It's little coincidence that the last of these facilities is housed in Utah. The Church of Jesus Christ of Latter-Day Saints contributes a significant amount to family history work, which includes reading and deciphering old records written mostly in cursive. They also encourage their members to contribute to this work. Around half of Utah's population are members of that church, so there naturally is a much larger pool of qualified cursive readers to select from there.


Not sure if this remains true today, but back when I was young you could go to your local LDS church and use their computers to search genealogical records. It may not be as useful today, but in the 80s it was some of the best stuff you could easily access.



I went to clicking around on this thing for several hours and found out that I am a descendent of Lady Godiva. Pretty awesome to see.


So there are still FamilySearch center in many meetinghouses of the Church of Jesus Christ of Latter-day Saints. They have also put almost all their records online at www.familysearch.org

If you go to the meetinghouses however you can also get access to the records of several other organizations that have partnered with FamilySearch for reciprocal resource sharing.

Disclaimer: My primary client is FamilySearch


“after 70 hours a week it becomes like second nature”. Yikes, that is a lot of time working.


After 70 hours a week staring at terrible hand writing you reach a second level of consciousness. Involves drooling.


  "It can be hard, especially when you first start," Bousha explained. "A lot of them are not used to reading cursive and interpreting what they see. But after 70 hours a week it becomes like second nature."
70 hours a week just reading and typing addresses? Hopefully that's hyperbole.


Fascinating---most of the examples they showed were immediately legible to me, and not at all examples of "bad handwriting".


That’s just editorialization on the source’s part: the facility exists because cursive is legible to humans, but not easy for machines to automatically classify.


Google Lens is able to read my handwriting, and I think the tech for it has been around for 10+ years.


My handwriting is not your handwriting, nor is it millions of other Americans' handwritings!

There's also plenty of edge cases: letters get wet, they smudge, get torn or folded, and so forth. A postal system that processes 420 million letters a day[1] with a 99% true positive rate on OCR still needs to divert 4.2 million letters for manual review.

[1]: https://facts.usps.com/one-day/


And as a postal service, writing off 4.2mil letters is absolutely unacceptable so thus we must dedicate the resources to mitigate this tail risk.

This is a common pattern, and why I think AI risk is overstated…unless we suddenly decide to start wholesale writing off the consequences of tail risk.


It will be little solace for the 98% of employees that get laid of by AI to know that at least 2% if their co-workers remain employed.

In other words, AI doesn't have to do anything perfectly to replace a large percentage of the workforce.


That's why this is the last facility that does this job—all the other ones are now unnecessary because machines can process enough of the mail.


"Cannot reproduce the bug on my local development environment. Ticket closed"


It is a little depressing that near-perfect cursive writing is now regarded as “illegible”. I suppose the same thing happened to copperplate, though.


Having been belittled by my second-trade teacher for my abysmal cursive, I’m pretty happy to see that particular bit of arcana go by the wayside.


Amen. I live in the Czech Republic and they still teach it here as standard, and everyone writes that way, but I just can't read it at all. Such a pointless thing in the 21st century.


My 12 year old daughter has not been taught cursive but she read it just fine.


I couldn't be happier about it. It's annoyed me ever since the third grade. And every time someone tells me "your signature must be in cursive". It's nearly as bad as daylight savings time.


I hold the generation as highly regarded


They mention in the video that they can't show real mail for privacy reasons, so it's not obvious to me what we're actually seeing, or if it's representative of what the workers see.


Its kind of wild how critical infrastructure a reliable postal system is.

I moved to a country without one and it’s a massive headache. Letters may or may not arrive. If they dont arrive they may or may not be sent back to the original sender. I imagine some room in a government building with millions upon millions of undelivered mail.

I couldnt even get important mail from the US government. I ended up using a family member’s address in the US.


Yes, here it is paid couriers or nothing. Even street names and building numbers are lacking. The postal system is just a gotcha.


One of these centers existed in a city in which I lived, Duluth, Minnesota, from 1995-2005. It was a fairly big deal for a city the size of Duluth, employing 1,000 people to start and ranging up to 1,400 at the peak. A former employee started a FB group that has a good photo of what these centers looked like in the 90s: https://m.facebook.com/photo.php?fbid=10155932292850485


Related HN discussion about handwritten mail delivery from March ("What is the minimal possible UK address?"): https://news.ycombinator.com/item?id=34995370


Old English jokes about lost mail to "degenerate bawd" finally making delivery to the electricity supply company


Maybe they should create a mobile application that uses the API of their OCR system and allows users to take a photo of the handwritten address in order to determine, before sending the letter, if it is readable or not.

