Eircode design

antoinolachtnai

Sam Russell wrote: »

I agree with the above, particularly the bold bit.

It is obvious to me that the routing code should cover about equal populations of about 1,000 addresses which would equate to about 2,000 routing codes. Coincidentally, that is about the number of An Post postal towns. It is about the number of codes that could be squeezed into a 3 character code of X99 format but if you allow XX9 or even X9X as well, then it fits quite easily.

When the redesign happens, they may go this route (pardon the pun).

I don't think they will redesign the actual code.

The best thing to do would be to redesign the commercial model, by issuing a free file which showed which eircodes were in which small areas. This would not fix the code design but would solve some problems with getting uptake on the code.

PDVerse

antoinolachtnai wrote: »

The more I read, the more obvious it is. The whole design process was captured by An Post. An Post got to dictate their requirements based on their own business needs. The design document states as much in an appendix which lists out An Post's ''requirements' which are really more like demands. These requirements define almost every single aspect of the code.

(The sarcastic hypothetical newspaper article above gives an idea how how this political power might have been brought to bear.)

This made political sense, but it does not make any technical or business sense, because An Post is not and will not be particularly significant in terms of volume of deliveries to homes - there are other organisations that deliver services to homes just as frequently - and there are many other organization that have much higher value deliveries. The reality is that An Post is in terminal decline, and that the daily flat postal service on which the code is oriented will not survive the decade.

This was the political reality of what happened. I am not knocking anybody involved, because they had to cope with the political reality and build something that fitted in with that political reality.

But the post-hoc technical justification is a bit tiring. The idea that the design was a response to the use of red envelopes on Valentine's day is very hard to take without feeling that one's intelligence is being insulted. Equally hard to stomach is the proposition that manual sorting at routing code level is critically important (and so there had to be a routing code), whilst manual sorting at the local level is not important at all (and so the code at that level was randomised).

Antoin, I could answer your post, providing as much information as I can, to help you better understand how mail sorting works and why the Routing Code is so important. However there really isn't any point in me doing so if you're going to dismiss it as "post-hoc technical justification". The main reason I post here is to counter misconceptions that I know are untrue, and to provide the rationale for decisions that were made. That is all, I'm not trying to change anyone's mind. This thread will not materially affect Eircode adoption, irrespective of whether it is pro or anti Eircode.

If you are willing to engage on the basis that we may agree to disagree on the weighting of pros and cons of design decisions then please ask away, but this time without accusations of lying.

plodder

oscarBravo wrote: »

It doesn't say much of anything, never mind "it all". Routing keys were never designed to be geographic areas. Showing geographic areas that represent routing keys is always going to require some arbitrary decisions as to where to draw lines, because the idea of a border between routing keys is arbitrary anyway.

That's a bit of a red herring tbh. PDverse has produced a map. Gamma produced a nice map also and they look much the same. So, the boundaries do exist and there isn't much doubt about them (except maybe for offshore uninhabited islands ..)

However, in some respects being able to represent the exact boundaries on a map doesn't actually matter that much. What matters most (as I keep hammering on) is the public aspect of the code and what it represents. Take an example of house price statistics and how open postcodes like the UK's relate to it.

Say someone phones an estate agent and asks what their house might sell for. The agent just asks - what's your postcode? And the person says for example BT99 3PJ (example made up) (and note not revealing the exact address, which means the query can be anonymous). The agent operates in the BT99 area, so he has a spreadsheet or some file on his computer with average prices for each postcode within that area. So, he doesn't need to look at a map (in this instance). He just relates the postcode given on the phone to the file of information on his computer and sees that houses there currently sell for X.

I'm sure someone will pipe in now, that Eircode doesn't prevent you from doing the same. Well it does, outside the granularity of routing keys, unless you have licensed access to its data and some suitable software to make sense of it. And routing keys are too big mostly for this purpose.

I don't know if a public property price register exists in the UK like here. But, if it did, you could compile detailed property values using UK postcodes, without needing any license from anybody. It would be public knowledge and not depending on estate agents, or anyone else.

ukoda

plodder wrote: »

Say someone phones an estate agent and asks what their house might sell for. The agent just asks - what's your postcode? And the person says for example BT99 3PJ (example made up) (and note not revealing the exact address, which means the query can be anonymous). The agent operates in the BT99 area, so he has a spreadsheet or some file on his computer with average prices for each postcode within that area. So, he doesn't need to look at a map (in this instance). He just relates the postcode given on the phone to the file of information on his computer and sees that houses there currently sell for X.

