National Postcodes to be introduced

chewed

ukoda wrote: »

Have you tried turning it off and back on again?

#ITsupport

I actually did turn off and on the laptop and it's working now!

BailMeOut

marmurr1916 wrote: »

The post office in Wilton Shopping Centre is:

AN POST
BISHOPSTOWN POST OFFICE
WILTON SHOPPING CENTRE
SARSFIELD ROAD
WILTON
CORK
T12 TV02

And the Bishopstown Credit Union in Wilton Shopping Centre is:

BISHOPSTOWN CREDIT UNION LIMITED
UNIT 27
WILTON SHOPPING CENTRE
SARSFIELD ROAD
WILTON
CORK
T12 X096

Boots chemists is:

BOOTS
UNIT 37/39
WILTON SHOPPING CENTRE
SARSFIELD ROAD
WILTON
CORK
T12 YK23

I could send a package to T12 YK23 with no other address details and the delivery company would be able to figure out that I meant Boots, Unit 37/39 Wilton Shopping Centre - or whatever shop(s) might lease Units 37-39 in future if Boots ever relocates or closes.

This is an excellent example on how useful/good this is going to be.

formerly scottish paddy

Deleted User wrote: »

Stick an extra character (or two) onto the end of the UK postcode and the problem is solved, simples!

Essentially that is why you need a house number for postcodes in the UK. I remember when they were introduced in Northern Ireland and every little country lane had to be given a name and then every house on it had to be given a number.

dr.fuzzenstein

formerly scottish paddy wrote: »

Essentially that is why you need a house number for postcodes in the UK. I remember when they were introduced in Northern Ireland and every little country lane had to be given a name and then every house on it had to be given a number.

And that's why this will never work here. Even if every country boreen had a name and every house a number, people would resist the change, they would refuse to use the new address and how many houses in towns that actually DO have a number, have that number up on the house? Very few.
If you look at houses in the countryside, you will see no identifying marks on them whatsoever, no name, no nothing. The general attitude is "well it's no one's business who lives here!". Honestly, no kidding, that's what people have said about Eircodes and why they won't be using them.
Cutting off your nose to spite your face comes to mind.

OU812

dr.fuzzenstein wrote: »

And that's why this will never work here. Even if every country boreen had a name and every house a number, people would resist the change, they would refuse to use the new address and how many houses in towns that actually DO have a number, have that number up on the house? Very few.
If you look at houses in the countryside, you will see no identifying marks on them whatsoever, no name, no nothing. The general attitude is "well it's no one's business who lives here!". Honestly, no kidding, that's what people have said about Eircodes and why they won't be using them.
Cutting off your nose to spite your face comes to mind.

I agree with this. They do need to pass legislation to enforce every dwelling having a number attached to it (of a specified size to avoid 2mm high numbers) and every road have a name (even if that name is "R168" etc.)

Make life a lot simpler for everyone. Won't ever happen though.

SpaceTime

I'd say though leaving off the first line of the address is going to make the postman's life harder than it should be as they'll have to look up the code to figure out where you're on about.

Unit X Wilton Shopping Centre
T12 ****

Would suffice.

The sorting systems will get the item to the delivery person / postal worker, but it's not going to provide them with something they can identify on the ground, so it might start to become a bit annoying if the top line of the address is gone entirely.

formerly scottish paddy

dr.fuzzenstein wrote: »

that's what people have said about Eircodes and why they won't be using them.

Well "they" may not use them, but government departments and utilities etc will be using them to send bills etc to "them".

BailMeOut

dr.fuzzenstein wrote: »

The general attitude is "well it's no one's business who lives here!". Honestly, no kidding, that's what people have said about Eircodes and why they won't be using them.
Cutting off your nose to spite your face comes to mind.

They will use the fast enough when out of home heating oil on some cold winters night and the delivery guy is calling looking for their Eircode so he can find their house!!!

TheBustedFlush

Why can't Eircode provide a map of the Routing Key areas so that people can see the area they are part of and how that relates to other areas?

It's public information. The addresses and codes are all available. So why not provide a simple map to give people a visual idea of where the various postcode areas are?

Answers on a postcodecard please to P01 NTED.

BailMeOut

TheBustedFlush wrote: »

Why can't Eircode provide a map of the Routing Key areas so that people can see the area they are part of and how that relates to other areas?

It's public information. The addresses and codes are all available. So why not provide a simple map to give people a visual idea of where the various postcode areas are?