And then lose their jobs.


As if the people who write unreadable and nigh impossible to understand addresses would care to use such a system.

These are not “doctor’s handwriting” issues.


I would estimate that the Venn diagram of people who would use such a system, and people hand-addressing postal mail, is two separate circles.


Oh definitely, if you are willing to make it readable you would write it in all caps or even use a printer. It's either separate circles or a Venn diagram with the center containing only dozens of people.


There are plenty of details online of what makes an address easily readable. Just look at the specifications for getting bulk sender discounts.


I wonder how long this system will exist before ML gets good enough that it just makes more sense to send illegible mail back to the sender and say, "sorry, whatever you wrote is not a valid address, try again".


A lot of the stuff they process are actually machine printed, but for some reason the OCR just rejected. Had a bud that worked at the Utah facility, and the same sentient can be found through out the comment section of the Tom Scott video.


If it's so easy for a human to fix it, I wonder how long the pipeline is between feeding that as training data into the next iteration of the OCR, and getting it deployed live?


No idea, I heard it was a multi year problem though.


Never; this facility will likely continue to serve for as long as the people exchange mail.

Sure, you might say "the sender should write better", but that's needlessly discriminatory and flies in the face of what the United States Postal Service stands for. Not being able to immediately read the delivery address is not an excuse.

The USPS considers it an absolute failure to do their duty if they fail to deliver mail that was otherwise furnished properly, and I commend their commitment.


I really think you're overestimating the average postal worker's sense of "duty" as well as how the post office works as a whole system. The USPS once delivered a postcard I sent to myself, because their system confused my "from" address with the "to" address. When I complained, they told me to go buy another stamp and send a new postcard. My postcard was crystal clear to a human, but somehow their automated system got it confused, and yet they acted like it was my problem, not theirs. This was about 15 years ago so maybe their automated systems are better now, but I don't think their "duty" or "commitment" has changed that much.


The professionalism of a given postal worker, unfortunately, varies considerably from what I've experienced, I've had the pleasure of interacting with exemplary workers and the bad apples.

For something like this I don't want to base it around the bad apples, because this really is an example of the postal service being exemplary.


>This was about 15 years ago

How can you type that out and not feel a hint of self-awareness?


The Surface handwriting recognition has been good enough for many years to recognize my cursive handwriting, which is fairly impressive, but I do like the idea of a public service striving to be usable even by people with handwriting not easily read by a machine, just like it's usable by rural users and others who are probably not the best targets from a pure business-minded perspective.


I thought the vast majority are OCR and the humans are for the ones it fails at.

Regardless, all you have to do is tell people, say at the post office, that if they type the address and cut it out and tape it on, the mail will get there faster. Alternatively they could use a label maker to print out the address and then affix it to the envelope.

I bet people still writing address labels by hand probably mostly don't know that there's a speedup for fast machine legibility.


If the destination address is illegible, chances are so is the return address.


But not nearly 100%, it could’ve been as simple as an oil smudge from the machinery or water hitting a few characters on a specific portion.


Makes sense. Mormons are used to reading horrible handwriting from all the time spent looking at old census sheets.


Can't believe how many hours I spent doing that xD

TBF the software tools for it were very helpful...plop down exhausted at the computer after 6 hours of church meetings and burn through a bunch of formal-ish handwriting data entry while half-braindead, necktie knot pulled down to the chest, black synthetic socks petting cat down on the floor, and bowl of favorite cereal at the ready. (Mormon alcohol equivalent is basically sugar)

I have friends who set regional records for processing those old forms. I want to say hundreds of thousands of pages or some wild amount like that. There was a lot of interesting geography and plenty of bio-stats to read, so for natural readers it was quite a comfortable hit in its way.

I did get into handwriting analysis later and that was more consistently fascinating on the other hand. There was always an open invite to bring found samples to the next meeting (this was not a Mormon thing, to be clear).

Everyone would then take turns pointing out possibly valid observations, ooh there's a felon's claw, anybody else see that, etc. Then at the end of the meeting they'd ask the submitter to share the general or specific identity of the sample. "That's my uncle so-and-so, the rascal was on a chain gang for 10 straight years" and so on.

But sometimes you would remember seeing the sample somewhere, so you had to brace yourself because it was definitely a serial killer and you'd spend the rest of the week thinking about that sample afterward.