This is the red herring.

If you want to know an average house price for an area you just Google it

No estate agent is going to entertain you using them as a search engine

They will want to know all sorts of info before they would even think to commit to a price. (House type, bedrooms, garden, etc etc) if you think they'll simply answer a question like that when you call is ridiculous. Any estate agent worth their pay would have your contact details and be pushing for your actual address. Do you think they'll say "its 400k thanks for calling bye"?

PDVerse

plodder wrote: »

That's a bit of a red herring tbh. PDverse has produced a map. Gamma produced a nice map also and they look much the same. So, the boundaries do exist and there isn't much doubt about them (except maybe for offshore uninhabited islands ..)

However, in some respects being able to represent the exact boundaries on a map doesn't actually matter that much. What matters most (as I keep hammering on) is the public aspect of the code and what it represents. Take an example of house price statistics and how open postcodes like the UK's relate to it.

Say someone phones an estate agent and asks what their house might sell for. The agent just asks - what's your postcode? And the person says for example BT99 3PJ (example made up) (and note not revealing the exact address, which means the query can be anonymous). The agent operates in the BT99 area, so he has a spreadsheet or some file on his computer with average prices for each postcode within that area. So, he doesn't need to look at a map (in this instance). He just relates the postcode given on the phone to the file of information on his computer and sees that houses there currently sell for X.

I'm sure someone will pipe in now, that Eircode doesn't prevent you from doing the same. Well it does, outside the granularity of routing keys, unless you have licensed access to its data and some suitable software to make sense of it. And routing keys are too big mostly for this purpose.

I don't know if a public property price register exists in the UK like here. But, if it did, you could compile detailed property values using UK postcodes, without needing any license from anybody. It would be public knowledge and not depending on estate agents, or anyone else.

Once the Google API has been upgraded to support Eircode then there will be many scenarios similar to the one you have outlined, that will be possible without licensing data or software.

On the specific scenario, we actively went out of way to ensure that the postcode design could not be used as you outlined. It wasn't an oversight, it wasn't treated as unimportant, it was treated as incredibly important that this type of discrimination was not possible by just inspecting the postcode itself.

plodder

ukoda wrote: »

This is the red herring.

If you want to know an average house price for an area you just Google it

On what basis will google "know" this information? The only way would be if the kind of public information that I just described were available.

No estate agent is going to entertain you using them as a search engine

They will want to know all sorts of info before they would even think to commit to a price. (House type, bedrooms, garden, etc etc) if you think they'll simply answer a question like that when you call is ridiculous. Any estate agent worth their pay would have your contact details and be pushing for your actual address. Do you think they'll say "its 400k thanks for calling bye"?

They might ask is it a 3 bed or 4 bed, semi d or detached? That information with the postcode tells them all they'd need to know.

plodder

PDVerse wrote: »

Once the Google API has been upgraded to support Eircode then there will be many scenarios similar to the one you have outlined, that will be possible without licensing data or software.

On the specific scenario, we actively went out of way to ensure that the postcode design could not be used as you outlined. It wasn't an oversight, it wasn't treated as unimportant, it was treated as incredibly important that this type of discrimination was not possible by just inspecting the postcode itself.

I've asked this before. Who decided this, and why?

PDVerse

plodder wrote: »

On what basis will google "know" this information? The only way would be if the kind of public information that I just described were available.

They might ask is it a 3 bed or 4 bed, semi d or detached? That information with the postcode tells them all they'd need to know.

When you go to myhome.ie or daft.ie they display the actual property price sales for an area so you can determine this information now, for free. http://www.daft.ie/price-register/
Eircode makes this process much simpler for daft, they already use Google Maps.

ukoda

plodder wrote: »

On what basis will google "know" this information? .

Daft.ie, myhome.ie, property register, CSO published stats, Google is pretty smart.

PDVerse

plodder wrote: »

I've asked this before. Who decided this, and why?