Answers on a postcodecard please to P01 NTED.

Eircode partners will do this. In a few week/months we will be bombarded with all sorts of apps and utilities that will be slicing and dicing this data every which way you can.

TheBustedFlush

BailMeOut wrote: »

Eircode partners will do this. In a few week/months we will be bombarded with all sorts of apps and utilities that will be slicing and dicing this data every which way you can.

I'm sure they will/might.

My question was why don't Eircode do it? It's an obvious piece of information to provide to give people a visual map of where their code fits into their locality of their routing key. The phone directory used to have a useful map of where the STD codes were, and it gradually built up a picture in people's minds of what related to where.

I presume there must be some kind of understandable pattern to how the Routing Keys are named and where they're located.

So why don't Eircode provide this as the creators of the code?

BailMeOut

TheBustedFlush wrote: »

I'm sure they will/might.

My question was why don't Eircode do it? It's an obvious piece of information to provide to give people a visual map of where their code fits into their locality of their routing key. The phone directory used to have a useful map of where the STD codes were, and it gradually built up a picture in people's minds of what related to where.

I presume there must be some kind of understandable pattern to how the Routing Keys are named and where they're located.

So why don't Eircode provide this as the creators of the code?

There are links in this thread showing these already.

Eircode should stay in the data business and let business now purchase access and find all sorts of cool things to do with it.

The same questions were asked here when the housing pricing register was put out there and there was a whole load of people complaining that there were not presenting it in all sorts of useful ways. Well within days and weeks of it going live there are so many sites and apps using and displaying this info in a far better way than they could ever do.

SpaceTime

I was accidentally listening to C103 in Cork earlier on and they'd a director of eircode on who mentioned that they are in talks with Google at the moment and seem to think they should be on Google Maps and other platforms within 2 months.

ukoda

SpaceTime wrote: »

I was accidentally listening to C103 in Cork earlier on and they'd a director of eircode on who mentioned that they are in talks with Google at the moment and seem to think they should be on Google Maps and other platforms within 2 months.

To be expected, No mapping company is going to turn down a database of every single address in a country with a national official postcode to pin point it on a map. It's ideal for them. Google will be foaming at the mouth for this data. And once Google have it, the sat nav companies will have no choice as they are in serious trouble of being made redundant by Google maps and need to keep up. Sat nav sales have dropped a staggering 60% in the last few years

SpaceTime

I wouldn't be surprised if the hold up was the clarification of the legislation.

There were potential data protection issues, which have now apparently been resolved.

ukoda

SpaceTime wrote: »

I wouldn't be surprised if the hold up was the clarification of the legislation.

There were potential data protection issues, which have now apparently been resolved.

I would also say it was down to negotiating pricing, like I've said before, ericode are developing a completely different non transaction based product for mapping companies

TheBustedFlush

BailMeOut wrote: »

There are links in this thread showing these already.

Eircode should stay in the data business and let business now purchase access and find all sorts of cool things to do with it.

The same questions were asked here when the housing pricing register was put out there and there was a whole load of people complaining that there were not presenting it in all sorts of useful ways. Well within days and weeks of it going live there are so many sites and apps using and displaying this info in a far better way than they could ever do.

No there are no links in this thread showing a proper accurate map of the routing keys. Finding "all sorts of cool things to do with it" is not what I'm asking.

It's very simple - they have a very good mapping product for their finder. The almost obvious thing to do in helping people to understand and use the code (which is not the task of "cool' companies looking to make money) is to provide an overview map of the country of where they are.

One would be forgiven for thinking that this is something that Eircode would prefer people not to see right now for fear of some possible criticism.

Perish the thought.

BailMeOut

TheBustedFlush wrote: »

No there are no links in this thread showing a proper accurate map of the routing keys. Finding "all sorts of cool things to do with it" is not what I'm asking.

It's very simple - they have a very good mapping product for their finder. The almost obvious thing to do in helping people to understand and use the code (which is not the task of "cool' companies looking to make money) is to provide an overview map of the country of where they are.

One would be forgiven for thinking that this is something that Eircode would prefer people not to see right now for fear of some possible criticism.

Perish the thought.

here is one

https://www.google.com/maps/d/u/0/viewer?mid=zrmXDfjvem7g.k2eSAnvagEdQ&usp=sharing

ukoda

BailMeOut wrote: »

here is one

https://www.google.com/maps/d/u/0/viewer?mid=zrmXDfjvem7g.k2eSAnvagEdQ&usp=sharing

I think he wants to see the boundary lines drawn. Although I don't think the public care about that, not one mention on Twitter of any joe public looking for it.