In the society library there were texts about individual letters...like people have authored multiple books about the written letter T, and so on.

For my part I was paid for some analysis on the side and learned the hard way that you have to kind of hold back sometimes. It's nice to know you're accurate but a lot of people just. Don't. Want. To know that others can tell this or that about them. Lol. You can effectively front-run an individual's own self-reconciliation/processing capabilities and this often brings unwanted results.

One of my own specialties was that I could usually get a good idea of someone's MBTI type through their handwriting. I had already finished my certification in Jungian typology at the time. So then, bouncing back to handwriting traits, I could identify a kind of sub-type, and this would give ideas of the various cognitive-function perspectives that were probably relevant to the individual's current stage of personality development.

I never met anyone else in the field who I could really talk to about this, though it was very useful. Sometimes you have problems in life that are less conscious, so you can't Google them, LLMs don't help, etc. As a result, finding someone who knows "people _who think like you_, and how they made it" can be really helpful.

Even took an IQ test based only on my handwriting once, pretty neat experience.


My grandfather probably spend 20 hours a week doing home genealogy work.

Going back to I guess the 80s, he always had the latest and greatest computer system in his home.

He wasn't super technical, but he knew how to use them for the things he cared about.

One year he hit an issue with a store-bought computer. I forget what the issue was, but he probably spent $5k on the computer, and it had hundreds, if not thousands of his work hours saved on it, and customer support led him down a path that wiped his hard drive.

After that, he would say, "Computers are light light bulbs, you can't trust them after a certain point..."

He complained about the customer support being so bad, so I told him we could just build one from pars if he wanted.

His eyes lit up, "You'd build me a computer?!"

So every 2 years, we would get together and build him a new computer.

He bought all the parts, and we'd build identical computers, one for him one for me... some of my fondest memories.

He got such a kick out of running performance benchmarks. 90-something year-old kid in a candy shop. He had to have the best parts.

From 1996 to 2016 or so, he had probably the most powerful gaming rig in Midland, TX... He had 3x 4k monitors, surround sound, and crazy silly amounts of RAM. Like... 128 GB was the last machine we built, back in 2016.

He'd always donate the "old one" to the church. I remember one year the kid they sent out had some IT knowledge. He was trying to write up a donation receipt, and Grandpa was like "64 gigabytes..." and the kid, thinking it was some old clunker since it was just sitting in an old Costco fruit-box at that point, was like, "Oh, you mean 64 megabytes of memory..." and Grandpa was like, in the most assertive tone an ultra-polite Mormon could muster, "No, sir. I mean gigabytes." This was like 2008 or so, when most high-end computers didn't have more than 8 GB of RAM. And Grandpa was just like, "It's got some old Nvidia GeForce 7950 GX2 Quad-SLI video cards..." And the kid was like, "Wait, how many video cards?" And Grandpa's attitude was like like, "Take that trash away, it's old and busted. Get it of my sight!" Ha.

He may have also liked playing flight simulators just a little. (=


Love it! That's too funny.

I can imagine his computer ending up in a variety of places depending on the personalities in his local church leadership xD

I learned to carry a PortableApps USB stick and a Puppy Linux USB stick with me every time I went to church meetings, just in case. One night I ended up tucked in a church library alone (under certain circumstances a random dude needs to stay in the building while other meetings happen), found a donated computer (nothing like your grandpa's!) and after digging around for cables for a while, had my Linux desktop up and running, streaming some prosaic public-access TV or something while reading Slashdot.

I can't imagine how that moment would have felt if the computer turned out to be a 64GB RAM monster...I think I seriously would have considered bringing my home computer to swap with it. I remember others being told to do the same, in cases where the waste was even less obvious (and I mean I even once had to turn away a guy who tried to donate a vial of gold he panned in Alaska to the church)...but wow, yeah that's a big deal.

(There were actually so many genealogy-computer shenanigans at the churches...someone could easily write a very interesting book about that)


Stories like this are an awesome reminder that advancements in computer technologies are absolutely worth it. Computers are tools, a means to an end, and if they make someone smile then they served their purpose magnificently.


> I have friends who set regional records for processing those old forms. I want to say hundreds of thousands of pages or some wild amount like that.

Yeah, it's insane! I loved how diligent they were! They take scans of every census record, and all the supporting forms that go into the record... then they distribute the forms to at least 2 independent people to extract data from. If there's a conflict, they send the data to a 3rd person... then they check all the entries against one another for accuracy. Retired folks are encouraged to participate, and it's pretty common for them to get "really" into it. 20-hours a week... pretty common amount of time.

