(评论)
(comments)

原始链接: https://news.ycombinator.com/item?id=41326604

法国建立了国家基本地址(BAN),这是一个结合精确 GPS 坐标的综合性全国地址数据库。 免费 API 支持搜索地址、地理编码或反向地理编码。 地方当局维护本地数据库并将其输入到主要 BAN 中。 数据来源于La Poste(相当于英国皇家邮政)和OpenStreetMap。 BAN 不包含更细粒度的细节,例如建筑结构、单个单元或内部布局。 尽管存在个人用户地址详细信息,但准确性会根据本地输入质量而变化。 法国用户注意到各个地址组成部分的结果不一致,并且缺乏门牌号码或公寓名称等具体信息。 虽然禁令提供了简化交付和紧急响应等好处,但由于当地市政当局的报告不一致而产生了限制。 实际住宅小区与 BAN 中相应条目之间的差异就是一个例子,导致用户的个人地址不明确。 相反,复杂寻址方案常见的国家(例如英国)的用户报告的重复地址问题较少。

相关文章

原文


On the other side of the Channel, the French government has managed to create the "BAN" (Base Adresse Nationale - National Address Database), a database of detailed postal addresses in the country along with precise GPS coordinates: https://adresse.data.gouv.fr/base-adresse-nationale

On top of the database they have provided an interface to view the data, interfaces for towns and cities to keep the data up-to-date, free APIs to search addresses and performing geocoding or reverse geocoding (https://adresse.data.gouv.fr/api-doc/adresse) and the data is openly licensed and available to download.

Feeding the BAN has been enforced by law, localities are required to put together and upload their "Base Adresse Locale" (Local Address Database)

The original data was obtained from multiple sources, including "La Poste", the French Royal Mail equivalent, and OpenStreetMap !



A cautionary example of how data meets reality…

My address in France is listed in the BAN… but only to the granularity of my street number (e.g., 123 Main St.).

Unfortunately, that number corresponds to at least 7 different structures, 5 of which are apartment buildings.

Of those 5 buildings, each has multiple stairwells with their own door and no line of communication between them—they might as well be separate buildings.

My particular building has 8 levels with 2 flats per level. No flat has a door number or letter, meaning I must say 'Nth floor, door on the right' to give directions to a visitor. And I could not receive mail until I affixed my name to my postbox on the ground level.

None of that is in the BAN as far as I can tell.

Finally, on OpenStreetMap, the coordinate for the the street number address in the BAN actually corresponds to an island in the street that happens to face a private road that enters the property. There is more than one entrance :)



This sounds like bad design by the property developer and a sloppy building authority. The first is corroborated by the lack of unit numbers. Who does such a thing?

The BAN actually only tracks down to the plot level, so I assume all your structures are on the same plot. From there on it is the building authorities job to check building plans and to enter the substructures into the cadastre, where they are usually lettered. It's the developer's job to mark the buildings and entries. Sloppy work, all around. So sad.



You could be right, but I think it's a little beside the point.

The challenge illustrated in the blog post is that it's practically impossible to build a really accurate address dataset since the real world is messy for the reasons you listed. Just like falsehoods programmers believe about names [1], you shouldn't put much faith in anything that claims to normalize addresses either.

As other commenters have said in the replies, my situation is not uncommon in Europe.

As they say, 'the map is not the territory.'

[1]: https://www.kalzumeus.com/2010/06/17/falsehoods-programmers-...



As long as it shows that your address corresponds to that plot of land it's still a perfectly accurate address dataset. Your address just kind of sucks. That doesn't make the dataset less accurate, just less useful.

Still a lot better than some other parts of the world though. In Asia you sometimes have addresses that boil down to the nearest landmark and a phone number for the mailman to call



> it's practically impossible to build a really accurate address dataset since the real world is messy for the reasons you listed

Different entities will have orthogonal needs when it comes to your address. First responders want a door, the post office wants a mailbox, assessors want a plot number, etc.



I don’t know what’s usual in France, but it’s usual in Germany for apartments to not have numbers. You have to put your name on your mailbox, and there’s no way to address something to someone who doesn’t live in the apartment. If you’re filling out government forms, you sometimes have to put in something like “third floor left side” so they know where you actually live.



Britain also has "dwelling designations" like "3FL" (third floor left) commonly used to describe unnumbered flats (which may well have numbers or not). I suspect this way of referring to flats is unofficial, but it is commonly seen on letters.



Costa Rica doesn’t have numbers on the buildings, and many streets lack street signs, if not names. You’ll have addresses like “50 meters north of the old church” or “behind the banana stand.”¹



Depending on where you are in France (especially places with lots of housing stock being older buildings), it's common (if not the norm) for there to be no unit numbers and to direct people to apartments by floor number / door position relative to stairwell.



That doesn't surprise me, same thing in Germany. However having multiple buildings with the same house number (without distinguishing letters) sounds like the much worse oversight here



Though at least in Berlin it's pretty common for multi-family houses to have a separate wing (Seitenflügel) or rear house (Hinterhaus) that are reached by entering the street door of the front house (Vorderhaus) and then exiting through a door behind the staircase into a courtyard before entering the second building, and at least in some cases each building has its own set of mailboxes, all with the same address.

I regularly have the problem that deliverers don't read my delivery note and don't listen to what I say on the intercom, and go all the way to the top of the front house before realising I'm in a different building altogether.



> This sounds like bad design by the property developer and a sloppy building authority.

This sounds like every day reality.

> Sloppy work, all around.

It's a system that explicitly relies on the cooperation of several independent entities. You were never going to achieve anything better than this.



> The first is corroborated by the lack of unit numbers. Who does such a thing?

The entire country of Germany for example. It's super annoying.

Although they have the decency to assign distinct numbers to stairwells and when you register where you live for administrative and postal purposes you give description at which floor and on which side the door is located.

The funny thing is that in Germany you have to pay TV license which is paid "per apartment". But since apartment doesn't have its own number, just street name, building number and freeform description then the authority responsible for collecting tv license fees doesn't know it a fee for this apartment is already being paid. So when you move in anywhere they always send you a letter so that you either start paying or provide TV license I'd number of a person living in this apartment who's already paying.



Would not the opposite be true? If you have to write your name out just so the mail can find you, you are less anonymous than if you just have a number that gets mail directly to your mailbox.



> the lack of unit numbers. Who does such a thing?

Everyone in Germany. Units are identified by the surname of the person who lives there. If there's more than one person living there, too bad, pick one or write them all.



Yeah wait how is it the BAN's fault that you don't have unit numbers, that's like complaining that you never receive your letters "just because" your house just fully doesn't have any street address and the post office needs to figure it out better without any involvement on your part.



That sounds like chaos. Who thought constructing multiple apartment buildings without any kind of sensible post code or address was a good idea? Sure, this being reality BAN does not apparently meet reality, but it does sound like someone had the opportunity to keep reality sane here, and they didn't.



In Finland in similar case, each stair well has own letter and each apartment has different number. So those are used always with the street house number.

Though the later case is bit messy with cross roads. As building can have two different addresses. Or same complex of multiple building have two different addresses for each building. With in my case one having A-C and other D-F stairwells... Oh, and numbers also are not restarted at least sometimes.