TheBustedFlush

ukoda wrote: »

To be expected, No mapping company is going to turn down a database of every single address in a country with a national official postcode to pin point it on a map. It's ideal for them. Google will be foaming at the mouth for this data. And once Google have it, the sat nav companies will have no choice as they are in serious trouble of being made redundant by Google maps and need to keep up. Sat nav sales have dropped a staggering 60% in the last few years

Really? I thought that Google already license GeoDirectory for their mapping products - in which case they already have most of the key data.

So all they would really need are the Building Identity Codes that the state is planning to use i.e Eircodes

Why would Google want to use the official address data since it is evident that for an awful lot of addresses, they're not in the format that actual users use? Why would they be "foaming at the mouth" as you put it? (foaming at the mouth means they would be highly angry)

BowWow

SpaceTime wrote: »

I was accidentally listening to C103 in Cork earlier on and they'd a director of eircode on who mentioned that they are in talks with Google at the moment and seem to think they should be on Google Maps and other platforms within 2 months.

I'm sorry, I don't really understand this? I can already use Google Maps on my phone to find my way to an Eircode location - so are they not already on the Google Map platform?

seamus

BowWow wrote: »

I'm sorry, I don't really understand this? I can already use Google Maps on my phone to find my way to an Eircode location - so are they not already on the Google Map platform?

No. What's happening there is that the eircode website is telling your phone to go to GPS co-ordinates, which Google Maps understands.

What they're talking about is being able to search for an eircode within Google Maps, and that will bring you direct to the location.

It's a subtle difference, but not everyone uses Google Maps on their phones, plenty of people like to use the desktop version and print out the directions.

It also means that other websites can use the Google Maps API and use eircodes in their mapping applications.

OU812

Any ideas why Dublin 21 & 23 were not used ? I know they weren't before, but it seems dumb to loose them altogether when they could have used them.

TheBustedFlush

ukoda wrote: »

I think he wants to see the boundary lines drawn. Although I don't think the public care about that, not one mention on Twitter of any joe public looking for it.

You may have missed this, Ukoda, but the public is not defined by those who appear - inanely or corporately - on Twitter.

The Google map provided above is too cluttered for easy public understanding. A map that outlined the boundaries of the 139 areas would make more sense to people - with perhaps the more detailed information lying behind that for those who wished to see it.

The map should also include county boundary lines, so that people can see that the RK areas are not linked to counties. And it might also show the location of the post-town in order to help explain the otherwise inexplicable postal addresses that accompany people's eircodes on the website.

ukoda

TheBustedFlush wrote: »

Really? I thought that Google already license GeoDirectory for their mapping products - in which case they already have most of the key data.

So all they would really need are the Building Identity Codes that the state is planning to use i.e Eircodes

Why would Google want to use the official address data since it is evident that for an awful lot of addresses, they're not in the format that actual users use? Why would they be "drooling" as you put it?

You seem to be contradicting yourself in this post?

Google already use the base database (geodirectory) and then you go on to question why they would want it??

In theory, they don't need the address, just the ericode and geo's. That's correct.

ukoda

TheBustedFlush wrote: »

You may have missed this, Ukoda, but the public is not defined by those who appear - inanely or corporately - on Twitter.

The Google map provided above is too cluttered for easy public understanding. A map that outlined the boundaries of the 139 areas would make more sense to people - with perhaps the more detailed information lying behind that for those who wished to see it.

The map should also include county boundary lines, so that people can see that the RK areas are not linked to counties. And it might also show the location of the post-town in order to help explain the otherwise inexplicable postal addresses that accompany people's eircodes on the website.

Twitter is a good representation of a cross section of society. I made the point that no joe public types on Twitter have requested a map of the routing keys. Companies have yes, companies with an interest.

I'm not saying it wouldn't be a good idea to have the map, im just making the point that I don't see the demand for it.

Aimead

GJG wrote: »

But we should never sacrifice the good on the altar of
perfection.

But we didn’t even end up with the good. By adding random (or pseudorandom if you prefer) characters they have removed functionality (i.e. the functionality that comes with hierarchy) for no reason whatsoever.