They have to pass an eye exam every year to do it. My grandfather was heartbroken when his eyesight got to the point where he couldn't do it any more.

But yeah, the distributed work, with sanity checks, and conflict resolution baked-in, it was all so well organized!


I didn't know about the eye exams, that's pretty fascinating.

It kind of reminds me of some of the senior couples that were driving church vehicles abroad in mission territory...one of those couples got their car stuck in a narrowing alley (all dented up by literal alley walls, and bits of infrastructure on the walls) within the first week, then within the first month the trunk was held mostly-shut with rope.

It looked like a wreck on wheels, having very recently been new...but by god could that same driver tell a Japanese fairy tale in a way that got kids SO excited. Funny memories.



Sorry, that's not how that works, and I'm not taking on new requests from the general public right now--haven't been for years at this point.

If you want to test someone's skill in this area, you should always confirm their availability first, then ask them what kind of sample they want, and how the process will work from their end.

You should be willing to put your identity on the line in some way as testing quality collateral.

Otherwise, for one, it can look as if you, new user K69bLrdd5K28vX, may be likely to cut and run if the analysis doesn't turn out the way you thought / suspected / etc. Analysis of any quality takes a lot of time and effort and this kind of outcome is very disappointing.

You should also be willing to explain how you arrived at your personality type, i.e. is it only the reported (tested) type based on a free internet test, or is it "best-fit type" based on a recommended, integrative process designed and tested by a recognized publisher; were the test statistics made available to you for accuracy / reliability / test-retest, etc. and can you share those?

Anyone certified in type should know to emphasize this aspect, or else any testing based on a given type-constant is qualitatively compromised. (I hope you can see why a result that's simply derived from a free internet test with poor or unknown accuracy & reliability would complicate matters)

That's not to say it's not worth a test. A lot of idea-oriented types are interested in the concepts behind it, too. I always found that most INTPs were game, for example. Good luck to you.


Very curious to hear more about this - how does processing census sheets fit in to LDS?


Oh... um... I mean look, it's a bit controversial.

They document family trees, and then post-humorously baptize the dead so everyone's kin can get to heaven.

They mean well, and it's coming from a good place, but it's a little weird and it can be off-putting to know that they don't think any other religion's baptisms count so they have to do it over the LDS way. But look, they mean well.

Some background on the controversy.

https://www.nytimes.com/2012/03/03/us/jews-take-issue-with-p...


Posthumously, rather than post-humorously.

As you note, many take a rather dim view of the practice (and it is very much not coming from a good place).


> and it is very much not coming from a good place

Can you elaborate? I know nothing about this, but from the description it seems well-intentioned so I'm curious what isn't?


I posted a link in an earlier part.

But basically instead of just "Oh, we're just saying some prayers for your dead granny..." -- which I think everyone would be fine with, it's the Mormons saying, "Oh, we're saying some prayers for your dead granny... because without true Mormon salvation she'd be burning in hell forever."

The Mormons also have "stand-ins" that they baptize in place of the person. Usually kids. So a kid goes up and says, "I'm so-and-so, grandmother of such-and-such..." and then they get dunked in a tub. And it also serves to further indoctrinate the kids... who are re-baptized a bunch of times.

My view... all religions are nutty. Not sure where we draw a line on any of them. But I don't really care what people do in the privacy of their own temple, or what internet archives they want to spend their time transcribing. If it lets them sleep a bit better knowing my dead great grandmother is with her family in Heaven... that's great. Or at least, it's no going to hurt her any. (=


> Posthumously

Good catch. This is how we'll be able to spot AI in the future. It won't make nearly as many dumb typos as humans do. (=


From that article:

> Church policy is that people baptize only their own dead relatives

but indeed, some adherents apparently do proxy-baptise unrelated people.


I long for the day when we treat all religions’ baptisms as having precisely identical value.


There are a few parts to it.

First, Latter-Day Saints believe that one needs to be baptized in order to enter into heaven (based on the teachings of Jesus Christ recorded in John 3:16). For this to come about for those who never had the opportunity to be baptized (let alone the countless individuals who never even heard of Jesus Christ while alive), they perform proxy baptisms on behalf of dead individuals. Obviously they don't baptize corpses; instead, they have a member of their faith baptized again, on behalf of an individual who has passed away.