I live in Finland nowadays, and this system is nice.

I moved from Scotland where there are frequently buildings containing multiple apartments - tenements - there are there are two systems for the labeling of the apartments.

The first is the obvious one, "flat 1", "flat 2", "flat 3" (often this would be written after the number of the street - so flat six at number seven example road would be called 7/6 Example Road).

The second approach is the more physical layout. I used to live in "TFL, 7 Example Street". "TFL? Top flat - left side". You get "GFR" for "Ground-floor right", and similar examples. This worked really well if there were three floors to a building (top floor, middle floor, and ground floor) but the confusion got intensified if the building were higher.

There were times when you'd enter your postcode into an online service, ordering a home delivery for example, or setting up a new electricity contract, and you'd be presented with one/other of these systems. And broadly speaking it would always be the same. When I lived at TFL it was *never* called Flat 6, although I'd often enter it as 7/6 Example Street a time or two just to keep the posties on their toes!

To be honest most of the time the postal delivery people were smart, if I got mail addressed to "Steve, 7 Example Road" it would end up at the correct apartment. Either because the postal delivery person knew - they tended to have fixed routes - or one of my neighbours would do the decent thing and redelivery if it was sent to them in error.



> but it does sound like someone had the opportunity to keep reality sane here

What is "sane" about reality? People want a place to live, they don't care about government databases.



You're right. Since the postal worker knows his route, he knows my name. So snail mail works perfectly well. Same for the Amazon delivery person (took a few visits). Same for the local pizza place.

It's online address suggestion/validation/one-time deliveries that don't work well. E.g. Uber Eats and DHL drivers always require a phone call so that I can guide them along the final hundred meters of their delivery. I usually go downstairs and meet them at the curb.



Frankly, that just sounds like a fire code / building code issue. Are these "apartment buildings" legal for habitation, with actual legal separate apartments, and not some weird subdivision/subletting situation?

In every place I have ever lived, having a clearly marked addresses and door numbers for apartments is required by the fire code. If there's an emergency that requires a fire or ambulance response, smoke in the air, etc, then "Nth floor, door on the right" is not a good thing to be explaining over the phone.



> Are these "apartment buildings" legal for habitation, with actual legal separate apartments, and not some weird subdivision/subletting situation?

Yes. In fact the 'résidence' (the conglomeration of apartment buildings) is considered one of the nicer, more desirable, places to live in the city. In the US, each apartment would be called a condominium [1], i.e., most are individually owned and not rented out.

[1]: https://en.wikipedia.org/wiki/Condominium



In several countries in Western Europe there's hardly a tradition of apartment numbers in multi-apartment buildings. Instead the apartments are identified by family name of the owner. Or the main person living there. Or the person who used to live there some time ago. Or some guy backpacking in Asia and (illegally) subletting the apartment.



> GPS coordinates

*coordinates

There are four GNSS constellations, of which GPS is only one...... a statement that negates the fact ones position on Earth may be calculated using a variety of other means.

EDIT: In response to replies below; One isn't questioning the coordinate system (!), rather the assumption as to how they have been calculated.



In this context, it's not terribly hard to divine that they probably mean EPSG:4326 coordinates. I was going to comment that one of the ETRS89 UTM zones might be easier to work with, but on second thought the data almost certainly includes the DOMs if not the TOMs, so a global coordinate system is probably best.



The BAN provide fields `long` and `lat` which are WGS84, and also `x` and `y` which are coordinates expressed in "the appropriate local CRS" (without much elaboration on what that would be).



That would be the French national grid system, no? The UK has the ordnance survey grid which is based on the OSGB36 datum. I'm pretty sure France will have a similar national datum to create their own local grid coordinates as planning and building works needs to be done in a more accurately aligned local datum than WGS84.



For mainland France it's reasonable to assume the French national grid. But what about French Guiana in South America or Mayotte in Southern Africa (an island north of Madagascar)?

France still spans the globe, with many places treated as equals to the French mainland.



Not to mention that “latitude” and “longitude” cannot uniquely describe an address, regardless of the datum or ellipsoid. Maybe that is not the intent of storing the coordinates. Lat/Lon says nothing about floor number in a multi-story apartment.



Fun fact: the word Νερό (nero) means water in greek. The actual meaning is fresh (I think it's the source of the word "new" too). It turns out, that many years ago you meant something else than fresh water by saying just water, so you have to be specific when you're talking about fresh water. In ancient greek water is ὕδωρ (hudr, think hydro, water) and fresh water is νεαρὸν ὕδωρ (neron hudr). Sometime in the past, the ancient Greeks were sick of saying 2 words to say water. So they dropped the second one.

Something similar happens with GPS coordinates. People are just saying GPS when they mean coordinates. even though the logical thing to do is drop the GPS (neron) and just say coordinates (hudr).

Personally, I think that language is just a bunch of symbols that have no real meaning. Each symbol means something only in a context, no matter how broad or specific. I would argue that it doesn't matter which word is more logical to use because logic is just a part of the context.

But you are right.



Australia is similar, howeve, irrespective of how perfect your national addressing standards are, companies ingesting this data providing any sort of to-the-premise service still have to mash and clean and dissect it to fit whatever legacy system they are running.

I am aware of one utility provider that is locked into a custom network modelling solution that was officially sunset in 2014 and employs 3 ftes to manually create and delete addresses because the old address import tool broke.



So many Australian sites use some data source that has an old name for the building I'm in, and sites are so convinced their address databases are right that I can't do anything about it! Mildly frustrating



Our previous apartment was listed under the wrong postcode. Annoying for Uber Eats because they would get lost.

Our current building is one of those 56-66 style buildings. Different service use a different number (e.g. postal is 56, gas is 58). We've had a few cases where our address doesn't match so the system rejects us. And when I vote I have to read which number they have upside down!



In the US, I had a family member's address change zip codes (approx similar to larger area postal codes) and associated city.

It took a surprising amount of time to cascade through systems, as in years.

I think we're at +8 years now, and Google Maps still has the old zip and city. Which means many websites do too.



Very cool. Nice effort by France.

For a while I played around with that kind of data here in Belgium, it's not easy to get it all standardized and "usable".



The big difference is that US postcodes describe very large areas. A 5-digit US ZIP code describes a town or neighborhood, with on average 8200 people living in each ZIP code.

Most European postcodes are far more precise, often describing a single street, part of a street, or even part of a building. Postcode + house number is usually enough to uniquely identify a mailbox. For example, in The Netherlands on average only 40 people live in each postcode. That makes the dataset far more valuable for geolocation.



They are public, but the post office changes the last 4 digits every few months so there is no point in telling anyone what yours is. These days the post office can look up your street address and give you all the information they need - which is an 11 digit bar code good for the next week.



Based on my personal experience, I really doubt that the last 4 digits of the ZIP+4 are changing more often than once per decade or longer. I could see the delivery point of the 11-digit code changing every few months, but you are already aware of that code system so it is not simple confusion between the two on your part.

Could you provide more information or a source?



20 years ago they changed all the time. Wikipedia doesn't mention this though. These days the post office can read the street address via computers and get the 11 digit code they need, so I suspect they don't need them. (for PO boxes the 9 digit code apparently doesn't change)



Some context, for people not located in the UK - A full British postcode typically aims to cover around 15 buildings (sometime a single building, sometimes a street of 50 houses). This is in contrast to many other postal code systems which cover relatively broad areas).

Or to put it another way -

UK - 2,643,732 codes, 1 code per 25 people,

USA - 41,700 codes, 1 code per 8000 people,

Germany - 8,200 codes, 1 code per 10000 people,

This means that post-codes are often used as a proxy for an exact location - e.g. if I am going to visit a relative, I can enter their postcode into my sat-nav, and be confident that most of the time I'll get to within a hundred meters of their location.

This doesn't work so well in rural area or on large estates where the access point may different from the location, leading to places sometimes advertising a different postcode to put into your sat-nav (e.g. of where the site entrance is) to that of the location itself.



8,200 in Germany is way too low, and I'll add some fun facts. According to Wikipedia 30,000 of the theoretical 98,901 are currently in use. The number of people per postcode does vary a lot, from zero (no one lives in a company that has its own postcode) to none living in a demolished village of Billmuthausen (right on the inner-German border) or the two people living in a district that had no postcode until the problem was fixed in 2015. Yes, they forgot Gutsbezirk Reinhardswald (a quarter the size of Frankfurt/Main) which is almost all forest but has a forester hut with two people. There are even four Austrian villages that also have a German postcode in addition to their Austrian one, and a swiss one. There are even still four-digit postcodes with no five-digit update in use: Feldpost, the Germna army postal service.



Some larger retail stores in Germany ask you for your postcode during checkout, presumably to learn a bit about their customer base. I don't mind telling them mine, there are about 16K people with the same postcode. But I'm pretty sure I would not tell them if I was one of the two forest rangers in Reinhardswald. (And yes, I do pay cash whenever I can.)



Interesting, is the German postcode not used for transaction validation? I know the American payment processors definitely use ZIP codes for validation - see anecdote 1.

That said, there are definitely situations where the payment processors don't require the ZIP code - see anecdote 2.

Anecdote 1: When I worked in food service as a kid, I used card terminals that connected directly to a phone line. I remember a couple of times when I entered the ZIP code incorrectly - the card terminal would print out a receipt with an angry message saying the transaction got rejected. So, I know they were using the ZIP code to validate the transaction.

Anecdote 2: With those same card terminals, you could skip the ZIP code and it would run the transaction as usual. But, my manager always told me not to do that. Maybe I never asked him why, or maybe I forgot his answer. Regardless, I don't remember why we he required us to enter the ZIP code, even when it didn't seem to be necessary.



The answer to anecdote 2 is probably that if the seller chooses to skip validation measures on the transaction, then they become liable in the event the transaction is deemed fraudulent.



ZIP codes are used as a weak "something you know" factor in payment card processing.

The card is (for card-present transactions) "something you have". And the ZIP complements that. ZIP code is optional, but the merchant gets a data integrity score back from the network ("AVS/address verification service response", from no match to full match), and can accept/decline the txn at their discretion.

Because it's optional and at merchant discretion, all it really does is give the merchant some additional ammunition when disputing a chargeback. And of course to build a demographic database.



> is the German postcode not used for transaction validation?

No. The only time I have ever been asked for a post code was when a petrol pump in the US demanded my zip code. I have no idea what it meant, I just put some random zip code for the general area I was in and it was accepted. I've never been asked for my post code in Europe; I can't speak for the whole of Europe though, just UK, Spain, Ireland, Portugal, France, Belgium, Netherlands, Italy, Germany, Poland, Denmark, Sweden, Norway.



This might help explain in-part the discrepancy. Perplexity.ai[1] also says 8,200 German postal codes. I set Claude 3.5 Sonnet in the LLM settings on Perplexity but it looks like it might use a ChatGPT model for the initial search of sources? At least we can see what it is sourcing to fetch the value of 8,200. Interestingly, asking Claude 3.5 Sonnet directly at claude.ai returned 16,000.[2]

1. https://www.perplexity.ai/search/how-many-german-postal-code... 2. "There are approximately 16,000 postal codes (Postleitzahlen) in Germany. These five-digit codes cover all areas of the country, including cities, towns, and rural regions.

To break it down a bit further:

1. The first digit represents one of 10 postal regions. 2. The second digit typically represents a sub-region within that area. 3. The last three digits identify specific delivery areas or post offices.

It's worth noting that the exact number can fluctuate slightly over time due to administrative changes, urban development, or postal service reorganization. However, 16,000 is a good approximation for the total number of German postal codes.

Would you like more information about how the German postal code system works or its history?"



I felt your reply was too harsh, but after few moments I realized that I instinctively think the same I treat any output from ChatGpt as garbage until checked in other sources. So effectively, not worth looking there in the first place.



I think that it's often easier to verify an answer than to find an answer with nothing to go on, so perhaps not entirely garbage but certainly not reliable.



Dutch post codes actually do specify the street.

It's four digits and two letters. The digits cover an area (can be a city, town, neighborhood) and the letters cover the specific street or part of the street. Technically, they cover a range of house numbers, which in 99.9% of the cases is (part of) a street.

So just like in the UK a postcode is enough to get you pretty close. A postcode and house number will get you to the front door.

To get back to the article. I always feel like the UK manages to take privatization of public services to a next ridiculous level. This being a good example.

Another one is the rail network where the company that owns all the infrastructure and is responsible for maintenance (Railtrack) was fully privatized and even stock listed. This of course did not go very well in as far as actually properly maintaining the network. Resulting in it being nationalized again where now Network Rail is responsible.

In the Netherlands the company that owns all rail infrastructure and is responsible for it (ProRail) is a private company but with just the government as a shareholder. Meaning it is still effectively a public company, so things did result in such dire conditions as the UK.



the problem with UK privatization is the same as with California PG&E ... it's private in name, but the incentives are all bad.

there was (is) no point for optimization on costs as the profit was a fixed percentage (so it ended up quite the opposite) instead of a price cap. (ideally the cap would be a simple formula based on input prices, to at least make the lobbying transparent. sure, this also has a built in profit percentage, but the important difference is that the profit is not fixed, so the private company is incentivized to push the costs down.)

see https://www.noahpinion.blog/p/energy-bell-the-sketch-of-an-i...



In some cases, your zip+4 is uniquely your address, too. My townhouse was a new development and after complaining for over a year that I wasn't able to sign up for Informed Delivery, I was assigned a new unused +4.

That said, most people don't use the +4 when getting directions or the like, it's just used for postal service.



> it's just used for postal service

Even then, to my understanding the USPS has for some time now not relied on zip codes at all. They have a really good address database, they match on that, and then stamp the mail with a routing barcode at the origin post office. The zip code is extra, mail flows just fine if you leave it off.



Hence why a house number and postcode constitutes a complete address in the UK, we’ve sort of already got What Three Words with “a number and 5-7 characters” - not quite as catchy though



We needed an ambulance off road in the middle of Richmond Park where a postcode would also not help. We didn't have WTW either, which they asked for and would have helped immensely.



I know - my van broke down recently, had the same experience even though I could describe the location exactly by intersection of roads, or grid reference.

As I had a smartphone they did at least have a link I could click which would give me my w3w location, which I had to read back to them.



w3w has been sold to most UK emergency services to deal with that exact scenario. I know Richmond Park well, and it's hard to direct anybody to anywhere in it using addresses, so it makes sense.

The problem is that w3w is privately owned and has multiple issues with it, as well documented elsewhere.

They could invest in a solution that allows for an OS grid reference to be discovered by sending you a link (a bit like they do with w3w), or some other open (already paid for) reference. That still has limitations if you don't have a smart phone with GPS on it, but I'd argue it's better than what they have right now.

Of course none of this solves for the fact the most useful location dataset in the country is the PAF, and we can't use that without spending a small fortune on licensing it.



Centimetre, and after the next big earthquake are all numbers off, sometimes even by several meters. Now what you do? New addresses for all, or wrong numbers to new buildings?



> Now what you do?

When the ambulance arrives wave your hands and say "over here!". So they can do the "last several meters" of navigation by homing on your visual presence.



In the best case your GPS is off by far more than the worst case GPS. There are GPS receivers that can get you to within 2cm, but they cost thousands of dollars and are not used in phones.

In the context of navigation that is good enough - if you are within 100 meters you can look to see your destination.



Lat/long coordinates and metres are actually linked quite closely: the metre was originally defined as "the arc from equator to North pole is defined as 10,000 km". That is, 90 degrees is 10,000 km.



And if the French had had their way, we’d use grads not degrees and latitude would instead be 100 grads per 10,000km, so each grad of latitude would be 100km.

That kind of sanity was, of course, unacceptable to the rest of the world.



The French were really into decimalization for a while. They tried decimal time (10 decimal hours per day, each 100 decimal minutes, each 100 decimal seconds), and a new calendar with equal 30-day months (the extra days at the end were national holidays, in September in the Gregorian calendar). Also 10-day 'décades' instead of weeks.



I noticed that W3W showed up in a number of places, typically on advertisements, at least around London. Is it used universally across the UK? I don't know of anyone using it in the US aside from a few enthusiasts. I had forgotten about it until a few weeks ago when we were visiting London.



> sometime a single building

Fun fact- some postcodes cover only a fraction of a building. There’s buildings where 3 postcodes are used. Same street number, same main entrance, but different post codes.

Edit: A 2-postcode building example is “M3 7GW” and “M3 7GX”, both go to 55 Queen St, Salford.



Apparently the worst case in the opposite direction is the University of Warwick, where a single postcode (CV4 7AL) covers 5000 individual residences (so probably about 5000 people given these are likely to all be student rooms, and sharing a room is uncommon in the UK).



Though tbf (assuming they haven't changed it since I left), Royal Mail's responsibility stopped at the post room. The individual residences were delivered by UoW staff using a pigeon hole type system; anything larger than a letter and you had to go to the post room to pick it up.



US 9-digit postcodes ("ZIP+4") were introduced decades ago (1983?) and are publicly findable online given an address.

The US also has a full 11-digit address code that is printed (by the USPS) on mail in a bar code when you mail a piece. This should take the mailpiece to a unique address.



You can put Ireland at the top of the list, one postcode for every address, every house has one, every apartment has one, every building has one, even some old ruins have one.



> Irish posties mean you don't actually need them

I worked as a postie for a few weeks as a Christmas job when I was at Uni a looong time ago.

My GF used to write to me regularly (yes, writing letters was a thing back then), we came up a nice scheme: instead of using my actual address she wrote to a made-up non-existant address but with a valid postcode ("501 Any Street, Town, AB1 1AB" on a street with only a dozen houses) that was in one of the streets I'd be sorting letters for/delivering to.

Worked like a charm, would find her letter waiting for me in the pile to be sorted when I rocked up at the sorting office at 5am.



Case in point, going to a funeral the post code for the crematorium was for a 2km stretch of road, and going by foot I realised my folly and so had to run to make it time.



For what it's worth in the Netherlands you have about 1 postal code per 21 addresses. Typically one code is a street or the even/odd half of the street.



As he points out, this was a profoundly stupid mistake made when privatising Royal Mail. It would have been trivially easy to do at that point, but now it's a lot harder. If the government decided that it does want to do this, it can't just pass a law that says "the PAF is now free" without paying hundreds of millions of pounds in compensation to Royal Mail. That's quite apart from the ongoing costs of maintaining the data. At a time of cuts of budgets this would be a hard sell.



Another part to this is that there's a certain amount of cooperation between Royal Mail and councils street numbering and naming. Councils are the first authority over new streets/locations, changes like a property being split or merged (i.e. landlords converting to a property of multiple occupation and not telling them for various reasons, and then residents have issues getting post), residential/commercial, etc, and then that gets passed onto Royal Mail to update the PAF. If there's an issue with an address you've got to check with the council first, so there would be some good fit for centralization there.



> If the government decided that it does want to do this, it can't just pass a law that says "the PAF is now free" without paying hundreds of millions of pounds in compensation to Royal Mail.

You can pass the law, get sued and pay whatever the PAF is worth. But that's just.. fair? The govnerment spent 5 mio just for a survey concluding that it's impossible to recreate the PAF. So hundreds of millions sounds like a good deal.



Just to be clear: UK Parliament is sovereign. If it passes a law forcibly legalising it, the privatised Royal Mail can sue the government but would need to find an international treaty obligation to win. Even then, if Parliament flagged it and said "we're ignoring this treaty in this case" then the courts are bound to the law, not treaty obligations.

If it has knock-on impacts in other areas, it's hard to say, but that's separate to the law.



International courts - especially US ones - have de facto overrides on UK sovereignty in many domains.

If Royal Mail put the asset into a US subsidiary then US courts must protect it.

Regardless of what the UK domestic authorities say, very few entities would be willing to risk violating US copyright laws.



They can try to protect it but anything inside the UK is subject to UK law and no other law. As I said, there may be practical concerns (read, consequences) for a course of action but this is separate to the domain of law.

Inside the UK, if a law explicitly counteracts, say, an international treaty, the British courts MUST find in favour of British law.



> anything inside the UK is subject to UK law and no other law

This is true in theory but not in practice. If you’ve been sanctioned by the US then you won’t by able to get a bank account in the UK. Even if you’ve not violated any UK law.

The Americans can give a British bank an ultimatum between dealing in US dollars and dealing with a particular individual. Every bank will always choose the former.

Similarly anything with US assets is subject to the US court system any its interpretation of copyright. Naturally this means US companies need to obey its rules.

But also given the reach of copyright law, so do foreign companies that interact with US ones. Visa, Mastercard, Google, FedEx, and Stripe can’t do business with someone who openly violates US copyright.

So perhaps a local council using self hosted services could use nationalised data but that’s about it.



Article 1 of Protocol 1 of the ECHR covers exactly this, so the Supreme Court (and ECtHR if it came to that) would probably find in favour of Royal Mail if this were to be done without compensation.



If it were done by an Act of Parliament, no they wouldn’t. If that Act explicitly overrides external concerns, the Supreme Court must find in favour of the Act and the ECHR would have to find against it, which the UK Parliament is free to ignore, if it so chooses.

As I point out, this isn’t “free” because it can invite consequences from other parties. But that’s a different domain to law.



No, I'm pretty sure this one was incompetence. It came at the same time that the government was going all-in with open data in other areas, and this was a really stupid omission.



> profoundly stupid mistake

Surely the PAF formed part of the sale price when privatising Royal Mail?

So if you removed it before selling (at a lower price), or you buy it outright after, is there really that much difference making it a profoundly stupid mistake?



I doubt they thought about it at that level of detail. I think it was just sold off on the cheap through a share offering with an initial offering[0] of underpriced shares? There was some kind of scheme where a private individual could buy a small number of shares before they went on general sale. Could be wrong though.



To never have given would have simply required the government to say "this is not part of the sale."

To take it would likely require either lengthy court battles or legislation. Given the priorities of the Labour Party, the latter isn't likely to happen within the next 5 years (when they'd have to add it to their manifesto).



Privatisations, like most IPOs, are deliberately under-priced, and I'd very much doubt that the valuation of Royal Mail would have been affected by adding "The universal service provider must maintain the postcode address file and make it available under the Open Government Licence" to the Act



> it can't just pass a law that says "the PAF is now free" without paying hundreds of millions of pounds in compensation to Royal Mail

Why not? Parliament has the ability to make whatever laws it wants, no?



In the US, the Fifth Amendment of the Constitution says the government cannot take private property for public use without providing just compensation. I don't know whether any similar right exists in the UK.



"without paying hundreds of millions of pounds in compensation to Royal Mail" < the Royal Mail makes about £3m/year in profits from selling the data. It would cost a lot less than hundreds of millions to bring it back into govt



They made about £18m in revenue though. It depends who'd be taking on the costs. If they still need to maintain the PAF (they need it themselves), then it's the revenue that would need replacing. The best solution would probably be to say that the OS will take over the maintenance (presumably funded by central government if it's going to be open data), and then Royal Mail will have access to the data and can be paid a lot less in compensation.



The government gave up its property rights when the Royal Mail was privatised.

Taking it would be like taking back a Rolls Royce factory, BP oilfield or other formerly state owned asset.

The state could nationalise the asset in theory, but that’d involve leaving international treaties and causing foreign copyright issues.



I have no idea what UK law is. In the US the data itself is public domain, but the compilation is of data is copyright. Maps commonly would intentionally have errors in to detect copying - the error is creative work and so a copyright violation to copy so if someone copies your map you can sue them for copyright violation for not just the errors but also that compilation. If you take someone else's map and then use that to create own map off of (thus finding and fixing the errors) it is legal, but that is as much work as just creating a map from scratch.



> compilation is of data is copyright

Not by default, at least in the US. The database has to actually be more than just a compilation. It's not a high bar to clear, but it's there. Europe and the UK have the "sweat of the brow" doctrine however.



> Why can't it be taken for free?

It can be, but that has unknown long term effects. If you do this it shows everyone your government cannot be trusted and so other good ideas will not happen because people cannot trust the government. We probably do not agree on what is a good idea so I'm going to leave this vague - whatever your political side there is a good idea that is suddenly unworkable because the government cannot be trusted to hold their end of the deal.



> it can't just pass a law that says "the PAF is now free" without paying hundreds of millions of pounds in compensation to Royal Mail

Parliament absolutely can, legally. The issue is that it’ll set a bad precedent that’ll get brought up by the buyer the next time the government want to privatise something.



> Parliament absolutely can, legally. The issue is that it’ll set a bad precedent that’ll get brought up by the buyer the next time the government want to privatise something.

Great, maybe they'll be more wary of taking advantage of this kind of blunder if they can get corrected.



> The upshot of the research then, is that building an accurate database is really hard. OS concludes that it would have to check the 4.2m bad addresses manually to make its PAF-less database a viable dataset that would actually be useful.

The secret to the Royal Mail's success with the PAF, and the reason why only the Royal Mail can maintain the PAF, is that the Royal Mail has people walking and driving to all those delivery points six days every week.

Compare the Freedom of Information requests to Royal Mail from OpenStreetMap contributors concerning the locations of post boxes, which were refused ultimately because that information was handled only by local sorting offices.



Maybe they can use the TV detector vans used for TV license enforcement to collect the data, if they're already surveilling every single building in the country on a daily basis!



The simple solution here is a threat from the government to Royal Mail.

Give us your postcode file for free, or we will simply make up a new numbering scheme, send an address card to every house telling them of their new number with their next council tax bill, and postcodes will become a thing of the past.

The new numbering scheme will be unique to each house too, and have a check digit so the number alone is sufficient for 3rd party logistics companies like Amazon to use it for deliveries.



Every property already has a UPRN (unique property reference number). If you go on a council website and find a recent planning application it will be linked with this UPRN in the council's database. If I ever want to find a postcode I go to the find a planning application map and look it up there. I've not checked this in England, but it's definitely the case in Scotland. e.g. here's a random example; the entry for St Mungo's Cathedral in Glasgow:

https://publicaccess.glasgow.gov.uk/online-applications/case...



To have a unique id for each house is neat but I think there are loads of situations you’d have to account for so that there isn’t any ambiguity in the assignment of unique ids. If any ambiguities exist inevitably you will have exceptions in the system which defeats the point. For example -Subdivision of a lot. -Joining of lots. -You said every house… what about two houses on the same lot? -What about apartments buildings? -What happens when one or more houses are demolished and an apartment building goes up? Etc etc

I work in manufacturing and this sounds a lot like the problem of part numbering, and let me tell you, it’s not a trivial problem and the company I work for thought it was and got it wrong.



entity resolution is hard everywhere. Because the world is dynamic, but the common understanding of "entity" is a static object.

and the only perfect description of the world is the world, just like on a more trivial scale the only perfect description of what a piece of software does is to run it and see what it does.

So the best I know is to find a level of abstraction that captures enough stability to be useful, with enough flexibility to enable the classification to adopt.

In math, phylogenetic trees might be an example; think Dirichlette processes and exchangeable stochastic processes.



I'm unable to think of any reform in British history where 'throw everything out and start again' had successful outcomes. The British state runs on two principles: maximum effect for minimum effort, and the Ship of Theseus.



When did this almost Reaganite sentiment ("I'm from the government and I'm here to help") make home in the UK? I know it's not recent: I remember similar arguments coming from the No2ID camp in 2005 at-least.



For ID cards specifically most of the hostility was towards Blair's specific implementation which had a wide-ranging database that pretty much everyone and their dog in the public sector and beyond would have access to. While the arguments are perhaps a bit weaker in the modern day where the government taps the internet backbones and surveillance is a major category of business model, there were definitely good arguments against Blair's proposals that weren't necessarily applicable to ID cards in general.

I don't think it's necessarily Thatcherism that made people like this, just a slow erosion of trust that the government has the competency to carry out the tasks of a modern country that's accelerated as time's gone on. Anecdotally Liz Truss's episode as Prime Minister seemed to be the final straw for a lot of people's goodwill towards the government.



Quite a lot of it is Reaganism, via Thatcher. Probably dates from the Winter of Discontent.

It's not entirely without merit, but only because there's a tendency to drastically underfund and micromanage state services. And things like the Post Office Horizon fiasco do not make the government look good here.

On the other hand GDS is excellent - but that's almost entirely as a result of staff professionalism, rather than being driven by whichever ministers had the leadership of the civil service.

An odd outcome of the ID discourse is that we now have an extremely high tech biometric identity system .. but only for immigrants.



No. The Post Office is not a private company, it's a public limited company with the government as sole shareholder.

It was changed from a government department to a statutory corporation in 1969. It was then changed to a public limited company in 2000.

Furthermore:

- Post Office Ltd owns and runs Post Office Counters Ltd which runs the post office branches. This is the company that uses Horizon (since 1999)

- Royal Mail delivers mail to addresses, and owns the Postcode Address File. Royal Mail was separated from the Post Office and privatised in 2013. It has never used Horizon.

Horizon is an EFTPOS/accounting system, nothing to do with mail delivery. It was introduced to the Post Office in 1999 after Fujitsu/ICL were originally commissioned by government to build an accounting system for the Benefits Agency, and it was so awful and buggy the Benefits Agency rejected it, so the government asked them to retool it for the Post Office.



It's the other way around here; the Post Office is effectively a government department, cosplaying as a commercial business. It has never posted a profit. It's up to the government to bail it out, every time. It's controlled at arms length by a body called UK Government Investments (UKGI) who crack the whip at it and try to ensure "value for the taxpayer".

The rest of the UK government is capitalism on stilts, and is forever outsourcing everything to the private sector. There was a scandal when the outsourcing firm Carillion went bankrupt - we learned that the cleaners in Parliament were under four layers of subcontracting - i.e. four sets of middlemen taking a cut between the government paying for cleaning Parliament and the people who actually do the cleaning. One of those middlemen was Carillion, which had just paid £79m of dividends to investors and then collapsed with £7000m in liabilities and £29m cash. That's because capitalism is perfectly efficient, and it's not just a bunch of crooks cooking the books to appear to be perfectly efficient, right?



There's certainly been distrust/mild distain for the govt in Scotland, Wales, and The North since Reagan's gender-swap, Thatcher, for broadly similar reasons Reagan is maligned



I'm saying this as quite a strongly left-wing person. I am very much in favour of competent government intervention and regulation of markets. But the current government, probably since Thatcher, has shown themselves to be incapable of delivering large-scale national projects.



Take the aircraft carriers for example, we’ve currently got 2 but only purchased enough aircraft for half of one.

Whether you view this as a mistake of over investment or under investment, it’s clearly a mistake of some sort.

See also HS2.



True, but it's specifically the modern UK government - with its penchant for outsourcing jobs to ministers mates and bloated contractors - whose competency at large scale projects I dread.



so modify the law to deprive an owner of their legal property which was given to them by the law?

Not sure that's a precedent I'd want set in a common-law country, and not sure that would hold up to judicial review under common law.

The government made a bone-headed mistake when they included the postal data as an asset in the sale. The solution is for them to admit their mistake and pay for it. It's fiat money anyway, so it doesn't really cost anything. Having them abuse their government power to cover up their mistake is not an approach I endorse.

Not that this hasn't happened before, think postal scandal or yesterday's comments on the Hawke and Curacoa https://news.ycombinator.com/item?id=41285275



It would just be a change of law or regulations. Governments change these all the time, and sometimes it costs people or businesses money.

In any case, nationalisation has a long history in the UK, so it would hardly be setting a precedent.



I'm in favour, but that leaves RM holding a database of non-copyrightable addresses.

One way or the other, a private asset must be either nationalised or compelled to be released.

Gradual renationalisation of the rail network was in the manifesto. That's not particularly contentious, as rail franchises have fixed terms. But the manifesto is all about steadying the ship, and militant nationalisation risks spooking investors, so whether the government has any appetite to nationalise anything by fiat is questionable.

Nonetheless, there's public support for renationalisation; and, for such a low-value asset, this might be a nice test of the waters.



The postcode-to-coordinate data is now freely available as "CodePoint Open"

So there's already data for people who want to know postcode AB10 1JL corresponds to the area around 57.14677,-2.09873

The PAF is a more detailed data source, as seen in https://www.royalmail.com/find-a-postcode which can tell you that AB10 1JL specifically covers the addresses

102-104, Union Street, Aberdeen

82, Union Street, Aberdeen

Timpson Shoe Repairs Ltd, 86 Union Street, Aberdeen

Smart Mobile, 88 Union Street, Aberdeen

92 Union Street, Aberdeen

98 Union Street, Aberdeen

The PAF is useful if you want to provide a "quick address entry" option on your website - and to validate address data. But if you just want postcode-to-location conversion, that info is already available.



Thanks for the clarification. I’d forgotten that ernestmarples only offered postcode-to-location lookup: it was a long time ago.

I suppose this is encouraging! It shows that the forces of openness are gaining ground in this battle.



This also leads to some very interesting issues, as third parties who automatically ingest the data have a habit of just reading the docs and making the wrong assumptions about what it means in reality.

One example I often encounter myself is Google Maps trying to geolocate my address (city, street name, house number), and then reverse-geolocate that into my postcode. Which sounds like it would work - until you realize that the postcode polygons can overlap. I live in a building where (roughly) each floor has its own postcode, so whenever I try to fill in my address on a website which uses Google's API, it'll "helpfully" auto-fill or "correct" my postcode from 1234AB to 1234AZ. It'll essentially pick a random postcode, because all of them share the same coordinates!

That's Really Really Bad, because the postcode plus house number combination is supposed to uniquely identify a mailbox: it's only a matter of luck that the house numbers aren't reused in the set of postcodes used for my building. They could've just as well reused the numbers at the individual building entrances...



This creates a very special Dutch thing —- my neighborhood had the roads on the map before the map itself was updated to show landmass instead of the body of water.



At least the Uk has the the defence that postcodes are 60 years old and that the legacy cruft that comes with that is part of life.

Meanwhile Ireland introduced Éircodes less than 10 years ago, chose an opaque format that uses a central database that you have to pay for access to for anything more than a handful of lookups, only covers homes (so you can’t give an eircode of a park, or a walk). It’s pretty much what you’d expect to be designed by a modern government.



Postcode data is already freely available. You can even get coordinates of where the properties are. What is lacking is the actual addresses within a postcode.



I’d expect a modern government to design something as clear and well regarded as the GDS stuff in the U.K.

I’d expect a corporation like ibm etc to design the total mess we see with any large project



Transport for London is a pretty tightly run ship. Only capital city in the world that doesn't receive operating subsidy for its public transport. Not that that is a good thing necessarily as the tube is expensive to use relative to Paris or Berlin but a pretty impressive achievement considering the ancient complexity of the whole thing.

Scotrail is run by the Scottish government and has been steadily electrifying the Scottish rail network and because of the slow and steady nature of the work, between them, Network rail and the OHLE contractors they have got the cost for this down to 5 times less per km than typical UK costs previously e.g. the great western main line.

The moral of the story is get good people, give them stability and a clear goal and they will do great work. It doesn't really matter if they are working for the government or the private sector.



Well, how many buildings, and of what sort, varies enormously, but yes it won't be a whole town or region.

Most of my street is a single post code. Once upon a time it was a street of single family dwellings, so that's maybe a 3-4 dozen homes, but this is a city suburb so densification means some of those homes were modified and cut up to form flats, one large family home becomes six smaller homes - and some were purchased, knocked down and replaced by buildings which don't look out of place but aren't what they were before. I live in a purpose built four storey block, but it's designed to look superficially like a big house, the bottom floor is below street level (it faces out over the hill), the top has only loft-style windows at the front like somebody did a loft conversion.

It's all still one postcode though, so I share a code with maybe 100+ households. Recoding is disruptive and it's not really worth it, so they mostly don't do it.

Remember for actually delivering the post the postcode is just a convenient human readable part of an address, the machines (with occasional human help) turn any arbitrary address into a unique destination code, and then that's literally barcoded (albeit not in a code you're used to from like UPC etc.) onto the post. So for the Royal Mail the postcodes not being as descriptive as they were fifty years ago isn't a big problem.

Take some mail you've received, preferably over several days and study the outsides carefully. Two fluorescent orange bar codes have been jet printed onto the mail during sorting. The upper code is "just" a temporary unique ID, every piece of mail in the sorting system is issued a code, when they run out they start over, this helps with debugging and statistics. The lower case is in some sense the successor to the postcode, it'll be identical for every item delivered to the same address and distinct for other addresses. In fact it's encoding the "Delivery Point" which is what PAF handles, the location to which the Royal Mail employee delivers mail. https://en.wikipedia.org/wiki/RM4SCC

The use of these "real" postcodes also enables the Royal Mail to more readily accede to impractical "vanity" postcode requests. If the rich people in this part of Dirt Town think they ought to have postcodes from the adjacent and posh sounding Upper Niceton, RM can allow that, because in reality their teams are working from the purely numeric code which will still treat all these new "Upper Niceton" homes as being where they actually are, in Dirt Town.



The British mail addresses are pretty interesting. We quickly learned that, as you say some postcode have just one or two houses, which may not have numbers, but names.

I'm sure there is a "falsehoods programmers believe about addresses" somewhere.



Ireland can have:
  Foo House
  Townland
  Large town somewhat nearby where the mail comes through but only tangentially near the actual house
  County Bar
Where Townland is optional.

There's a bank address in my town: PTSB Kennedy Road Navan Co Meath

Kennedy Road is about 2 blocks long with ~ 30 shop fronts, and there are numbers on all but one of them.



> Where Townland is optional.

Not really optional in most cases if you're not actually in the "large town somewhat nearby". I would say the large town part is more optional.

You're not going to get post delivered to "Lakeview, Cavan, Co Cavan", but you should be ok with "Lakeview, Killeshandra, Co Cavan".



Kilshandra is a town, the townland for Lakeview would likely be "Portaliff or Townparks". Though to be somewhat fair, Lakeview in Kilshandra is really only unique vs things like "Pond View", "Lough View" or "Yet another body of water view".

In Meath, there's a House address near Garlow Cross where it's Foo House, Johnstown, Co Meath, but Johnstown is 7km away or so.

For those who have not been near there -- It's karst topography with basket of eggs hills where the water table is above ground in many of the valleys.



Amazing. For a e-commerce site I argued that we would save ourself a lot of trouble by simply making the address field one large text field, rather than attempt to making a form that would work for every country and city (looking at you Mannheim).

But apparently that would make data analysis to complicated.



There is a middle ground and some common patterns that can help.

The address field names are fairly standardized[0] and Google has an open dataset (used by Chrome and Android) describing which countries need which fields[1].

I have an older PHP library[2] and a newer Go library[3] that build upon this, while crowdsourcing fixes (since Google hasn't updated their dataset in a while). The Go library allows me to serve all address formats and state lists in a single HTTP request, which can then power a very fast JS widget.

[0] Initially by the OASIS eXtensible Address Language (xAL) which trickled down into everything from maps to HTML5 autocomplete.

[1] https://chromium-i18n.appspot.com/ssl-address

[2] https://github.com/commerceguys/addressing

[3] https://github.com/bojanz/address



Essentially the same deal in Ireland, with Eircodes. They were originally created as private dataset with ownership, and now you have to license access to it to use it.

Eircodes are better than postcodes, in that there's 1 per building/address/apartment, however they're discontinuous, so adjacent buildings will have distinctly different eircodes.

The article highlighted the difficulty of shopping centers and apartment buildings, from my experience trying to validate a large number of Eircode <-> addresses for a project, this is definitely an issue. The worse issue is that there's no way to just send someone out to check, because the eircode isn't like a house number that's posted somewhere. (Leaving aside the problem that valid Irish addresses can have no numbers outside of the eircode, and eircodes are a recent, and therefore non-traditional addition)



Eircodes also aren't used by An Post, to add insult to injury.

> The worse issue is that there's no way to just send someone out to check, because the eircode isn't like a house number that's posted somewhere. (Leaving aside the problem that valid Irish addresses can have no numbers outside of the eircode, and eircodes are a recent, and therefore non-traditional addition)

The HSE National Ambulance Service (NAS) National Emergency Operations Centres (NEOCs) have a GIS package that resolves Eircodes (and other traditional and colloquial addresses) to actual buildings and building entrances in real-time, which actually quite impressive. The directions can be transmitted to ambulances and other assets in real-time and has reduces delays in clinical services due to address confusion enormously since 2016.

So the country is capable. Eircode is what we chose as a country, not what we were limited to.



As the tenant, once, of a new build home in the UK it’s not just the file that’s important: it’s the channels to patch the file too.

I spent almost a year having to enter my address manually because the postcode DB — or whatever old version pets.com, cameras.com, and looroll.com had — lagged behind the reality of my infill bungalow for seemingly forever. I’m 8A godamnit, not 8. (Thank you Mrs. No8 for accepting my packages throughout those dark months btw.)

It’s just like tzdata. A precious resource not just because it compiles the history of geopolitical wallclock settings, but also because it is meticulously updated, on time and on budget.

It’s all very well liberating PAF.v2024_08finalfinal_v3_final.doc, but who is also going to keep it up to date?

I’m not a hater, just a realist. TFA is spot on: we’ll never be able out compete with or recreate or leak a sufficient version of the PAF. It, and it’s entire infrastructure, needs to be wrested into public hands ASAP. And we should be prepared to fund the updates.



As censorship for FOIA requests is done manually, it may be beneficial to request the missing figures directly without noting you have them in a censored context. Censoring is subjective, so that would at least draw out either the figures or a justification.



Not quite. G-NAF is a government owned enterprise, separate to the privitised but government body of Australia Post.

G-NAF is the equivalent to the UK's National Address Gazette.

It's a separate body of data, that sometimes disagrees with the "source of truth" that is Australia Post, and all the post systems that rely upon them.

For example, it took two years for G-NAF to notice that Winter Valley, Victoria, is not within 3356, but actually has its own brand new post code of 3358.



Sadly I don't think this would happen, particularly if Ordnance Survey is responsible, all their data is paid for access.

We have a very different model for access to data produced by government agency use to that in the US.

USGS Topographic maps: public domain / free

UK OS Topographic maps: paid access, and it's not cheep

US National Weather Service: Public domain / free commercial use

UK MetOffie: Payed access for commercial use



I remember asking a USGS person about this. They remarked that the other difference was that, compared with the OS, the USGS data was a bit rubbish (I may be paraphrasing).

The USGS is funded by some shard of the US federal budget, and does commendably good stuff with the budget it gets; it's there for both high-minded and commerce-supporting reasons. The OS is now (in a sequence of reorganisations from 1990 to 2015) a private company with a government-owned golden share, and is expected to be revenue-positive. The fact that it has more money per square metre of country, means that it's able to be _very_ thorough, mapping down to the level of individual bits of street furniture.

Sidenote: the context I was hearing this included a talk by someone from OS describing using reasoning software to do consistency checking of their GIS: for example, if you find a river bank in the middle of a field, something has been mislabelled. I thought that was cute.

When you buy a data product from OS, you're buying some subset of the layers of the database.

As the other reply pointed out, some of these layers are available for free, and in the last few years there's been some review/churn/debate in the data subsets made available that way (I see there are more details on the Wikipedia page). One can form a variety of opinions on whether those subsets are as big as they could or should be, but there does seem to be a substantial point that the level of the detail in the master map is there because it's profitable for the company (and thus income-generating for the government) to develop it from surveys, and it wouldn't exist otherwise.

I think the Met Office is organised in a similar way.

There are a number of questions of principle and practice here, but the OS seems to me to be claimable as an example (rare, in my opinion) of a privatisation which has produced net positive outcomes.



Unfortunately the British mindset these days is to either rent it out or sell it but, whatever the hell you do, don't grow it.

Somehow these idiots managed to strike a deal to keep the sovereigns figurehead on stamps (which has no economic value whatsoever, and actually the Crown should be compensated for this) but, in this data age, didn't safeguard such a critically important database to e-commerce

It's like selling off the Tower of London because you can't afford to repair the roof and forgetting you left the crown jewels inside



It blows my mind how many public services have been privatized in the UK. It just feels like they're selling off the shoes they're standing in. When their railways got privatized, the service didn't improve, the price just ballooned.

Even in the states, the USPS has resisted privatization this far. For the love of god I hope it continues to. Protect our boys n girls in blue and tell your congressman you want postal banking.



Yup. The Post Office, the railways, the water system, for heavens' sake!

The tories, as a matter of religious faith, see privatised => efficient, whilst being unclear on the difference between 'efficient at creating shareholder value' and 'efficient at serving the public good'. The political mood music, over the last few decades, has meant that the Labour party has repeatedly found itself obliged to say positive things about privatisation, as part of the process of Being Sensible About The Economy (there is a much longer alternative version of this comment!).

The US -- the world temple of capitalism -- seems to be oddly principled (viewed from outside) about keeping certain things such as the postal service, or USGS, as part of the service to the public realm.

The one service probably immune from privatisation is the Health Service. It's only the most frothing-at-the-mouth right-wingers, the provocateurs just one step away from a rabies injection, who'd even admit out loud to a desire to do that. A politician talking about privatising the NHS would I think be pretty much equivalent to a US libertarian politician talking about privatising the armed forces.

(there's a longer version of that comment, as well...)



Minor pedantic correction:

The Post Office (shops, services, government forms, etc.) is still fully government-owned. It's the Royal Mail (delivery) that was privatised. They used to be the same company but split in two before Royal Mail was privatised.



Why bother paying attention to all this legal mumbo-jumbo?

Just have someone exfiltrate the file and post it on Anna's Archive. Or extract them from Open Street Map.

Just bypass the Royal Mail altogether.

Or just ignore the postcodes. For most private individuals an ordinary street address works perfectly well and is needed anyway even if the postcode is provided.



This doesn't seem correct:
    The problem is it’s not an easy dataset to get hold of, as it cost a lot of money.
    This is because the data has to be licensed from Royal Mail ...
It seems to be talking about the National Statistics Postcode Lookup UK, which is officially published here: https://www.data.gov.uk/dataset/7ec10db7-c8f4-4a40-8d82-8921...

It's been there from at least 2017, which is when I first came across it.

There are later version of the data set online too: https://open-geography-portalx-ons.hub.arcgis.com/datasets/o...

The license: https://www.ons.gov.uk/methodology/geography/licences

    Under the terms of the Open Government Licence and UK Government Licensing Framework
    (launched 30 September 2010), if you wish to use or re-use ONS material, whether
    commercially or privately, you may do so freely without a specific application for a
    licence, subject to the conditions of the Open Government Licence and the Framework.
    If you are reproducing ONS content you must include a source accreditation to ONS.
If the article is talking about a different postcode address file though, then the above doesn't apply. ;)


I am a bit surprised by how hard this article makes out the problem to be.

Crowdsourcing should make short work of the problem, with the right incentives, which the government will be able to offer.

Additionally private map providers (e.g. Google, Apple) must surely have this data (since they are able to route navigation to private addresses). Why not just negotiate with them?



> Additionally private map providers (e.g. Google, Apple) must surely have this data (since they are able to route navigation to private addresses). Why not just negotiate with them?

They licence it from Royal Mail



Probably, but not necessarily.

The article points out that the PAF is kept up to date by virtue of thousands of postmen and postwomen physically visiting the rows in the database on a daily basis, as part of normal business, and logging updates. That level of routine maintenance is what any non-PostOffice PAF alternative would have to also do.

Amazon, and probably Google Maps, are two of the very small number of organisations which _might_ have the resources to build this postcode->GPS mapping, as a sideline to their current business.

They probably do license the PAF, of course, but they illustrate the sort of scale required to assemble that data independently.



I was a postie for a short while. A particular row of houses had no number 63, 61 and 65 were next door to each other. I always wondered if I posted something to 63 would it land in my sorting rack? Sadly I never tried, but I am fairly sure it would have. I often observed manual intervention to resolve addresses, from years of collective postie knowledge.



How would crowdsourcing solve this problem?

> Oh, and it wouldn’t even be legally allowed to include, er, postcodes, as they are specifically owned by Royal Mail



I can't find any good information post-privatisation, but at least before 2013 the postcodes themselves were copyrighted by Royal Mail (likely Crown Copyright as with government data). There were attempts to enforce this in 2009[0]. I suspect the copyright is now owned by Royal Mail Group Ltd.

That aside, a practical issue is that Royal Mail still retains the rights to _allocate_ new postcodes for any new properties. Yet another failure of this particular privatisation.

[0]: https://www.techdirt.com/2009/10/06/uk-royal-mail-uses-copyr...



Our company started operating in the UK recently, and some of our customers were very surprised we didn't charge for a subscription for part of our product. The idea would have no legs in Australia (our homeland) but is completely normal in the UK. So, new revenue stream for us, and some learnings about the UK culture.



great article, this demonstrates just how bad the civil service & politicians are when it comes to negotiating contracts with private investors… or trade deals, or brexit if it comes to that



Recent history teaches that the Post Office should be the last company on earth to be anywhere near creating a nationally important IT system. Their technology team have been useless for decades.

联系我们 contact @ memedata.com