The rationale is something I've explained many times. To quote a section of a blog I wrote in April 2015

Avoids "Postcode Lottery" issues
"Hello I'd like to draw an imaginary line around your house which may affect your property price, insurance costs, your school catchment area, etc. Any problem with that?". The secondary uses of postcodes appeared long after postcode boundaries were introduced in other countries. As we are late adopters these will be immediate. Current rumblings of lack of sequencing/hierarchy in the design would be replaced by local protests against the negative impact of public groupings of properties. Reaching agreement on where to draw the lines could prove impossible.

Imposing a public grouping of properties that introduces postcode discrimination is fundamentally unfair. As Eircode is unique to each property it allows grouping of addresses specific to each use rather than a "one-size-fits-all" approach embedded in the postcode. Therefore hierarchy in the postcode design is not only undesirable, it is simply not required.

The following links highlight issues with the introduction of postcodes in Fermanagh that the Eircode design ensures won't happen here:

Postcode Lottery Issue
http://fermanaghherald.com/2014/06/garvary-resident-out-of-pocket-over-p...

Low adoption
http://fermanaghherald.com/2014/11/only-30-of-fermanagh-residents-using-...

Address change resistance
http://fermanaghherald.com/2012/06/battle-to-save-townlands-moves-up-ano

plodder

PDVerse wrote: »

When you go to myhome.ie or daft.ie they display the actual property price sales for an area so you can determine this information now, for free. http://www.daft.ie/price-register/
Eircode makes this process much simpler for daft, they already use Google Maps.

How does eircode make that simpler? They are presumably referring to areas using the names that people already know.

The only thing that Eircode can add, that people might know about, is the routing key and we can see that routing key areas are mostly quite large.

On the postcode lottery question. Where was that declared in the requirements for your work? Where were the issues debated before that happened? This is an important question of public policy. It shouldn't have been left as an arbitrary design decision in a project like this.

ukoda

plodder wrote: »

It shouldn't have been left as an arbitrary design decision in a project like this.

Who are you to arbitrarily decide that it shouldn't be a project decision?

PDVerse

plodder wrote: »

How does eircode make that simpler? They are presumably referring to areas using the names that people already know.

The only thing that Eircode can add, that people might know about, is the routing key and we can see that routing key areas are mostly quite large.

On the postcode lottery question. Where was that declared in the requirements for your work? Where were the issues debated before that happened? This is an important question of public policy. It shouldn't have been left as an arbitrary design decision in a project like this.

Eircode allows the address to be placed at the exact location without any effort with a simple Google API call.

The process was a two year collaborative tender, a final tender and tender process, a design process and an implementation process. I understand that you disagree with some design decisions, but there wasn't an absence of input from affected parties, and the interests of the public were paramount. We researched the pros and cons of other postcodes extensively to ensure the public interest was properly weighted rather than just a list of requirements from industry and Government. The weighting of pros and cons of design decisions are detailed in the design document.

As an example, there is currently a lot of comment about what would have been a better design for CSO. What would have been achieved by a public debate about what CSO might want or need? CSO gave their input, and the design suits CSO requirements 100%. The structure and content of the Eircode Address Database is where most of these design decisions impacted. The actual postcode format of 7 characters, while a relatively small part of the overall design, was discussed and debated extensively.

oscarBravo

plodder wrote: »

That's a bit of a red herring tbh. PDverse has produced a map. Gamma produced a nice map also and they look much the same. So, the boundaries do exist and there isn't much doubt about them (except maybe for offshore uninhabited islands ..)

No, the boundaries don't exist, because routing keys don't define geographic areas. Eircodes belong to routing keys, and you can draw arbitrary lines between the locations of Eircodes that have different routing keys, but that doesn't make the routing keys geographic areas, and they don't have boundaries.

You're arguing that because it's possible to create arbitrary geographic areas that closely approximate the extent of routing keys, that that somehow proves that routing keys are geographic areas, and therefore should be treated as such for the purposes of criticism.

It doesn't matter how much you want routing keys to be geographic areas; they're not.

So, back to the point that started this discussion: the fact that the lines that border arbitrary areas used to visualise routing keys are ipso facto arbitrary doesn't "say it all"; it doesn't really say anything at all. It's basically a truism.

PDVerse

oscarBravo wrote: »

No, the boundaries don't exist, because routing keys don't define geographic areas. Eircodes belong to routing keys, and you can draw arbitrary lines between the locations of Eircodes that have different routing keys, but that doesn't make the routing keys geographic areas, and they don't have boundaries.

You're arguing that because it's possible to create arbitrary geographic areas that closely approximate the extent of routing keys, that that somehow proves that routing keys are geographic areas, and therefore should be treated as such for the purposes of criticism.

It doesn't matter how much you want routing keys to be geographic areas; they're not.

So, back to the point that started this discussion: the fact that the lines that border arbitrary areas used to visualise routing keys are ipso facto arbitrary doesn't "say it all"; it doesn't really say anything at all. It's basically a truism.

Just by way of clarifying the issue, here's a current screenshot (errors and all) with background mapping turned off. Where you decide to draw lines is arbitrary, we went with Small Areas unless they needed to be split, but that is but one approach.

plodder

oscarBravo wrote: »

No, the boundaries don't exist, because routing keys don't define geographic areas. Eircodes belong to routing keys, and you can draw arbitrary lines between the locations of Eircodes that have different routing keys, but that doesn't make the routing keys geographic areas, and they don't have boundaries.

You're arguing that because it's possible to create arbitrary geographic areas that closely approximate the extent of routing keys, that that somehow proves that routing keys are geographic areas, and therefore should be treated as such for the purposes of criticism.

It doesn't matter how much you want routing keys to be geographic areas; they're not.

So, back to the point that started this discussion: the fact that the lines that border arbitrary areas used to visualise routing keys are ipso facto arbitrary doesn't "say it all"; it doesn't really say anything at all. It's basically a truism.

"arbitrary geographic areas that closely approximate the extent of routing keys" is a contradiction. They can't be both arbitrary and closely approximate the extent of routing keys.

So, even though we can know that the boundary between two routing key areas lies between two houses, but we can't say exactly what the path is, you are saying that means the boundary doesn't exist...

It's a bit like saying, a carefully drawn circle using a compass, is not a circle, because something hand drawn doesn't (and can't) follow the 100% exact path of a circle.

plodder

PDVerse wrote: »

Just by way of clarifying the issue, here's a current screenshot (errors and all) with background mapping turned off. Where you decide to draw lines is arbitrary, we went with Small Areas unless they needed to be split, but that is but one approach.

Two points about that. First, as soon as you assigned a colour to each of those points (however it was done), the areas were defined.

Second, it really doesn't matter exactly where you draw lines so long as each colour is always on one side of a line.

PDVerse

plodder wrote: »

Two points about that. First, as soon as you assigned a colour to each of those points (however it was done), the areas were defined.

Second, it really doesn't matter exactly where you draw lines so long as each colour is always on one side of a line.

Step 1. Thematically shade ECAD buildings with their Routing Key
Step 2. Draw polygons around the buildings with the same colour
Step 3. When new buildings are assigned a Routing Key check they are inside their designated polygon
Step 4. Modify polygon boundaries as required
Step 5. Repeat Steps 3 and 4 forever.

This is the process that is followed in the UK, where postcode boundaries are continuously changing.

oscarBravo

plodder wrote: »

"arbitrary geographic areas that closely approximate the extent of routing keys" is a contradiction. They can't be both arbitrary and closely approximate the extent of routing keys.

If you think "approximate" and "arbitrary" are mutually exclusive, you're just redefining words to suit your purpose.

So, even though we can know that the boundary between two routing key areas lies between two houses, but we can't say exactly what the path is, you are saying that means the boundary doesn't exist...

No. I'm saying that you can arbitrarily define a boundary somewhere between two things, but that that doesn't logically imply that the boundary exists in its own right.

It's a bit like saying, a carefully drawn circle using a compass, is not a circle, because something hand drawn doesn't (and can't) follow the 100% exact path of a circle.

It's nothing whatsoever like saying that.

Routing keys aren't geographic areas. You can create approximate areas that encompass the contents of routing keys, true. I could create an approximate circle encompassing the place where you're standing, but that doesn't mean you're a circle.

plodder

oscarBravo wrote: »

If you think "approximate" and "arbitrary" are mutually exclusive, you're just redefining words to suit your purpose. No. I'm saying that you can arbitrarily define a boundary somewhere between two things, but that that doesn't logically imply that the boundary exists in its own right. It's nothing whatsoever like saying that.