To find out who of their family tree needs to be baptized, Latter-Day Saints perform tons of genealogical work, which includes indexing handwritten censuses into digital form. Hence the census sheets.


I know a few people who have been hired at this facility but washed out in the training and testing program; the speed and accuracy required to work here is outstanding and evidently quite difficult to achieve.


I think it must be at least a decade since I’ve written an actual complete sentence by hand, a paragraph certainly must have been even farther back. Maybe I should pick up a pen.


>What’s the Point of Teaching Cursive?

https://archive.ph/WuZhn

>What Killed Penmanship? We’re all texters and typers now, so if you can’t read that grocery list you scrawled, you are not alone.

https://archive.ph/6mm4j


The world needs such facility for processing doctors' doodles, or to start teaching handwriting in medical universities.


Definitely a good use of everyone's time in a world full of electronic medical record systems.


Someone unable to write is half-illiterate, still. Not always and not everywhere printer is available or sometimes doctors don't bother issuing printed receipts.


Printers? Who's printing? Every prescription I've gotten in the past few decades has gone electronically, from the software they use to keep charts in.


Experience will vary widely from country to country, depending on insurances subscribed and on medical establishment. Unfortunately I dealt with half-illiterate doctors in country with public healthcare and they were too arrogant to accept feedback.


An aside but "Making Money" by Terry Pratchett references a similar sub-office of the Post Office in Ankh-Morpork--the Blind Letter Office. Amusingly, Lord Vetinari is able to assist with some as their addresses are mis-spelled vague directions rather than strict addresses.


People after 2010 can't read cursive? Really? So weird to me.

The examples in the article are crystal clear to me, no?


It’s not taught anymore. I’d be interested in hearing from someone who didn’t learn cursive if the example in the article is actually hard to read.


I half-assed learning cursive but was never really graded on it, and printed my whole life.

I hit "View all photos" and got this

- Mrs. Grace Dawn, 2110 Minsp (?) Church Drive West, Dr. Koven, Ken

- Nevada National Bank, W. Sahara Avenue, Las Vegas, Nevada, 89102

- M A Hill, Post Office Box 41, Whitesville, GA

- Knight and Associates, Park and Main, Butte, Montana (this one I guessed based on likely words)

- Twyla Sharp, HWC Box 7, Wyola, MI, 59089

These felt easy. I see some people's cursive that just looks like a ton of loops, with tiny squiggles differentiating the loops into actual letters.

I'm sure it's fast for them, but it's not accessible. And as a lefty, cursive doesn't flow well for me anyway, so I only print.


Yes, as someone who learned cursive but never really had good handwriting (and can't really write properly at all today), the examples seemed pretty straightforward for the most part.


> Twyla Sharp, HWC Box 7, Wyola, MI, 59089

That would be Wyola, MT. 5xxxx is not in Michigan (and neither is Wyola).


One of the things the UK's postcodes do that US ZIP codes don't do is they're alphanumeric but all the confusables are eliminated, so e.g. SS2 5QF exists, this means S52SQF, 5525QF, SS2SQF and so on are just misreadings of SS2 5QF - so both a machine and a (suitably knowledgeable) human can just correct them automatically.

The use of alphanumerics is much more common everywhere in UK systems than many foreign systems and I think they're almost always a benefit. Postcodes, National Insurance numbers (~ tax IDs), Map references, Car number plates ... In many of these systems the initial introduction of alphanumerics substantially predates automation, so perhaps other countries figured numbers would be easier for machines, but this was a very fleeting benefit.

You do need to eliminate confusables though, having 1I1Il and I1l1I both be valid and signify different things is clearly an enormous problem.


The pictures are probably all sample/training images to protect the privacy of the mail.

(Was the case on another article I read about this system.)


Well after 2010 it was dropped from most schools. So, yeah.


Just because kids aren't taught how to write cursive doesn't mean they can't still read it. Cursive letters are no harder to recognize than any other stylized font; the only potentially weird ones are "r" and "z".


In my experience as a regular user of cursive that is not true.


Conveniently right down the road from the [NSA Utah Data Center](https://en.wikipedia.org/wiki/Utah_Data_Center).


Why don't we have postal addresses with reasonable checksums?


A zip/post code adds redundancy to a state, town and often street, sometimes more.


That's exactly why I added "reasonable" to my comment.


With all these advances in AI, why can't we get automated handwriting recognition? I'm still trying to decipher my grandma's letters.