This isn’t a case of, as you rather inaccurately portray it, anyone requesting more consultation. This is a case of an entire fecking industry (logistics and transport) being ‘consulted’ and then being utterly ignored. I cannot speak to other industries since I am not familiar with what submissions they did or did not make, but I do know the submissions and representations made by the transport and logistics industry – and we may as well have sat on our arses singing dixie for all the good our consultation did.

Consider this from a submission by CILT and tell me that it doesn’t make sense:

“Principles for a National Postcode
The Institute considers the following principles to be important in the decision to adopt a National Postcode:
It should be granular: A postcode should be capable of providing a detailed, precise and unambiguous identifier for each individual address.

It should be scalar: A postcode should be capable of expressing a greater or lesser degrees of granularity by extending or reducing the number of characters in the postcode. (For example, a scalar system such as OpenPostcode divides the country into successively smaller 5x5 grids. Five characters would identify a postal area, six characters an individual address, and seven characters a part of a property, such as a gate.) A scalar code would be able to accommodate any material privacy concerns by specifying the level of detail that could be made available in any particular database.

Open-sourced: A postcode should be derived from a source that is publicly available and free to use. It should be unlicensed, or if licenced, the licensor should be the Minister or ComReg; who would provide it to individuals or commercial bodies free of charge.

It should be decodeable: A postcode should be directly translatable to longitude/latitude coordinates. This would allow the postcode to interact with other data sources, improve its usefulness, and “future-proof” it.

It should be “Beyond Post”: The An Post GeoDirectory already has address identifiers for approximately 2.2m unique addresses in the Republic of Ireland. An Post does not require postcodes to deliver mail (although market liberalisation requires that a system to identify and efficiently deliver to postal customers be available). Since most of the justification for the adoption of a postcode is outside the letter mail sector, a value-for-money postcode must be capable of delivering the requirements of these services. A Post-Town system is not capable of these additional levels of functionality, and would be a waste of money. ”

That last point is, for me, the most crucial. Deliberately and artificially decoupling the code from GPS was in-fecking-defensible. Do you seriously want to try claiming that Rabbite, in all his wisdom, was correct in ignoring such common-sense principles as being “obstructionism”?

Whenever you see Alex White talk about how the stakeholders where ‘consulted’ you should probably read that as ‘ignored’. As I said earlier, I’m not familiar with the views of other industries, but I haven’t been able to find anything in any submissions that asked for ‘random characters so that the code isn’t hierarchical’.

A 9 character alphanumeric code gives 33^9 possible locations. Say we take Ireland as being 56,-11 to 51,-5 on GPS this gives about 215 billion square metres (actual land area is about 85 billion, but I’m deliberately including more for this exercise). So if we assume we want a resolution of 5x5 square metres this means we need 8.6x10^9 codes. 33^9 is on the order of 4.6x10^13. Suppose, for uniqueness, we add a ‘height’ component. For 9 entries this would need 4.1x10^10 codes. Three orders of magnitude more ‘possible codes’ is more than sufficient for single digit error correction and error correction for when two digits are reversed. Instead of having the silly point locations that Eircode has, each database entry would define a polygon for a property, so any self-generated codes can be mapped to a database entry (you could even display the closest entries for non-property locations too).

The above calculation is completely off the top of my head, and it may well be possible to achieve the same results with an 8-digit code (would need to crunch the numbers a bit more closely for that). The result is a code that has all the advantages of Eircode (some of which, imho, aren’t really advantages at all) with all the benefits listed in the CILT submission.

To put it bluntly, that the rather obvious advantages that come with being hierarchical were dropped (and for piss poor reasons at that) was a bit of a shock to us in the transport industry.

---

BailMeOut wrote: »

People will, computers will not. It will be computers and software that will be created using Eircode that will make deliveries very efficient.

We’ve had plenty of salespeople drop by our office who talked a good game about how their software can ‘optimise’ our deliveries, only for their software to choke when we tried out actual examples of what our deliveries are like. A single van with twenty deliveries has tripped up plenty of these salespeople. When (and this is usually the case in practice) we added in multiple loading locations and multiple vans/trucks the ‘optimised’ solutions being generated by the various software packages was often outright junk.

The overwhelming majority of research into the travelling salesman problem involves the simplest case (i.e. one loading depo, one truck, no transhipments, no packing space restrictions, etc.), and even then the computational load is massive. I’ve seen some impressive algorithms when it comes to this simplest case, but as soon as you introduce the load space restrictions, multiple vehicles and multiple load points then whole thing comes tumbling down.

The underlying issue is that the raw number of permutations are just too massive. The better algorithms try to achieve improvements gradually (eg: genetic algorithms), but this can have a tendency to get stuck in local minima within the solution space. The simplest case has relatively few local minima, and those that are there can usually be overcome by carefully tweaking how the randomness portion is processed (although, even here I’ve seen some good algorithms get stuck in nasty local minima). But when you are dealing with more complex (which are what we deal with in our business) cases then the whole shebang just goes to pot.

So many people ITT keep making comments like that quoted above. Genuine question for ye – do you actually know anything about the mathematics and computer science behind the travelling salesman problem?

---

Grand_Rajnah wrote: »

If only they could effectively get the message out there that they didn't just throw scrabble letters at a map and see where they fall - a lot of thought has gone into this!

Consider the calculation I did above with a hypothetical 9-digit code. Assuming we kept the first 3 digits as some sort of routing keys, in designing the algorithm you’d have free rein how to bake-in your offsets for doing the error correction. A set of 2.2 million sounds quite large, but you could crunch different offsets and calculate to see how well those offsets satisfy the rules described in the document. Given that for the vast majority of address the full choice of 9 (which I described as a ‘height’ component, but is really a way to have multiple addresses on the same physical location) would be available, so it would be relatively simple matter of picking out which space within the 9 best satisfies the rules. The starting point would be getting the offsets that best follow the rules for addresses in the same physical space, and then choosing the best of the 9 spaces for each remaining address.

The punchline here is that it was perfectly possible to preserve hierarchical structure while having the error correction and the unambiguous codes for clustered addresses. Maybe a bit more thought was needed…?

---

ukoda wrote: »

Pretty comprehensive defence of eircode here from Gamma, they seem to be tweeting this link all over Twitter in defence of eircode, I guess someone had to do a bit of damage control

Shame absolutely none of that applies to anything I have said. It is actually rather disappointing when you consider that, behind An Post, the transport industry would have the most experience with, and would benefit the most by the solving of, non-unique addresses.

---

plodder wrote: »

Whatever about paying money up-front, I have heard they don't like paying subscriptions, which makes sense given they don't charge for their own lookups.

I’ve noticed that some of the aerial photography that Google uses has been downgraded. There was one farm that we used Google satellite view to get an idea of where the outhouses where, and when I checked the same location a few weeks ago the quality had downgraded massively to the point where you couldn’t even see where the lanes were. Was wondering if the copyright holder of the original photography was charging them too much or cancelled their contract? I don’t think a yearly subscription would be a problem, but a per-user fee would be a no-no.

---

FishOnABike wrote: »

Hierarchical and coordinate based systems are not mutually exclusive.

To be more accurate, a coordinate system is an example of a hierarchical system.

---

plodder wrote: »

There was no public consultation involved in the process ….

There was plenty of consultation, the problem was that all of that was tossed out at the 11th hour for no justifiable reason.

plodder

TheBustedFlush wrote: »

I'm sure they will/might.

My question was why don't Eircode do it? It's an obvious piece of information to provide to give people a visual map of where their code fits into their locality of their routing key. The phone directory used to have a useful map of where the STD codes were, and it gradually built up a picture in people's minds of what related to where.

I presume there must be some kind of understandable pattern to how the Routing Keys are named and where they're located.

So why don't Eircode provide this as the creators of the code?

I don't think they particularly want to do it as there are quite a number of strange anomalies, such as parts of county Dublin finding themselves in D13 which nobody was expecting. There is also an enormous disparity between the biggest routing key areas and the smallest. These are only the ones that we know about.

The FTAI (who are behind loc8) have said they will produce a map.

ukoda

plodder wrote: »

The FTAI (who are behind loc8) have said they will produce a map.

You have to wonder why wouldn't you? As their stance was "it's of absolutely no use to our members"

Oh right yeah, it's because they've changed thier stance now to "its of some use to our members"

And they put a request out for companies help them map the routing keys

plodder

Aimead wrote: »

There was plenty of consultation, the problem was that all of that was tossed out at the 11th hour for no justifiable reason.

Regardless, there was no public consultation. The public wasn't asked if it wanted a hierarchical code or a random code. And there are various experts around who could have contributed usefully to the project. I'm talking about people like the NUIM experts who created the small areas dataset for the CSO. I'd say they were the primary technical experts in this field in the country, but they weren't involved.