Here's the plan to actually liberate the Postcode Address File
What the government needs to do to end the tax on innovation
This week only! If you take out an annual subscription to Odds and Ends of History before the 12th January, and send me your postal address, I’ll send you a “FREE THE PAF” badge that you can wear with pride!1
As Rishi Sunak taught us last year, the new year is a good time to set goals. And regardless of whether or not we manage to lose weight, read more books, or stop the boats, it’s always wise to have an aspiration in mind, as something to aim for in the year ahead.
That’s why this year I’m going to set myself and the readers of Odds and Ends of History a goal. Not five, just the one, as we’re going to remain realistic about our ability to control inflation.
But this one goal is, I think, important. And I hope that as you’re reading this, you're raising your hand and pledging yourself to the cause too.
I’m talking, of course, about how 2024 will be the year that we finally liberate the Postcode Address File (PAF)2.
For the uninitiated, this is something I’ve been banging on about for ages. To cut a long story short, the database of every building address in Britain is currently not freely accessible. It’s owned by Royal Mail – and to access it, businesses and app developers need to pay some pretty outrageous fees3.
For example, if a bedroom coder has a great idea for the next big thing, but it happens to use address data in some way then… bad news. Before you can launch your business, you’ll need to pay around six grand to licence the address data held in the Postcode Address File4.
Now I know what you’re thinking – yes, this is madness, and no, of course I do not get invited to parties.
So as I’ve said before, in my view, the PAF should be made available for free for people to build on, as the fees amount to a tax on innovation.
And as crazy as it sounds, important people are actually taking the issue seriously.
Last year, in a ridiculous turn of events, I memed myself into a meeting with Jonathan Reynolds, Labour’s Shadow Business Secretary, to make the case directly to him. And I’ve also heard from a bunch of other people who work with geospatial data about how the existing PAF data licensing regime is a nightmare for them too. I think word is getting around, at least a little bit.
However, banging the drum is the easy part. That’s something any idiot with a Twitter account can do. The actual difficult part is figuring out exactly how the government could make it happen.
So I’m very pleased to say that some actual clever people have been thinking about this.
At the end of last year, the Centre for Public Data published a new briefing paper, co-authored by ‘PAF Avengers’ Anna Powell-Smith, Peter Wells5, and Amber Dellar, digging into some of the knotty policy details about how to reform the existing PAF regime, and actually liberate the data from the clutches of the Royal Mail.
So let’s dig into the briefing, and find out exactly what the government needs to do to ‘repeal the innovation tax’6.
If you value my writing please consider subscribing (for free) to get more politics and policy takes direct to your inbox. And if you really enjoy my work, take a premium subscription to get something new every single week.
Why freeing the PAF is complicated
The first thing to know is that address data is bewilderingly complicated.
Thanks to several historical accidents, something as simple as maintaining a database of postal addresses touches a bunch of different organisations, each of which have their own priorities and incentives.
For example, imagine a house is built. How does it get an address? First, the local authority has to assign it a street address - the human readable part, like “29 Acacia Road”7. But the council on its own lacks two other critical components: The postcode, and the Unique Property Reference Number (UPRN).
The latter is a basically a unique ID that has been assigned to every home or building. They were first introduced in the early 2000s. The thinking is that whereas a street or an entire block of flats may share the same postcode (with house or flat numbers used to differentiate residences), URPNs go down to the level of individual properties – so each flat in a building will have a unique one.
As you might imagine, they’re very useful for everything from mortgage contracts – so everyone can agree what building the paperwork is actually referring to – to more practical matters, like telling the fire brigade which specific house is burning down.
The only problem is that these two extra bits of critical data are assigned by two other, completely separate organisations: Postcodes are, of course, granted by the Royal Mail, and it is the Ordnance Survey (OS) that generates and assigns UPRNs.
The good news is that since 2020, UPRNs have technically been released to the public for free under the Open Government Licence – which means you can download the entire database yourself if you like.
Here’s an example of what the ~2GB dataset looks like inside:
However, the bad news is that the above is pretty useless on its own. And this is where things start to get expensive.
Say you wanted to build something that connects UPRNs to real postal addresses – perhaps you want to deliver packages to customers, or your app is designed to send a swarm of drones to assassinate your enemies. For those use-cases to work effectively, you probably need more user-friendly address data to make target identification easier. So you need to connect the UPRN to a real address.
At this point you either need to pay Royal Mail to licence the Postcode Address File data, or pay for a third party service like Ordnance Survey’s AddressBase (which itself, sub-licenses PAF data) to join the digital dots.
However, this is just scratching the surface of the complexity8. I recommend reading PAF-hero Owen Boswarva’s detailed explanation for all the grisly details.
The ultimate takeaway though is a simple one: Getting hold of address data is really fucking complicated and expensive. As things currently stand, multiple organisations have their fingers in the address pie, and each organisation has wildly different incentives – and the result is that for anyone who wants to build using address data, the status quo is basically a nightmare.
How to liberate the PAF
The goal of all of this, of course is clear: The PAF-tivists like myself simply want it to be possible for anyone, from small business owners to academic researchers to nerdy men who write Substacks, to be able to build on the data held in the PAF for free.
But when it comes to untangling the mess, there are actually multiple policy options for doing so.
It’s similar to how we can agree the principle of “free at the point of use healthcare”, without that principle necessarily implying whether it is better to have hospitals be publicly or privately owned9, because the mechanism is less important than the outcome.
So in the same sense, there are multiple routes the government could take, and Anna, Peter and Amber present a number of credible options.
First, the government could take address management away from Royal Mail entirely and create a new government body to maintain the PAF and publish it for free. I think this would probably be the structurally most dramatic option, as it would require the government to pass legislation to do it.
And then after there would be a machinery of government question about where to situate the new body – as you can imagine a whole bunch of government acronyms, from DLUHC, OS, ONS, GC, DSIT or presumably many others all being plausible homes for the team responsible (you can play a fun game deciphering which bits of government all of those letters refer to – answers in this footnote10).
This option may also cost a little money to effectively renationalise the address management part of Royal Mail – but the figures involved are literally low-ish millions. The sort of money that wouldn’t buy you too many local chess boards.
The second option is via regulatory tweaking. Royal Mail stays in the loop as the body that compiles and manages the PAF, but the government instructs it to make the data available for free – perhaps by including open access in the Universal Service Obligation, which is the same set of public service rules that force Royal Mail to deliver to every obscure little Scottish island or mountain hamlet, no matter how unprofitable.
However, in the briefing document the authors caution that this slightly more diet option may not provide as robust oversight – and may not entirely remove all of the intellectual property restrictions, which are at the heart of the PAF problem.
Finally, a third option could be change the way the Ordnance Survey works. At the moment, the OS is in the slightly strange position of being entirely government owned but operated like a private company, with a mandate to turn a profit to pay for itself.
But the authors argue that it could be possible to change this – presumably either by bringing OS back into the centre of government, under the control of a government department, or by government paying a subsidy to the company.
If I’m understanding the implications of what they’re saying correctly, then this would mean that Royal Mail would continue to maintain and licence the PAF for money – but the Ordnance Survey would effectively buy a blanket licence to access the data, and share it with everyone in the country for free.
Ending the innovation tax
So that’s how it could be done – there are multiple options on the table, each with political and financial trade-offs for any government to take into account. And I’m sure given the dizzying complexity of the government bureaucracy, there are other models of doing it that I or the other PAF-avengers haven’t yet thought of either. (Any ideas? Let us know in the comments!).
In any case, as a self-appointed loudmouth on PAF issues, I don’t have a strong preference for any specific one of these options. I think they’re all credible and viable routes to finally achieving the liberation of the PAF. And if the government were to pursue any of them, the ultimate outcome would be better address data for you, me and children who are as yet unborn11. And this, in turn, would help grow the economy and spur innovation.
And if you’re still not convinced? I strongly recommend you go and read the full briefing paper, which goes into much more detail than I have here.
Hopefully after reading, you’ll be as radicalised on this issue as I am. And perhaps if enough of us bore on about it enough then 2024 really can be the year that we liberate the Postcode Address File for good.
I promise not to write about postcodes again next week. Subscribe now (for free!) to get more politics and policy takes direct to your inbox. And if you really value my writing, taking out a paid subscription so that I can continue writing every single week.
Oh, and don’t forget to follow me on Twitter and I guess BlueSky too.
Assuming Royal Mail don’t sabotage my post because, well, y’know…
Yeah I’m afraid it’s another post about this again.
Reminder that we’re not talking about personal data here. No names or anything, just the literal physical addresses of every building in the country.
And that’s only for one year.
Anna and Peter came with me to meet Jonathan Reynolds, along with the technologist Hadley Beeman. Luckily, they did the vast majority of the talking.
I’m going to keep using this “innovation tax” line because I think it sounds pretty persuasive.
Yes this is a Bananaman reference.
Despite what I just said, the organisation that assigns UPRNs is technically not Ordnance Survey, but a subsidiary called GeoPlace, which it co-owns with the Local Government Association – which is a membership organisation composed of all of the local authorities in the country. Lots of stakeholders!
This is another reason why PAF-liberation is the perfect pro-growth, technocratic, political-consensus-building policy intervention.
Department for Levelling Up, Housing and Communities, to the Ordnance Survey, to the Office of National Statistics, Geospatial Commission and the Department of Science, Innovation and Technology.
Now that I’m technically an activist on this issue I’m really trying to ham-up the rhetoric, like a real campaigner.
Bloody hell, all the same circuitry and obfuscation from RM and OS, just the same as - can it possibly be true - *13* years ago: https://www.theguardian.com/technology/2010/jan/22/postcode-petition-fails-blocked-number-ten
Nothing's improved since then. The PAF isn't particularly profitable (2005/6: revenue £18m, profit <£2m). There might be more numbers in the latest RM report.
The argument for making it free is as strong as ever - I think it would be persuasive to have more examples from other countries of the before/after effect: the canonical example of "free data" benefit is GPS, which costs the US ~$1bn to run annually and brings colossal economic benefit.
The people who need to be persuaded are the incoming Labour MPs who will be running Cabinet Office (fight with trading funds such as OS) and DSIT (fight with RM). The way to persuade them is to make the "this will drive economic growth" case - the present licensing system only favours big (usually American) companies such as Google and Amazon. Labour wants to talk big on economic growth.
I don't know if there's any case studies of benefits from the OGL on the OS data; those could help.
The biggest benefit will come from putting the PAF into an organisation which is most motivated to provide accurate data while not screwing money out of people; for that reason I'd suggest taking it away from RM, which can have a free licence in perpetuity, and giving it to ONS. (Is ONS a trading fund? I can't find anything on that.)
The question of whether something needs legislation or not isn't that important, in my view - what's more important is finding the right structure and then driving toward that, because if the structure's right then everything can be made to happen around it. (Getting OS to release data under the OGL didn't require legislation, even though it meant a loss of revenue.) Again, the message to push to Labour is that this is the Path To Growth which will also encourage home-grown businesses.
The funding suggestion of a small rise in LR fees has been suggested many times (I think Bob Barr was a champion of this idea). Clearly, the Tories simply don't have any interest in this topic at all, as evidenced by the GC "not having got around" to looking at the PAF in 13 years. Another possibility is that there's a charge per house registered. Apparently Labour wants to build 1.5m homes in its first five years, so £60 per house (300k houses annually, £18m annual revenue to run PAF) covers it.
This stuff is all doable, and desirable. It just requires a bit of adjustment and movement, and the trouble is this government long ago ran out of ideas for how to adjust and move things.
Finally: "I recommend reading PAF-hero Owen Boswarva’s detailed explanation for all the grizzly details." Bear with me - I think the word you want is "grisly".
Joining this late, but I very much agree that the PAF data should be opened up. Last I heard, the PAF file itself is compiled by a 3rd party company (AFD based on the Isle of Man) which presumably has a cost attached to it in addition to the costs on Royal Mail's side. I don't see any problem with continuing to pay AFD or one of their competitors to do the compilation, as long as it's not bloody Experian, may they rot in a rainy hell of Monday mornings.
On the income side, I read here that the profit in 2005/6 was <£2m on revenues of £18m, but I would be very surprised if both those figures haven't increased hugely since then, not least because I've seen the bill to my organisation for our very modest number of lookups per year which is enough for us to consider not bothering with any address validation at all. I'm terrified to think what large online retailers pay each year for address lookups in their signup workflows: it must easily cover the cost of each address update in the PAF being typed in on a solid gold keyboard by a Nobel laureate or Pulitzer prize winner and checked twice.
If the PAF data was liberated, my organisation might still pay a 3rd party to provide the software to do the lookup (currently we get this as a web service, but in the past it was local software hitting a local PAF file) but I would expect most of the per-transaction cost to go away, assuming the 3rd party passed on the saving. If they didn't pass it on, it would make it even more likely that open-source software would be produced to work with the (free) PAF file at a lower overall cost than the web service.