This page goes absolutely haywire when I try to read it with the text flashing and jumping up and down. But that's a cool curiosity.


Seems like a good use case for machine learning.


ML is used extensively for this. These folks are handling cases that are low confidence or Id imagine building labeled training datasets.


Yann LeCun pioneered CNNs for this actually. OCR of zipcodes for USPS. Early 90's.


Always wondered why USPS didn't adopt a system for writing zipcodes similar to the one used in USSR:

https://en.wikipedia.org/wiki/File:Stamp_Soviet_Union_1977_C...

Add: https://en.wikipedia.org/wiki/File:Russian_postal_codes.svg


Some might refuse to write 4s (vs U+1FBF4) and 8s like that as they require two strokes or retracing.


... what?

Can you explain how this is an issue at all and a point of concern for something like USPS?


The (arbitrary?) 9 segment number examples from the Russian Postal Codes that were linked and were questioned as to why the USPS did not adopt them are ambiguous depending on how one chooses to fill out the segments of the numbers. For example 4 may be seen as 9. 8 is slightly less an issue, but amounts to 1 segment away from 0 and 9. If a machine or any and all perfectly rule following humans would be writing and interpreting the zip using 10 different characters mapped to the Arabic numbers, then it would be more optimal to try to maximize the mutual distance between them such as an anti-reflected binary code, which would be an increasingly challenging exercise with more segments that could start at 7.


> For example 4 may be seen as 9

Except no, specifically because 9 has a slanted ending compared to 4. If some idiot writes the symbols whatever they wants instead of the reference then it's the the problem of the idiot.

And any way, most people would write it properly which means more mail would be processed efficiently and require less human intervention.


You think handwriting is easier to recognize than digits of a standard form and size?


No


There used to be many of these facilities. This is the last one. The demand has gone down as OCR has gotten better and hand addressed mail volume has decreased.


These facilities exists specifically for the edges that the ML models are failing on.


Indeed, and this is why they continue to downscale the number of people doing such tasks.


They need 250 more as per the article (scroll to end).


Just before the end is the likely reason for that: "70 hours a week".

This is the last such facility, after all the others have been closed, so needing more people is not incompatible with the fact that they've scaled down overall.


Is it then Fed to the NSA servers in Utah as well?


So here we are discussing how AI will soon kill us all or at least solve the big questions of science, and at the same time we employ 800 souls in Utah working 70 hours a week to decipher bad handwriting.


I suggest you watch Tom Scot mentioned in another comment and it might change your opinion. When OCR system was initially started 50% were done automatically. Today that is greater than 99%. Less than 1% are now handled by the Utah center and they it still going down (After closing rest of the centers already and Utah center is last remaining one).


Exactly my point! Machine takeover is further than you'd think, because of the long tail.


I think people underestimate the long tail of most problems. Even if it’s 98% good, what do you do with that last 2%? The same applies for self-driving and other computer vision related technologies.


> what do you do with that last 2%?

My plan when I finally purchase a self-driving car is to get 800 people from Utah to help me navigate for that last 2%


Realistically, keep running Utah until the next round of sweeping budget cuts comes, then cook up some reason to return/destroy the remaining 2% as "undeliverable" and shut it down. Good service always gets sacrificed to save 2%


I remember a patio11 thread about exactly this — once your operations reach a certain scale, even a 0.1% of cases problem takes a whole department to deal with


Shows that some things just can't be provided by the private sector and need a govt agency like the USPS. A private corp would have shut this down long ago and just required in their terms&conditions that people write in an easily OCR-able style, else the letter gets dropped without notice. Letting the govt take care of this gives people more freedom - for example the freedom to get old and have a handwriting that a machine has issues with.



“We are processing the requested change to your cookies settings” with a timer taking multiple minutes programmed in.

What a cancer website.


There are serious businesses working on malicious compliance now.


They have accessibility from https://userway.org/ "Representative Accessibility and Consulting Provider".


Wow. My ad blocker killed the cookie modal, I guess, but the site freaks out detecting that the videos that are supposed to auto play are not playing, and just keeps trying to reload them. Cursed.


If auto play video ends up being broken, I consider that a desirable bonus feature of the ad blocker.


I missed that completely. I scrolled the page and the modal just moved up the screen, so I could read the article without interacting with it at all.

Vile, and badly done. Hmph.


Inarguably malicious, whereas the usual cookie popup crap can just be (barley) passed off as complete UX design incompetence.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: