Post/Zip codes and Ireland

fricatus

This discussion is very interesting, although some of it goes over my head.

One thing I will say is that this idea of basing the postcodes on the same letter combinations as county vehicle registrations is not a good idea in a country where one county or another provokes such strong sentiment (you need look no further than this post to know what I mean).

And basing them on provincial boundaries invites the same problems.

How for example would you treat the estate of Bishopsgrove, Ferrybank, Waterford? Half of it is in Waterford city and the other half in Co Kilkenny. Would one half have a W postcode and the other half a KK postcode? I understand that people living next door to one another have their bins collected on different days and pay different amounts for the privilege.

How would Waterford people feel about a KK postcode? Would they end up not using it? Seems to me that a neutral string of numbers is best, and the idea of basing it loosely on phone area codes is sensible too.

probe

antoinolachtnai wrote:

The phone areas are pretty arbitrary though. Wouldn't it be better to use the traditional divisions (province and/or county) to do the high-level numbering? Number the provinces from the north-west down to the south east and then number within th provinces for the second digit.

1. Irish counties don't have traditional numbers. When France was postcoded, it had a well established number assigned to each Department (read county)

2. If you live in Youghal and your telephone code is 024 and they tell you that the postcode will be 2400 next week - the relationship is obvious and easy to remember. You might conclude that someone in Mallow might have a 2200 postcode and someone in Dungarvan a 5800 postcode. And if you found in practice that these were correct assumptions, you would immediately know the postcodes for the places you had occasion to do business with regularly.

3. If you assign postcodes to strict geographical boundaries (eg counties) you will probably end up breaking the logic of the postcode system because manually sorted mail for a particular town will end up having to be sent to another neighbouring county - ie the county across the county line where its post is delivered from. Example:

Let's assume Ardmore, Co Waterford gets its post from Youghal or somewhere else Co Cork as a fictitious example. I suspect that it is not an issue for Ardmore, but the issue would arise in other parts of the country close to county lines. In which case all post for Ardmore should go into the Cork system. But if Ardmore gets a strict Co Waterford postcode some mail might find its way to a Waterford sorting facility due to human error and the rest of the mail for Ardmore goes via the Cork system. Duplication and delays will result. And the postcode will be less "reliable" (ie human error free) as a result.

If on the other hand Ardmore's postcode followed logistical lines rather than ancient county boundaries (like the phone NDCs) the problem would be less likely to arise. And people wouldn't connect numbering ranges strictly with geographical boundaries. Less "political" issues as a result.

.probe

probe

fricatus wrote:

This discussion is very interesting, although some of it goes over my head.

One thing I will say is that this idea of basing the postcodes on the same letter combinations as county vehicle registrations is not a good idea in a country where one county or another provokes such strong sentiment (you need look no further than this post to know what I mean).

And basing them on provincial boundaries invites the same problems.

How for example would you treat the estate of Bishopsgrove, Ferrybank, Waterford? Half of it is in Waterford city and the other half in Co Kilkenny. Would one half have a W postcode and the other half a KK postcode? I understand that people living next door to one another have their bins collected on different days and pay different amounts for the privilege.

How would Waterford people feel about a KK postcode? Would they end up not using it? Seems to me that a neutral string of numbers is best, and the idea of basing it loosely on phone area codes is sensible too.

You have several problems using the county codes as part of the postcode:

1) The political boundary one you mentioned.

2) The much higher level of machine recognition failure if you put alpha characters into a postcode.

3) Confusion between Zero and letter O etc.

4) It is not compatible with the system in use in the virtually all of the rest of Europe.

5) The cashier in your local supermarket can't enter your postcode into the system if they want to survey customers to see where they generally live unless they have a full alpha numeric keyboard at every checkout.

6) Numeric postcodes are faster to enter.

County codes are a non-runner!

.probe

probe

Hagar wrote:

Not quite accurate. La Poste uses a 5 digit code. Mine for instance is 83240.

PS:

If some people had their way with 9 or 10 digit postcodes!, and France followed suit with this idiocy, if, as you did give your postcode here, anyone who wanted to find out could determine if you lived on Promenade de la Mer, or Avenue de St Raphaël, or Avenue des Canissons or anywhere else in that town and your house number (I won’t mention the name of the town [where I have never ever been in my life] in case you value your privacy). With the result that many people would be reluctant to give their postcodes at all.

Making postcodes a lot less useful and acceptable in the real world. Nearly 40% of Irish phone numbers are non-published numbers in the phonebook.... (ignoring mobiles completely).

.probe

Hagar

Just look up the Pages Jaunes (Golden Pages to you) There aren't that many Irish here. What's the obsession with privacy? There is no privacy. If you think there is you are kidding yourself.

antoinolachtnai

probe wrote:

>>>I never said anything about naming all the roads.


1. If you break numbers up into chunks they are more user-friendly to remember. Give the house a number. Give the town a number.

It is easier to remember a house number 123 or whatever as one component and one is less likely to make a mistake transcribing it and 1000 as a postcode before the town name, than to remember 10002152 as the postcode.

Have you any evidence for this?

2. Updating Geodirectories and GPS devices becomes a huge operation if you give each building a unique number. Close to 100,000 new buildings go up in Ireland every year. You apply for electricity or a phone service or anything else and you must know what your postcode will be to arrange delivery and installation. Not only that but the supplier's database will have to have up to the minute access to new postcodes to use postcodes in their IT systems. Otherwise you might as well tell people to deliver the product/service to the Naas Road, Dublin (which has no building numbers and which is nearly as long as Wilshire Boulevard!). Might as well be in one of the more remoter areas of central Africa in terms of infrastructure!

If you buy a GPS navigation system for your car, it will rapidly go out of date for the same reason and be totally useless at finding new buildings if there are no house numbers to work on. None of the software written for any GPS system on sale works on the basis of premises unique postcodes. Britain is a far bigger GPS market than Ireland, and none of the GPS manufacturers have changed their GPS software (or the Navteq or Teleatlas maps) to use their full postcode system. GPSs generally only look at the first bit of the British postcode - which has the same level of resolution as the French or German or any other postcode system in Europe - ie down to postal district number. You must have unique building numbers for GPS to work properly for navigation applications.

So don't use the full postcode in those systems. Just use the first two, four or six digits. It's up to the manufacturer.

Inevitably, no matter what system you go with, you have to build a national address database, if you want to have unique addresses. Sure, this is a lot of work, but it's work that has to be done anyway, in order to keep the rates and electoral registers up to date, as well as the commercial databases.

No it won't and why should it be so? It would lead to large numbers of sorting errors. Mechanised sorting in Ireland and on the continent is done with full scanning and matching of each delivery address to the national address database. If this didn't happen, lots of mail would go astray for days in the system. They use a scoring process in most countries. If the system can't fully recognise the address, they weight it in favour of the numeric postcode because this is more machine readable and less likely to have a mistake in "spelling" - particularly if it is short. At the very minimum it gets the poorly addressed item to the correct delivery office where the postal workers with local knowledge can usually put it in the correct slot to complete delivery.

In terms of scanning, the more information the postcode gives, the better surely?

And I'd say it's more like 200,000 items hitting between 4 and 6pm, and I would say that at most 20 or 30 percent of these would be non-machine-readable. These could easily be keyed in this time.

If you have two townlands with the same name and code, you don't duplicate the building numbers across both townlands.

This is quite a messy hack. It's fine if you have to do this later on in a system. But to design a system in this way is unwise.

To make this system work, your townlands will have to fit neatly into the new postcode areas anyway. So why not just use EDs or aggregations of EDs?

>>>>Re the use of a sort code being irrelevant. An Post currently uses (or attempts to use) the address as a sorting code. Either that, or An Post really believes that Drumshambo is in Co. Roscommon, and that Shankhill is in Co. Dublin.

No they don't. The An Post system reads the entire address and matches it to their up to date geodirectory as it enters the sorting process. This process assigns the street number, building number (if any), town number, and GPS reference for the delivery address to 1 metre accuracy for that envelope. These data are stored in the An Post computer system referenced by a barcode serial number which you see printed on the envelope when you receive your post. Every time the barcoded envelope passes through a sorting machine the barcode is read, the address details are called up and the sorting machine decides which sorting pocket to dump the envelope into.

This is incorrect. For prompt delivery, you must use the correct postal address. (For example, Dromahair, Sligo, Co. Leitrim; The Illies, Lifford, Co. Donegal; Shanhill, Co. Dublin.) Many An Post addresses are geographically incorrect. They would not stand up if they were used on a legal summons.

The theory of fuzzy matching is all very fine, but generally speaking you should not depend on it to work, and An Post does not expect it to work.

The scanning process for Irish bound mail can also take place in France or Germany or wherever and their barcode is used by An Post. If the envelope can't be machine recognised, it is optically scanned and an image of the envelope is presented to human operators on a VDU. Herein lies the problem which can be solved by a simple unique numeric postcode in a standard position.

You might get the mail to the right delivery office, but you won't necessarily get it into the pile of the right postman. If the postman doesn't know the area well, then there will be a problem.

You post a letter in say Dublin 6 where it joins millions of others in the Dublin mails centre. If the handwriting can't be machine recognised and matched to the geodirectory, while it gets barcoded, it can't move to the next sorting centre (eg Little Island in Cork if it is for a Munster address) until a human being gets around to looking at an image of the address label on a VDU and reading the address visually so they can geographically code it to enable it to continue on its journey. Millions of letters are hitting the system in Dublin between 16 and 18h00 every day and many are failing machine recognition. They can't all be manually deciphered before say 22h00 which might be the cut-off deadline for the trucks leaving the Dublin mails centre for Cork. So it gets held over for another day. (It has a tight deadline to meet, because when the mail get's to Little Island it might be for a Limerick address or somewhere else distant in Munster and have to be out of Cork at another deadline if it is to be delivered on time.

This is just your assertion. It may well be true, but you have nothing to back this up. An Post management denies that there is any such problem. (Not that they are necessarily correct.)

Anyway, you are focusing on An Post's use of the code, and as I've said, I don't think An Post's needs are going to be a driver of the code.

If you had a short all numeric postcode on the envelope before the town name, it would have probably a 98% chance of machine recognition. With the result that the envelope could be put on the truck for Cork within the cut-off deadline - even though the full address hasn't yet been deciphered. By the time the truck reaches Cork the An Post people will have had several extra hours to examine the address on their VDU and properly route it. Even if they fail to meet the Cork cut-off deadline, it would be able to continue on to its final destination delivery office in Munster with 98% probability of accuracy and the guys with local knowledge at these locations could deal with it appropriately in a timely manner.

For sure, a code would have the benefit you mention. A granular postcode (to the level of ED, say) would have the further benefit that less local knowledge would be required in the final sorting office, without requiring a longer code.

>>>>Your point about the granular code in Copenhagen is well made and explains one reason why a granular postcode would be useful. There are many others (for example, the collection of statistics about businesses).

My point was that the extra granularity in Copenhagen does nothing for statistical gathering or postal delivery in the context of current technology.

That's because they have unique addresses and an address database. We have none. Anyway, it's still a useful benefit.

My analogy was intended to highlight the risk of user unfriendlyless of moving further away from well known norms (eg Dublin 2, Dublin 4, Dublin 24) and replacing them with 1024-9082 type postcodes. How many of your friend's postcodes would you remember if they all had 8 or 9 digits?

I'd just write down the first four or five digits, and that would do fine, especially if it was an urban address. Regular users wouldn't need to use the full 9. If I was sending a letter to someone who had recently moved to a rural area and might not be known in the locality, or if I was sending the item by courier, I might go to the trouble of looking up his postcode to be sure it would be delivered. (I might store this postcode in my addresses database, rather than trying to remember it in my head. This is what is usually done with phone numbers.)

>>>You need a postcode that is human readable and fairly granular if you want sorters in a central office who are not familiar with the localities to sort this stuff into the correct walks. You can't key every single address into the computer to find out what bag to put it in.

The best postcode for sorting anything manually is a four or five digit number that will get the item sorted to the destination delivery office level. Easy to read. No point in granularity beyond that. Ask the guys working in your local sorting office, they don't need postcodes to know that item X goes into Joe's bag because he delivers to XYZ road.

You are assuming that there is good local knowledge in the office. That is less and less the case, as deliveries are increasingly handled by non-local personnel.

The problem is that if you have a TNT driver delivering to an address in a rural area. There is no way for the person sorting or the person driving to know from the address which van that item should go on, and no way to determine a route.

Like what? Competitors in the postal arena will dump their mail into the An Post system at wholesale rates for final delivery. As is the case with liberalisation of the phone system - eircom lines are used to the subscriber's premises - except in the case of large business parks where it is economic for competitors to install their own fibre to the customer premises.

TNT Fedex, DHL and DX don't dump their mail into the An Post system at wholesale rates.

Even though the Dutch system uses 2 Alpha characters at the end of the code, the four digits give it a high level of machine readability. The rest of the code is redundant with current sorting technology. They postition the postcode before the town name like the rest of Europe. It doesn't cause any compatibility problems. Ditto for Portugal.

There must be some value in it if they persist with it. Why do they bother?

An Post actually use the grid reference - but the same thing. An Post defines the grid reference for each premises and stores it in the geodatabase. You can go out and define a grid reference which An Post rightly or wrongly might consider to be your kitchen rather than your front door and it doesn't matter a jot. Because the reference number assigned to your house in the Geodirectory is the only one that is used.

You can't get a grid reference for 40 percent of address in Ireland, because the addresses are non-unique. What does An Post actually use the grid reference for? A grid reference is useless for delivering mail.

>>>>A postcode based on railroads is a pretty silly idea in an era where no mail and almost no freight will ever be delivered by rail again.

Not really. It is spatially logical. If central Dublin was 9000 and Blackrock was 1234 and Dun Laoghaire was 4567 it would be really stupid! Doesn't matter if you use railway or roads.

What about places where there are no railway lines (most of Ireland).

The numerical relationship between the postcode assigned to one town and a neighbouring town or village is purely to facilitate human understanding and make it easy to remember. A computer based system can store any number for any town or house or anything else - even if they were all generated by random number generator. It wouldn't matter!

It's one way to do it, but it's not the only way, and it might not be the best way for our situation.

All I am speaking from is my own personal experience.

If it is posted in Britain or Ireland it often takes longer and the delivery lead time is more unpredictable - but it is far worse in the other direction because the continental sorting systems have problems dealing with the non standard address structure used in GB and IRL.

The fact that it's unpredictable for the UK and Ireland in both directions indicates that most of the problem is with the operations, not with the code.

No country in Europe has anything as like as poor as 75% next day mail delivery performance aside from Ireland.

Do you have figures for this? The only comprehensive numbers I know of suggest that Ireland has a higher next-day delivery target than 17 of the EU25 (which is what you'd expect, because we are a small country). Obviously, a target isn't the same as performance.

antoinolachtnai

probe wrote:

1. Irish counties don't have traditional numbers. When France was postcoded, it had a well established number assigned to each Department (read county)

So give them numbers.

2. If you live in Youghal and your telephone code is 024 and they tell you that the postcode will be 2400 next week - the relationship is obvious and easy to remember. You might conclude that someone in Mallow might have a 2200 postcode and someone in Dungarvan a 5800 postcode. And if you found in practice that these were correct assumptions, you would immediately know the postcodes for the places you had occasion to do business with regularly.

Works fine in Cork, but for donegal, it would seem to be pretty tough. Some of the NDC areas are very large and would definitely have to be subdivided. I suppose it would work if you just wanted a rough-and-ready code to suit An Post. It wouldn't be much use for rating packages though.

3. If you assign postcodes to strict geographical boundaries (eg counties) you will probably end up breaking the logic of the postcode system because manually sorted mail for a particular town will end up having to be sent to another neighbouring county - ie the county across the county line where its post is delivered from. Example:

You are talking about developing what is basically a sort code to suit An Post's needs, rather than to meet general needs. The postcodes would need to change in line with changes in delivery areas. Logistics providers would be dependent on getting regular updates from An Post.

There is no special Waterford sorting facility (that I know of). All mail is supposed to be sorted into bundles for postmen in the big sorting centres.

Using an arbitrary logistical boundary is going to cause a lot more political problems than using a hard, well-known, legally-defined administrative boundary. There have been a lot of shenanigans with Dublin post codes as a result of this over the years.

Victor

Some townlands are multi-polygons, i.e. detached from themselves.

Irish phone codes are at least partially laid out along railway lines.

probe

antoinolachtnai wrote:

Have you any evidence for this?

I have to admit that I have not carried out any empirical studies on the matter. But if you are in any doubt about whether giving a rural townland a house number (eg 123) or a big long unique postcode (eg 78290128) - take two rural areas and give rural area A house numbers and rural area B long postcodes. After a few years, drive around the locality stopping locals for directions to house number 123 and I suspect the typical response might be "my house is 155 and it is just over there, so 123 is just a little bit up the road there on the same side". In the alternative scenario, stop people for directions to the house with postcode 78291028 and you will probably get a blank stare or sorry can't help you etc.

Inevitably, no matter what system you go with, you have to build a national address database, if you want to have unique addresses. Sure, this is a lot of work, but it's work that has to be done anyway, in order to keep the rates and electoral registers up to date, as well as the commercial databases.

The national address database already exists - it is called the Geodirectory. You can download a sample copy of the database structure (MS Access format) from www.geodirectory.ie - it just needs to be updated. In it:
1) Every building in the country already has an 8 digit building identifier - call that a postcode down to building level if you want to.
2) Every "locality" already has an 8 digit locality ID.
3) Every group of buildings eg a terrace, apartment complex etc has a "group ID" (8 digits)
4) Every townland has a townland ID - both a geodirectory ID number (6 digits) and the "national townland ID number"
5) Every town has a town ID number
6) Every "thoroughfare" (by which they mean road, street, lane, place, park, etc has an 8 digit ID number).
7) Every building's street building number is shown where it has one. Rural addresses don't have any and that is one of the gaps I am suggesting that needs to be plugged.
8) There is no postcode field.
We are drowning in long postcode type numbers as it stands. We don't need another long postcode number because every address entity in Ireland already has multiple long, un-user friendly numeric identifiers pointing to its various address elements.

What is missing in the current environment?
a) 40% of houses and other buildings don't have house/building numbers to give them a proper street address.
b) There is no easy means of rapidly finding an address in a database or rapid address entry system - required in call centres, government agencies gathering and providing information on the phone etc. This demand can be filled by a short 4 or 5 digit postcode. Enter the postcode and the system "knows" the district you have in mind. Enter the first one or two letters of the street address and the system knows which street and town is being selected. Enter the house number and the system has the full postal address, which can be matched to the other data in the geodirectory (eg relevant ED number, grid reference, county name, and anything else you want to add). No need for involving the general public in complicated codes.
c) The existing An Post sorting system is failing to read the addresses on about 30% of single piece mail items - delaying their delivery for 24 hours or longer, giving Ireland the worst postal delivery performance of single piece mail and overall in Europe. A short numeric postcode appearing as the first item on the last line of the address would solve this problem in about 95% of cases - it works everywhere else on continental Europe.

In terms of scanning, the more information the postcode gives, the better surely?

The more information the postcode gives, the longer and more varied the postcode in each area and the higher the incidence of errors (number transposition, wrong postcodes etc). The system doesn't need more information than the postal district number because it is all the other data in the geodirectory. The issue is one of deadlines. Your unmachine readable letter posted in Dublin at 17h00 goes to the Dublin Mails centre. If it has the correct 4 or 5 digit postcode, it can pass through the mechanised sorting system and be put on the road to the next mail processing centre (ie Cork, Athlone, or Portlaoise) in a timely manner.
At any point before they start sorting the batch of mail at the destination mail processing centre (ie in the truck or while it is in the loading bay etc) the sorting staff can call up an image of the address label on a VDU and manually correct any recognition issue for that envelope. Every letter in the national sorting system has a database record with the envelope's barcode serial number and a scanned image of the address showing on the envelope. Once this database has the manually corrected address, it will fly through the mechanised system and can be sorted mechanically down to house by house delivery order for the postman in question if they want to do it to this level of detail.

And I'd say it's more like 200,000 items hitting between 4 and 6pm, and I would say that at most 20 or 30 percent of these would be non-machine-readable. These could easily be keyed in this time.

They obviously can't - An Post has had this delivery delay problem with single item mail since they first mechanised without postcodes or reforming the address system. Ireland was the only country in Europe to mechanise without postcodes. Or not use building numbers. If you have 200,000 items and it takes say 10 seconds for a human being to read an address and correct the routing information, that is 555 man hours without a break every night, performing at peak speed. Put 100 people on manual coding, and that will take 7 or 8 hours to do with breaks etc. In the meantime, this mail in question has to remain in the originating sorting office until the manual procedure on each envelope has taken place.

If you have your numeric postcode on the envelope, the work they will have to do is much reduced. (a) There is a far higher probability that the street name will be worked out by the software when it knows that it is only looking at streets in say Waterford city - even though the system mightn't have been able to read the word "Waterford" because of bad handwriting or whatever. (b) Rapid address entry can be deployed from the postcode so the full address can be put into the sorting item database (as described earlier above) with minimal effort in a second or two by trained staff.

This is quite a messy hack. It's fine if you have to do this later on in a system. But to design a system in this way is unwise.

The system I am advocating is the norm - used across Europe within and outside the EU, in Russia, and further afield. On the contrary, the idea of giving every house a postcode instead of a short building number and a separate postcode is a hack that will cost a fortune to set-up and administer with lots of consultants fees, special customised software required solely for the Irish market, and will cost a fortune to maintain and communicate to the public. And it won't be internationally compatible with other sorting systems. You don't need consultants to put house numbers on buildings that don't have any number yet! You don't need consultants to give a simple 4 or 5 digit number to each town. You just do it like every other country in Europe done - most of them 40 or 50 years ago.

To make this system work, your townlands will have to fit neatly into the new postcode areas anyway. So why not just use EDs or aggregations of EDs?

All townlands are assigned to posttowns already. The townland comes first, followed by the posttown. If you give every posttown a postcode, all the townlands that "hang off" that town get the same postcode. Where you have large groups of townlands they can be given their own postcode with the same town name if there is a worthwhile benefit. But generally speaking this should not be necessary.

This is incorrect. For prompt delivery, you must use the correct postal address. (For example, Dromahair, Sligo, Co. Leitrim; The Illies, Lifford, Co. Donegal; Shanhill, Co. Dublin.) Many An Post addresses are geographically incorrect. They would not stand up if they were used on a legal summons.[/unquote]
The correct postal address when you have a postcode system, does not involve a county name at all. (This removes one political issue from an address like this!). The fact that it is in Co Leitrim is irrelevant because in logistics terms it is virtually a suburb of Sligo if Lough Gill wasn't in the way. So assuming Sligo was say 7100 Sligo, and a postally correct address is Dromahair might be
1 MAIN STREET
7107 DROMAHAIR
Any other towns around Dromahair might also share the 7107 postcode. Similar principles apply to the other hard cases you mention - they get a code based on the location that serves them logistically and county names are irrelevant. This keeps it clean, simple and logical, easy to find and politically neutral.

You might get the mail to the right delivery office, but you won't necessarily get it into the pile of the right postman. If the postman doesn't know the area well, then there will be a problem.

There are four mechanised sorting centres covering Ireland. They can deal with all the post. 7107 DROMAHAIR will get your letter into the people's hands in Dromahair. If the postman in Dromahair doesn't know the area well (a very unlikely story) the question arises how did he get the job! Increasing the postcode for houses in Dromahair to 8 or 10 digits wouldn't make his job any easier. The existing system can still sort his post into bundles within his route structure if An Post or anyone else decides to machine sort down to a very low level.

This is just your assertion. It may well be true, but you have nothing to back this up. An Post management denies that there is any such problem. (Not that they are necessarily correct.)

I do have a basis to back it up. No country on the continent has a problem machine sorting and meeting delivery deadlines like an post, because they all use short numeric postcodes and house numbers. If you do any research on OCR handwriting recognition in sorting systems you will find that there is a much higher machine readability success rate for handwritten numbers compared with cursive alpha handwriting.

Anyway, you are focusing on An Post's use of the code, and as I've said, I don't think An Post's needs are going to be a driver of the code.

I am not. Anyone competing with An Post will share the same problems until this infrastructural mess is sorted out.

For sure, a code would have the benefit you mention. A granular postcode (to the level of ED, say) would have the further benefit that less local knowledge would be required in the final sorting office, without requiring a longer code.

If EDs for example were helpful in sorting post, the existing mechanised sorting system could print the ED on each envelope as it is passing through the system. It could print a mini map showing the location of the house if the postman was dyslexic using the grid reference. It doesn't because postman can generally read. The existing system causes the mail to be machine sorted and piled neatly into blue boxes ready for delivery. It is fairly idiot proof as it stands.

I'd just write down the first four or five digits, and that would do fine, especially if it was an urban address. Regular users wouldn't need to use the full 9. If I was sending a letter to someone who had recently moved to a rural area and might not be known in the locality, or if I was sending the item by courier, I might go to the trouble of looking up his postcode to be sure it would be delivered. (I might store this postcode in my addresses database, rather than trying to remember it in my head. This is what is usually done with phone numbers.)

What is the practical benefit of this? The person has just moved into the cottage at 133 TOWNLANDNAME in 7107 DROMAHAIR. The postman takes to to 133 TOWLANDNAME and puts it into the letterbox. He doesn't have to know the addressee to do this.

You are assuming that there is good local knowledge in the office. That is less and less the case, as deliveries are increasingly handled by non-local personnel.

OK take someone just arrived from Poland. He gets a job with An Post or DHL or anyone else. He is put delivering letters in Sandymount. He gets a bundle of letters for addresses in Claremont Road, Claremont Park, and Serpentine Road. If they are not pre-sorted by house number, he finds a corner and sorts them in house number order and gets on his bike with his street map of Dublin to find the three streets in question, and delivers them. Any "sophisticated" (ie complex) postcode system would only confuse the guy. Anyway the existing An Post system (as I have said before) can sort down to house number if they want to.

The problem is that if you have a TNT driver delivering to an address in a rural area. There is no way for the person sorting or the person driving to know from the address which van that item should go on, and no way to determine a route.

What will a big long postcode do to solve this? Give each driver a Garmin Nuvi 660 GPS, which has every townland in Ireland and at least he can get to the required area in a systematic and fast way. If they assigned house numbers to each house in the townland, his job would be even easier - he wouldn't have to stop and make enquiries from the locals as to where a particular house is. Additionally, there is nothing to stop TNT's sorting system from printing the co-ordinates of each address on the package, as it passes through their hub. If he is stuck, he can type those co-ordinate numbers into the GPS and it will take him to the door. All these data is already in the geodirectory, accurate to 1 metre.

TNT Fedex, DHL and DX don't dump their mail into the An Post system at wholesale rates.[/unquote]
No because there An Post still have a monopoly for under 50g letters and aren't therefore obliged to offer wholesale rates. But that is the sort of thing that happens in countries with more liberalised postal markets.

You can't get a grid reference for 40 percent of address in Ireland, because the addresses are non-unique. What does An Post actually use the grid reference for? A grid reference is useless for delivering mail.

A mechanised sorting system often uses the grid reference in parts of the sorting process. Each street is also defined by min and max grid reference points - they also define an itm_min and max for x and y - which presumably describes the shape of the street in more detail. When you think about it, addresses (O'Connell street) are meaningless to a computer. But if every street is defined in terms of its co-ordinate range (which is a metric postcode if you like) and every delivery point is defined in similar terms, the sorting order becomes mathematically obvious. If you look at the geodirectory sample download from [ http://www.geodirectory.ie/downloads/access2k.mdb ] you can see the various tables in the database structure. See also the manual: http://www.geodirectory.ie/downloads/GeoDirectoryTechnicalGuide_v8.pdf

What about places where there are no railway lines (most of Ireland)

.
I only used the example of railway lines to try and convey the logic of how the numbers are allocated. You don't have to have railway lines to assign numbers to delivery offices in geographically logical order.

The fact that it's unpredictable for the UK and Ireland in both directions indicates that most of the problem is with the operations, not with the code.

I don't propose to delve into what I have already said earlier again. One of the big problems the sorting machines have in for example GB and the US is finding the postcode on the address label! It's position is variable. With the standard European format, the position of the postcode is fixed - ie always on the last line and the first item on that line immediately followed by the name of the town. This is another reason why it works better.

Do you have figures for this? The only comprehensive numbers I know of suggest that Ireland has a higher next-day delivery target than 17 of the EU25 (which is what you'd expect, because we are a small country). Obviously, a target isn't the same as performance.

Have a look at the Comreg site - they produce a survey every now and again (using consultants) and the J+1 (next day delivery) single item delivery performance of an Post is usually around the mid 70%. This number is in the high 80s or mid 90s% in most other European countries. In Ireland if you look at An Post's delivery performance of machine generated mail (eg ESB bills and similar) there is very little problem because the computer printed addresses can be machine read with a high level of accuracy. This proves to me that the problem is not with postman not knowing their routes, or much of the other stuff you have brought up here. It is down to machine readability of handwritten addresses and the delays arising as a result.

.probe

probe

antoinolachtnai wrote:

So give them numbers

.
Why give each county a number? You can if you want to - but it should have nothing to do with the postcode structure. Your own example of Dromehair is a perfect example of why there should not be strict connection between the county and a postcode!

Works fine in Cork, but for donegal, it would seem to be pretty tough. Some of the NDC areas are very large and would definitely have to be subdivided. I suppose it would work if you just wanted a rough-and-ready code to suit An Post. It wouldn't be much use for rating packages though.

There is plenty of numbering space for Donegal and everywhere else. The whole of Northern Ireland is effectively divided into 1 or 2 digit district numbers of Belfast.

You are talking about developing what is basically a sort code to suit An Post's needs, rather than to meet general needs. The postcodes would need to change in line with changes in delivery areas. Logistics providers would be dependent on getting regular updates from An Post

.

What group's needs doesn't a 4 or 5 digit postcode meet? The postcode should be as permanent a fixture as possible. Otherwise people will become confused, and databases holding postal addresses will end up with incorrect postcodes. All the changes are in the detail of the geodirectory - not the postcode.

There is no special Waterford sorting facility (that I know of). All mail is supposed to be sorted into bundles for postmen in the big sorting centres

Did I suggest that there was a Waterford sorting facility? I have stated several times that there are only four in the country - DUBLIN, CORK, PORTLAOISE AND ATHLONE. All Limerick mail is sorted in Cork, for example.

Using an arbitrary logistical boundary is going to cause a lot more political problems than using a hard, well-known, legally-defined administrative boundary. There have been a lot of shenanigans with Dublin post codes as a result of this over the years.

So what? Turning Dublin 4 into 10492-0089 still smells like a Dublin 4 address. Unless you change it to 98021-8922 which would be totally stupid. At least if you have a properly enforced postcode system and address structure you won't get away with describing a penthouse apartment in Ballymun as Ballsbridge North.

.probe

Victor

probe wrote:

I have to admit that I have not carried out any empirical studies on the matter. But if you are in any doubt about whether giving a rural townland a house number (eg 123) or a big long unique postcode (eg 78290128) - take two rural areas and give rural area A house numbers and rural area B long postcodes. After a few years, drive around the locality stopping locals for directions to house number 123 and I suspect the typical response might be "my house is 155 and it is just over there, so 123 is just a little bit up the road there on the same side". In the alternative scenario, stop people for directions to the house with postcode 78291028 and you will probably get a blank stare or sorry can't help you etc.

Surely number 123 would be 78291123?

Turning Dublin 4 into 10492-0089 still smells like a Dublin 4 address. Unless you change it to 98021-8922 which would be totally stupid.

Of course, cos its 90210-8922.

Solair

I really don't think there's any issue with long codes. We all handle 10 digit phone nos with out the slightest problem.

087 555 5555
or 1-800 555 555

Can't really see why a long post code would be any issue at all.

Personally, I think they should be an entirely new coding system that isn't related to county boundries. We're obcessed with counties in this country in this kind of "GAA flag" mentality.

The beauty of having fully geographically accurate post codes is that you can then put whatever historically / culturally accurate address you like on the envelope, the code will find the location.

Also, as a matter of interest, the NI postal code system is not as "high resolution" as the system used in England/Wales/Scotland. The codes define addresses with a lot less accuracy.

antoinolachtnai

Probe, are you happy enough that when postal sorting is changed, the postcodes should change too? I do not think this is a very good idea. Lots of people other than An Post will be depending on the code, and when it is changed, it will cause a lot of disruption for them.

Why does An Post say it does not need a postcode if what you say about sorting accuracy is true?

Do you think Geodirectory should no longer be proprietary?

mackerski

Solair wrote:

I really don't think there's any issue with long codes. We all handle 10 digit phone nos with out the slightest problem.

That's because we have to. But postcode uptake by the public will take a lot of cajoling, and a big long number won't help this. Also, remember that a 10-digit number like 087-5555 555 is really a 7-digit number with one of four prefix digits in front. A lot easier to remember.

parsi

probe wrote:

Virtually all French mail (one of the biggest countries in Europe) arrives next day. Virtually all mail posted on the Continent arrives J+1 (ie post it Monday, it is delivered Wednesday).

I wouldn't call an average of 81.3% "virtually all French mail" ...

http://www.laposte.fr/IMG/pdf/Communique_de_Presse_-_Une_nouvelle_etape_dans_la_modernisation_du_Courrier.pdf

It's better than AN Post but not up to Comreg's desired 94%

probe

Victor wrote:

Surely number 123 would be 78291123?

Of course, cos its 90210-8922.

Not necessarily - because you are going to waste a lot of numbering space within any fixed length code structure if the last 3 digits = what should be the house number.

In addition, the minimum postal district number has to be at least 3 digits. If you are going to give a postcode to each house, each street will end up having a code - so that is at least another 3 digits to point to the street. And another 3 digits for the house/building. 3 digit house numbers locks you out of metric street numbering - because many roads are several km long bringing you into four and sometimes five digit house numbers. You also have to fit in apartment buildings with postcodes for each apartment, office buildings with multiple tenants, postcodes for PO Box numbers etc. etc. etc.

And someone has to maintain all these codes in a timely manner as 100,000 new buildings a year go up. For the entire country! €€€€€€ and mistakes all over the place and for what?

.probe

probe

antoinolachtnai wrote:

Probe, are you happy enough that when postal sorting is changed, the postcodes should change too?

What changes are proposed in postal sorting please? How often do district numbers change in the Dublin area? Virtually all the "changes" over the past 20 years or so amounted to the allocation of district numbers to areas that didn't have any (and usually previously called "Co Dublin" [eg Foxrock, Co Dublin became D18] When you have the entire country coded to district number level, the system remains very stable and requires very little maintenance in terms of postcode boundary issues.

Why does An Post say it does not need a postcode if what you say about sorting accuracy is true?

Surely that is obvious from my rather long posting yesterday! Sorting is based on grid references for streets and delivery points. If everybody computer-printed all their envelopes in a machine readable way with the correct address and every house had a house number they wouldn't need postcodes because 99% of letters would be machine readable and fly through the system. They would have no problem meeting Comreg's "targets" (sick joke). But a postcode has far more uses than sorting the post.... bla. bla. bla. And 25 to 30% of Irish mail will always have handwritten addresses... And Irish addresses are far too long anyway so they need to be rationalised.

Do you think Geodirectory should no longer be proprietary?

Ideally it should not - but it doesn't matter as far as I am concerned. What should not be proprietary is the postcode list. This should be freely downloadable from the internet as is the case for so many continental countries.

.probe

probe

Solair wrote:

I really don't think there's any issue with long codes. We all handle 10 digit phone nos with out the slightest problem.

087 555 5555
or 1-800 555 555

Can't really see why a long post code would be any issue at all.

There is a big difference between you presenting a super-easy to remember phone number to us in this discussion and someone sitting in a call centre typing in postcodes into a VDU all day, trying to listen to people with all sorts of different accents and disabilities and reading abilities. Either or both parties can make a mistake.

Would everyone remember their postcode and all the other postcodes they need for their lives?

You then have to set up time cocnsuming procedures to deal with cases where people can't remember or don't have their 9 digit postcode.

Anyway, nobody has demonstrated a logical need for a long postcode with 21st century IT technology.


.probe

probe

Aritcle in a Norwegian newspaper “Aftenposten” of 6.3.2007, cites recent Europe-wide research indicating that most European countries have better than 90% next day delivery performance for letters. One suspects that the survey didn't include Ireland which according to comreg runs at about 72%!*

Quote:

Postal service ranks poorly
Norway's postal service, Posten Norge, has been ranked as among the worst in Europe. It's also among the priciest.

A recent survey of European postal services showed that only 82.4 percent of first-class, so-called "A" letters mailed in Norway were delivered on time last year. The Norwegian postal service is supposed to deliver domestic first-class mail overnight.

Most other European postal services could report that more than 90 percent of first-class mail arrived on time.

Only Latvia and Germany had slower postal service than Norway. Twenty other countries had faster service.

Norwegian postal delivery has also gotten slower in recent years. Between 2001 and 2005, as much as 87 percent of A-post was delivered overnight.

A spokesman for Posten Norge admitted service wasn't good enough in Norway. He claimed, though, that Norway's large land area, isolated settlements and geography made mail delivery more difficult in Norway than elsewhere.

Postal rates just went up again, meanwhile. It now costs NOK 7 (about €0,85) to mail a domestic letter in Norway, nearly three times what it costs in, for example, the US.

Unquote

http://www.aftenposten.no/english/local/article1675269.ece

*http://www.comreg.ie/_fileupload/publications/ComReg0519.pdf

.probe

antoinolachtnai

Mail is not sorted on the basis of grid references, is it? As I understand it, it is scanned to figure out the unique ID for the delivery point, and then each item is sorted into the relevant walk for that delivery point. There are grid references in there alright, but that's more because the data is derived from OSi maps, than having anything to do with sorting and delivery.

Re stability, Dublin 6 was definitely split during the period you mentioned. Adding new areas to the numbering system generally happened because of a change in delivery practice as I understand it. I heard that there was an adjustment made out around Quarryvale. I've also heard about adjustments up around Dublin 15. It is also hard to believe that a country can go through this degree of economic growth and social change and for a number of delivery offices not to be opened or closed and for routes not to be changed.

Re long postcodes, you yourself have demonstrated the need for seven digits of code (or so) in many addresses. The only thing you are arguing about is that the digits should be split into two sections, with one part at the end of the address and the other at the beginning, rather than having the whole code at the end.

The other purpose of postcodes is for small area statistics. The CSO says they need this to produce statistics about areas smaller than the county and that they cannot do this economically using current techniques.

antoinolachtnai

probe wrote:

Aritcle in a Norwegian newspaper “Aftenposten” of 6.3.2007, cites recent Europe-wide research indicating that most European countries have better than 90% next day delivery performance for letters. One suspects that the survey didn't include Ireland which according to comreg runs at about 72%!*

What survey are they referring to? There are no citations there. n Post have their own statistics that put the next-day rate as being considerably higher than the ComReg figure. It depends what you measure.

antoinolachtnai

probe wrote:

What is missing in the current environment?
a) 40% of houses and other buildings don't have house/building numbers to give them a proper street address.
b) There is no easy means of rapidly finding an address in a database or rapid address entry system - required in call centres, government agencies gathering and providing information on the phone etc. This demand can be filled by a short 4 or 5 digit postcode. Enter the postcode and the system "knows" the district you have in mind. Enter the first one or two letters of the street address and the system knows which street and town is being selected. Enter the house number and the system has the full postal address, which can be matched to the other data in the geodirectory (eg relevant ED number, grid reference, county name, and anything else you want to add). No need for involving the general public in complicated codes.

You've put in a digit for area, down to the nearest 20,000 houses or so. You've put numbers on every house. Why not add one more digit to allow you to confirm the address?
[/quote]

If you have your numeric postcode on the envelope, the work they will have to do is much reduced. (a) There is a far higher probability that the street name will be worked out by the software when it knows that it is only looking at streets in say Waterford city - even though the system mightn't have been able to read the word "Waterford" because of bad handwriting or whatever. (b) Rapid address entry can be deployed from the postcode so the full address can be put into the sorting item database (as described earlier above) with minimal effort in a second or two by trained staff.

Sure. And if you had a more granular system you would be able to automatically sort the same mail into the correct bundles and the correct order, ready for the postman to deliver.

The system I am advocating is the norm - used across Europe within and outside the EU, in Russia, and further afield. On the contrary, the idea of giving every house a postcode instead of a short building number and a separate postcode is a hack that will cost a fortune to set-up and administer with lots of consultants fees, special customised software required solely for the Irish market, and will cost a fortune to maintain and communicate to the public.

I think I said that the idea of having separate blocks of house numbers where two townlands have the same name within the same postal area was a hack. Doing what you suggest with house numbers is the 'norm' in Ireland where two streets have the same name in the same area at the moment. However, it is very bad practice, and causes all sorts of confusion. It is definitely not the international norm.

The comments about consultants fees - I don't see why it would need many consultants to put a number on every county and electoral district.

I am not. Anyone competing with An Post will share the same problems until this infrastructural mess is sorted out.

They have the same problems, but they won't have the same solutions. A courier company doesn't have 'postal towns'.

What will a big long postcode do to solve this? Give each driver a Garmin Nuvi 660 GPS, which has every townland in Ireland and at least he can get to the required area in a systematic and fast way. If they assigned house numbers to each house in the townland, his job would be even easier - he wouldn't have to stop and make enquiries from the locals as to where a particular house is. Additionally, there is nothing to stop TNT's sorting system from printing the co-ordinates of each address on the package, as it passes through their hub. If he is stuck, he can type those co-ordinate numbers into the GPS and it will take him to the door. All these data is already in the geodirectory, accurate to 1 metre.

So it is necessary to pay An Post 80,000 euros a year just for the right to compete?

I only used the example of railway lines to try and convey the logic of how the numbers are allocated. You don't have to have railway lines to assign numbers to delivery offices in geographically logical order.

I thought you were assigning numbers to areas, and I thought you proposed to use the codes from the national dialling plan? What you are now proposing is a sort code to suit An Post's short-term needs. Why not just use the presort and postaim code that are in geodirectory already?

Have a look at the Comreg site - they produce a survey every now and again (using consultants) and the J+1 (next day delivery) single item delivery performance of an Post is usually around the mid 70%. This number is in the high 80s or mid 90s% in most other European countries. In Ireland if you look at An Post's delivery performance of machine generated mail (eg ESB bills and similar) there is very little problem because the computer printed addresses can be machine read with a high level of accuracy. This proves to me that the problem is not with postman not knowing their routes, or much of the other stuff you have brought up here. It is down to machine readability of handwritten addresses and the delays arising as a result.

.probe

None of this proves anything. Those aren't proper statistics to allow comparisons to be made across Europe. Stuff like ESB bills are only guaranteed to be delivered on the second day, not the next day, and are presorted in any case so they aren't that relevant. Are there actually any public statistics about the delivery times for these to allow a real comparison? You also haven't taken into account the effect of absenteeism.

You are mixing up speed and accuracy. Just because the machine gets a read on an envelope, it doesn't mean that it will code it correctly. A survey of next-day delivery times won't tell you all that much about overall accuracy. I'm sure there are internal An Post figures, but I don't know of any public survey of An Post's delivery accuracy.

There is little publicly available data, but anecdotally, I see a lot of problems with An Post's accuracy. My own experience (in an urban area) is that I to frequently received letters with printed addresses delivered to me that were addressed to completely different streets nearby. The mail appeared to have been sorted into the incorrect postman's box. Breastcheck complained that appointment letters sent to rural areas were occasionally delivered to addresses in the wrong county.

lostexpectation

any good makes of post or phone codes

perhaps one that breaks down to bouroughs(?)/phone area codes

probe

antoinolachtnai wrote:

Mail is not sorted on the basis of grid references, is it? As I understand it, it is scanned to figure out the unique ID for the delivery point, and then each item is sorted into the relevant walk for that delivery point. There are grid references in there alright, but that's more because the data is derived from OSi maps, than having anything to do with sorting and delivery.

The "unique ID" allocated and barcoded on each envelope after scanning is just a serial number (like a DHL or UPS air waybill number). These ID numbers are part of the ISO 15459 standard. This standard covers not only the barcode on individual letters etc, but also the containers in which they are transported, RFID tags, and lots of other related matters. If you look at the post delivered to your home, every barcode on each envelope is different. If it was an ID that referred to your home or street or whatever - the barcode would always look the same. It is the same situation everywhere else on the Continent. My postcode has nothing to do with the barcode on each envelope. In addition, the barcode is printed on the envelope even if the address can't be machine recognised. It is an item identifier - a link to a database of items in transit.

If you look at the unique IDs allocated to buildings, streets, etc by An Post, they appear to be serial numbers. There is no geographical structure or logic to the numbering. For example, if a new building is put up between building ID 10002456 and 10002457, what building number do they assign to it? It matters less if it is a serial number, but if it is the basis on which mail is sorted, you will end up with a dogs dinner in sorting terms, as new constructions spring up. At least the NGR to 1 metre accuracy (which they use) could deal with the task of delivering mail to a new doghouse inserted between two terraced houses, if FIDO was into correspondence via An Post. A "postman's walk" is probably also defined in terms of grid reference boundaries - like a street - only it is probably a wider box than a narrow street in most cases.

If you are only sorting down to the postman's route (rather than to street and house number level), it is a simple mathematical operation to compute which delivery points are within the boundaries of the postman's route.

Mail sorting software is complex and has to be robust. The companies that make the kit (eg Siemens Dematic) won't find it economic to develop special software to for countries that want to create non-standard systems. They would have Health Service type software fiascos on their hands for every project! So it seems to me that they develop a single platform that works universally - ie one based on geographical co-ordinates. This approach will work equally well for the "Sahara Desert Postal Agency" or someone delivering mail in Paris. It is down to the mathematics of a straight line and points along the line or a curved line or a "box". If a new construction pops up and it is simply a new point along the line or within the box in the scheme of things.

Logistics is about universality. You can't impose an "Irish solution" on the rest of the world. In the aviation industry, Dublin airport has an IATA code of DUB for reservations and ticketing systems, an ICAO code of EIDW (dumb American origin - the ISO2 code for Ireland is IE not EI), and 53 25'24" x 6 15'20" co-ordinates. The only really universal element is the co-ordinate reference, and this is what the flight management computer uses for navigation. Not dissimilar to the way that envelopes are "navigated" to the delivery point.

Re stability, Dublin 6 was definitely split during the period you mentioned. Adding new areas to the numbering system generally happened because of a change in delivery practice as I understand it. I heard that there was an adjustment made out around Quarryvale. I've also heard about adjustments up around Dublin 15. It is also hard to believe that a country can go through this degree of economic growth and social change and for a number of delivery offices not to be opened or closed and for routes not to be changed

So what? Everyone knows about Dublin 6 and 6W. Dublin 6W is really probably Dublin 26 in system terms. The changes are few and far between compared with what would be involved in maintaining a database of postcodes down to delivery point with about 1.8 million postcodes, and 600,000 more houses in the pipeline over the next 9 years!

Splits like Dublin 6 and 6W are also a product of bad planning in terms of allocating district numbers. If district numbering was laid out with potential growth in each area in mind, a district served by one DO at present might be allocated two or more district numbers today. When the time came to open a new delivery office to serve an expanded urban area, the codes would be well established and there would be no change from the users point of view. example: Let's assume Swords is Dublin 70 (ie 1070 SWORDS) and has only one postal district today, the opportunity could be taken to give people in Swords North perhaps 1071 and Swords West 1072 today. All three district numbers would be assigned to the same DO today. And when the time comes to open Swords West DO addresses remain unchanged. And if they never open a Swords West DO it doesn't matter.

Re long postcodes, you yourself have demonstrated the need for seven digits of code (or so) in many addresses.

I have not "demonstrated the need" for seven digits. I referred to the old German system which if replicated in Ireland would allow the existing Dublin district numbers to remain as they are. i.e. Dublin 4 would become 1000 DUBLIN 4 for all street addresses and all street addresses in the Dublin area would share the same 1000 postcode. That is very different from "demonstrating the need for seven digits". Functionally, there is no need for anything more than 4 digits.

The other purpose of postcodes is for small area statistics. The CSO says they need this to produce statistics about areas smaller than the county and that they cannot do this economically using current techniques.

Yes they can. The CSO have access to the Geodirectory. They can collect data to an accuracy of 1 M2 of land surface if they want to using this. If the CSO phone you up at random to find out how much pet food you buy or whether anyone in your house has been diagnosed with MRSA - they have several choices in terms of defining your geographic location.

1) They can use your postal district number (ie 4 or 5 digit postcode if/when it arrives). They can then produce statistics on how much MRSA has been found in Dublin 2, Dublin 4, Dublin 13, Cork 4, or 7107 Dromahair. These areas are far smaller than counties.

2) They can ask the interviewee for their full address, and with the postcode and about 4 more keystorkes pinpoint it to an individual house using rapid address matching via the geodirectory.

3) They can use the geographic telephone number they called. The footprint of the NDC + the first 3 digits of the local number is only a few km2 in most cases.

Anyway I have yet to see CSO statistical publications where the data supplied is granular even to postal district number! In most cases they use counties and urban boundaries.

If the CSO wants to impose a high resolution code for whatever reason on the country (hasn't happened anywhere on Continental Europe) let them write to each household and allocate them a "CSO code". They could fine everybody €3,000 if they don't know their CSO code - in keeping with the new Irish police state! But please don't let them warp any postcode system making it user-unfriendly and virtually useless for anyone else!



Finland Post Corp has taken advantage of this mail sorting technology to give postcodes / town names to major companies and organisation.

You can write FI 00045 NOKIA GROUP on an envelope and post it anywhere in the world and it will get to Nokia headquarters. Other Fin postcorp virtual company towns can be found here: http://www.posti.fi/svenska/transaktioner/sokpostnummer/foretagspostnummer.html


A breath of fresh air, compared with the complicated, antiquated postcode structures being proposed for Ireland, often by people who are drawing huge salaries from the state and have never lived in a country with postcodes, clean tap water, an efficient waste recycling system that doesn't involve dumping everything in China - exporting valuable raw materials and energy for nothing, and wasting zillions of kW of energy in the process, or a public transport system that is intermodal, efficient and comfortable enough for everybody to use, so that it becomes the normal method of travel rather than the private car, as is the case in Ireland. As for electoral divisions, I see they were introduced under the Poor Laws! "District Electoral Divisions originated as subdivisions of Poor Law Unions, grouping a number of townlands together to elect one or more members to a Poor Law Board of Guardians. The boundaries of District Electoral Divisions were drawn by a Poor Law Boundary Commission, with the intention of producing areas of roughly equivalent rateable value (the total amount of rates that would be paid by all ratepayers in the DED) as well as population. This meant that while DEDs were almost always contiguous, they might bear little relation to natural community boundaries." http://www.answers.com/topic/district-electoral-division

What an antiquated country!

.probe

probe

antoinolachtnai wrote:

What survey are they referring to? There are no citations there. n Post have their own statistics that put the next-day rate as being considerably higher than the ComReg figure. It depends what you measure.

I have no idea - but Aftenposten is a paper of record in Norway - they hardly made it up - particularly as it is somewhat critical of their postal administration. The way postal performance is measured across Europe is governed by a standard (EN 13850:2002).

An Post's figure does not comply with this standard.

.probe

antoinolachtnai

probe wrote:

If you look at the unique IDs allocated to buildings, streets, etc by An Post, they appear to be serial numbers. There is no geographical structure or logic to the numbering. For example, if a new building is put up between building ID 10002456 and 10002457, what building number do they assign to it? It matters less if it is a serial number, but if it is the basis on which mail is sorted, you will end up with a dogs dinner in sorting terms, as new constructions spring up. At least the NGR to 1 metre accuracy (which they use) could deal with the task of delivering mail to a new doghouse inserted between two terraced houses, if FIDO was into correspondence via An Post. A "postman's walk" is probably also defined in terms of grid reference boundaries - like a street - only it is probably a wider box than a narrow street in most cases.

I wouldn't think they do it that way. It would be a lot easier to have a list of which thoroughfares and settlements were covered by each walk. (You can do this with a database join, rather than having to do geographical lookups, so it is less processor-intensive and it is also easier to maintain.) Streets and roads are a common feature of all human settBut maybe you are right.

If you are only sorting down to the postman's route (rather than to street and house number level), it is a simple mathematical operation to compute which delivery points are within the boundaries of the postman's route.

It would be easier to just have a list of thoroughfare and building group id's that pertain to each walk.

So what? Everyone knows about Dublin 6 and 6W. Dublin 6W is really probably Dublin 26 in system terms. The changes are few and far between compared with what would be involved in maintaining a database of postcodes down to delivery point with about 1.8 million postcodes, and 600,000 more houses in the pipeline over the next 9 years!

You are proposing to number every home in any case. This amounts to a project just as big. You still have to build a directory or register of addresses.

Splits like Dublin 6 and 6W are also a product of bad planning in terms of allocating district numbers. If district numbering was laid out with potential growth in each area in mind, a district served by one DO at present might be allocated two or more district numbers today.

The point is that you said these divisions are stable. They aren't. They were devised as zone codes to allow easier sorting, nothing more. I think you are being uncharitable to the people who devised it when you say that it was badly planned.

I have not "demonstrated the need" for seven digits. I referred to the old German system which if replicated in Ireland would allow the existing Dublin district numbers to remain as they are. i.e. Dublin 4 would become 1000 DUBLIN 4 for all street addresses and all street addresses in the Dublin area would share the same 1000 postcode. That is very different from "demonstrating the need for seven digits". Functionally, there is no need for anything more than 4 digits.

You suggested applying a 3-digit house number to all rural unnumbered houses . You suggested a 4-digit code for areas the size of dublin postal districts and rural NDCs. That makes seven digits in total.

Yes they can. The CSO have access to the Geodirectory. They can collect data to an accuracy of 1 M2 of land surface if they want to using this. If the CSO phone you up at random to find out how much pet food you buy or whether anyone in your house has been diagnosed with MRSA - they have several choices in terms of defining your geographic location.

Why don't they do that then? They claim it is not practical. From my experience, they are probably right. The geodirectory is just not that accurate. The addresses in it aren't legal addresses and many of them are not commonly used. And it takes a lot more than just a database to match an address.

1) They can use your postal district number (ie 4 or 5 digit postcode if/when it arrives). They can then produce statistics on how much MRSA has been found in Dublin 2, Dublin 4, Dublin 13, Cork 4, or 7107 Dromahair. These areas are far smaller than counties.

Your postcode would be bigger than an ED and EDs would not fit evenly with it. That would mean that it would be impossible to make comparisons between the inter-census survey and the census. Without solid population figures, it's impossible to do accurate parametric statistics.

Also, regions such as 'Dublin 7' are enormous and very heterogenous. The average income for an area the size of Dublin 7 is fairly meaningless.

2) They can ask the interviewee for their full address, and with the postcode and about 4 more keystorkes pinpoint it to an individual house using rapid address matching via the geodirectory.

that doesn't work at the moment, even within the Dublin area. It would be haphazard enough if you kept using the current postal divisions in Dublin, for example. There are a lot of streets with similar names.

3) They can use the geographic telephone number they called. The footprint of the NDC + the first 3 digits of the local number is only a few km2 in most cases.

Do you think the CSO haven't thought about this? Your assumption about the geographical distribution of phone numbers is not correct.

Anyway I have yet to see CSO statistical publications where the data supplied is granular even to postal district number! In most cases they use counties and urban boundaries.

That's because they can't produce anything more accurate with the current addressing. The fact that they can't is an enormous problem, and there is a significant project underway to resolve this. As I understand it, it is a requirement that this code integrates with the postcode.

They do produce census statistics by EDs. (This is basically a constitutional requirement.)

They could fine everybody €3,000 if they don't know their CSO code - in keeping with the new Irish police state!

As for electoral divisions, I see they were introduced under the Poor Laws! "District Electoral Divisions originated as subdivisions of Poor Law Unions, grouping a number of townlands together to elect one or more members to a Poor Law Board of Guardians. The boundaries of District Electoral Divisions were drawn by a Poor Law Boundary Commission, with the intention of producing areas of roughly equivalent rateable value (the total amount of rates that would be paid by all ratepayers in the DED) as well as population. This meant that while DEDs were almost always contiguous, they might bear little relation to natural community boundaries."
What an antiquated country!

EDs are the basis of public admistration in this country for over 150 years. They may not be perfect, but it would certainly be cheaper and easier to use these than to make up a new set of boundaries from scratch based on the workings of An Post, which won't bear any relation to community boundaries and will have to be constantly modified and updated.

I think it is time to draw a line under the postal delivery discussion. Postal delivery is not really all that important for the future. Postal volumes are already in decline, and the decline is going to become more acute. Other uses of the postcode are a bigger focus.

probe

antoinolachtnai wrote:

I wouldn't think they do it that way. It would be a lot easier to have a list of which thoroughfares and settlements were covered by each walk. (You can do this with a database join, rather than having to do geographical lookups, so it is less processor-intensive and it is also easier to maintain.) Streets and roads are a common feature of all human settBut maybe you are right.
It would be easier to just have a list of thoroughfare and building group id's that pertain to each walk.

We are getting lost in technology irrelevancies here! Computer power is cheap in 2007. Sorting machines cost millions.
The big picture issues:

1) Labour is becoming more expensive everywhere, especially in Ireland. Dealing with items that aren't fully machine readable is very expensive in terms of labour cost and delays. Postal services everywhere are trying to optimise the machine readability of the mail they handle to get a grip on spiralling labour costs. Example, La Poste (post.be) is working on a web based address label generation service. Using this service, the customer will enter a postal address which will be machine verified for accuracy online using a web browser, and an address label with an item tracking barcode is printed out in the process. The address entered will be stored in the sorting system for a short period avoiding the need to scan the addresses on mail at all. They expect greater than 99% machine readability of all mail processed in this way.

An Post is bleeding financially and rural post offices are being closed all over the place - because there is no money to keep their nationwide infrastructure in place. Their resources are being soaked up in high wage bills dealing with labour intensive operations - the main one being sorting the mail.

2) Reducing delivery costs in all types of logistical services due to standardised, short, precise addressing, which complies with norms used in the rest of Europe to facilitate interchange of mail.

3) Keeping the postcode and address format as short, simple and geographically neutral as possible to make it more user friendly - thus facilitating maximum acceptance.

You are proposing to number every home in any case. This amounts to a project just as big. You still have to build a directory or register of addresses.

60% of houses already have numbers. You don't need a directory or register of addresses to use or benefit from house numbers. They are just there and have been for over a century. And there is space for building numbers in the geodirectory.

The point is that you said these divisions are stable. They aren't. They were devised as zone codes to allow easier sorting, nothing more. I think you are being uncharitable to the people who devised it when you say that it was badly planned.

I am not being uncharitable. Nobody seems to plan for anything in Ireland! Postal districts are one of the most stable elements in any country - far more stable than complex postcodes. If you have addresses with building numbers and postal zone postcodes, you can analyse data sets to any level of detail you require down to 1 M2 of land area.

You suggested applying a 3-digit house number to all rural unnumbered houses . You suggested a 4-digit code for areas the size of dublin postal districts and rural NDCs. That makes seven digits in total.

A house number is not a postcode. You are confusing very distinct elements in the postal address structure. Unfortunately you are not alone.

Why don't they do that then? They claim it is not practical. From my experience, they are probably right. The geodirectory is just not that accurate. The addresses in it aren't legal addresses and many of them are not commonly used. And it takes a lot more than just a database to match an address.

If you want to write "in the County of Wexford" in a summons or other legal document there is nothing stopping you. That doesn't make it necessary to appear on every envelope going through a logistics system for proper and efficient delivery. If there are inaccuracies or out of date items in the geodirectory, they should be corrected. As I pointed out earlier in this thread, An Post must have a version of the Geodirectory on their sorting system that is up to date - otherwise they couldn't sort the mail. Who knows, it may even be malicious intent by some people in An Post to keep the published version of the very expensive geodirectory out of date to make life difficult for their "competitors"?

Your postcode would be bigger than an ED and EDs would not fit evenly with it. That would mean that it would be impossible to make comparisons between the inter-census survey and the census. Without solid population figures, it's impossible to do accurate parametric statistics.

ED's are nothing more than artificial, antiquated boundaries based on "rateable valuations" for the Poor Laws! But if you are so hung up on them they are in the geodirectory so there is no problem matching datasets to them - providing you have the correct house number and postal zone code for each address.

Also, regions such as 'Dublin 7' are enormous and very heterogenous. The average income for an area the size of Dublin 7 is fairly meaningless.

This is getting boring. I won't repeat myself again!

that doesn't work at the moment, even within the Dublin area. It would be haphazard enough if you kept using the current postal divisions in Dublin, for example. There are a lot of streets with similar names.

So what? There are lots of Bahnhofstrasse's in Germany, and it doesn't cause a problem.
Every town name in France has been made unique by making slight changes to the original name in cases where there was duplication. eg

AIX EN PROVENCE
AIX EN ISSART
AIX NOULETTE

In the very isolated cases where you have duplication of street names in a postal district, you should amend them - eg Main Street North, Main Street South.
Postcodes have no role in correcting street name ambiguities. If you cut your finger, you put a bandage on the finger - not on your leg!

Do you think the CSO haven't thought about this? Your assumption about the geographical distribution of phone numbers is not correct.

Every country in Europe has statistical agencies. I don't hear any of them crying for more complex postcode systems so they can do their work!

Complex (ie "granular postcodes") seems to me to be yet another Irish re-invent the wheel exercise that will cost a fortune to implement, take years to achieve, and will end up a total mess. The best things in life are those that are kept simple - especially for the end-user.

.probe

antoinolachtnai

1. It's the same size project to give unique addresses whether you put the number at the start or the end of the address.

2. Postcodes (for example the German postcode) do change every year. For example, the US system changes around 1 or 2 percent every year because of changes in mail sorting practice (which are mostly driven by demographic change, not bad planning). As a result, after 10 years, over 10 percent of homes have changed postcode. You cannot expect postal sorting areas to be stable.

I think it is a good idea for postcodes to be stable, as do you. That is why I suggest not using postal sorting divisions.

3 EDs are the basis of public administration in this country for over 150 years and are stable in a way that postal districts are not and will never be. ED information has to be updated and maintained for constitutional reasons so there is no extra expense in maintaining them. There are published maps of EDs in existence. There are no published maps of an Post's proprietary sorting areas, and these maps would have to be produced and frequently updated at significant expense.

The La Poste system you describe is a complicated administrative solution to a simple problem. It might have a benefit in some circumstances but it's hard to see it being cost-effective. Royal Mail has a much simpler solution (and has had it for years). http://www.royalmail.com/portal/rm/content1?catId=600103&mediaId=15500215 . US mail has something similar - http://www.idautomation.com/usps-barcode-faq.html

antoinolachtnai

Regarding the problems of An Post, the company is definitely in a lot of trouble. The company really nees to innovate quickly, but it is hard to see how it can do so with the current IR situation. For example, if An Post really does have such a problem with scanning mail (which is in dispute) it could solve this problem within weeks by outsourcing manual mail coding to eastern Europe or India.

I think there is a way out of this, but improving mail sorting of regular mail is just a part of it and not the most important part. The big operational problem with mail delivery is at the delivery end - the company is too dependent on a small pool of skilled operatives -.

The big strategic problem with mail is that it is a dying business. By 2020, there won't be much old-fashioned mail at all. Really, An Post needs to be looking at rapidly developing the new businesses rather than trying to streamline the old ones.

larchill

It seems that the post code thing has become unstuck. A memo was submitted to the Government in May seeking approval for the implementation of a postcode system. Now it seems that another round of analysis to quantify the benefits is being carried out. A public consultation is to follow. All this is after two major reports, & an extensive public consultation which was carried out previously. Looks like another bloody Irish merry-go-round. See link for details ...

http://www.independent.ie/national-news/minister-cant-deliver-on-new-postcode-system-1062358.html