Routing keys aren't geographic areas. You can create approximate areas that encompass the contents of routing keys, true. I could create an approximate circle encompassing the place where you're standing, but that doesn't mean you're a circle.

Whatever. We'll agree to disagree then. I'll call them geographic areas, and you can call them "approximate areas that encompass the contents of routing keys" or whatever you like and we'll just regard it as a a strange co-incidence that two different people produced maps that look remarkably similar from areas with arbitrary boundaries that don't exist :rolleyes:

PDVerse

plodder wrote: »

Whatever. We'll agree to disagree then. I'll call them geographic areas, and you can call them "approximate areas that encompass the contents of routing keys" or whatever you like and we'll just regard it as a a strange co-incidence that two different people produced maps that look remarkably similar from areas with arbitrary boundaries that don't exist :rolleyes:

Two different people didn't produce maps that look remarkably similar. I specified both. Gamma and Autoaddress share an Eircode.

plodder

PDVerse wrote: »

Two different people didn't produce maps that look remarkably similar. I specified both. Gamma and Autoaddress share an Eircode.

Fair enough, though, there are differences between them. You're not suggesting I assume, that two people couldn't have produced such a map independently, using the ECAD? You described how it is done in post #289.

I'll come back to your other post about the design process and "postcode lottery" questions later.

Carawaystick

PDVerse wrote: »

Antoin, I could answer your post, providing as much information as I can, to help you better understand how mail sorting works and why the Routing Code is so important. However there really isn't any point in me doing so if you're going to dismiss it as "post-hoc technical justification". The main reason I post here is to counter misconceptions that I know are untrue, and to provide the rationale for decisions that were made. That is all, I'm not trying to change anyone's mind. This thread will not materially affect Eircode adoption, irrespective of whether it is pro or anti Eircode.

If you are willing to engage on the basis that we may agree to disagree on the weighting of pros and cons of design decisions then please ask away, but this time without accusations of lying.

Can you explain how sorting is done for P51?

PDVerse

Carawaystick wrote: »

Can you explain how sorting is done for P51?

The short version is that Routing Keys are not geographic areas, so the fact that it appears as two discrete "areas" has zero impact on sortation.

The following is a gross simplification of the delivery process for An Post or any large courier/distribution company.

1. Get items from many locations and bring them to one or a few central locations.
2. Group items and send to dispersed delivery centres
3. Group items for delivery personnel to deliver from these centres
4. Deliver

If you can't find the address, this causes a delivery delay at step 4.
What impacts most delays however is step 2. The stickers that An Post put on mail to complain about the wrong postal address being used is because it caused step 2 to go wrong, which causes delay.

Most commentary along the lines of "I have a unique address which no one has any difficulty finding so I don't need to use Eircode" ignores the delay caused by step 2. An Eircode dramatically improves the optimisation of step 2.

The Routing Key is designed to optimise step 2 process for An Post for items that require manual sortation.

Curly Judge

PDVerse wrote: »

The short version is that Routing Keys are not geographic areas, so the fact that it appears as two discrete "areas" has zero impact on sortation.

The following is a gross simplification of the delivery process for An Post or any large courier/distribution company.

1. Get items from many locations and bring them to one or a few central locations.
2. Group items and send to dispersed delivery centres
3. Group items for delivery personnel to deliver from these centres
4. Deliver

If you can't find the address, this causes a delivery delay at step 4.
What impacts most delays however is step 2. The stickers that An Post put on mail to complain about the wrong postal address being used is because it caused step 2 to go wrong, which causes delay.

Most commentary along the lines of "I have a unique address which no one has any difficulty finding so I don't need to use Eircode" ignores the delay caused by step 2. An Eircode dramatically improves the optimisation of step 2.

The Routing Key is designed to optimise step 2 process for An Post for items that require manual sortation.

I'd like to thank you for bringing in a system that An Post didn't ask for, doesn't use and has no intention of using in the future.
Could you inform us of how An Post 's deliveries can be effected by a system it doesn't even use?

oscarBravo

Curly Judge wrote: »

I'd like to thank you for bringing in a system that An Post didn't ask for, doesn't use and has no intention of using in the future.
Could you inform us of how An Post 's deliveries can be effected by a system it doesn't even use?

An Post do use Eircodes.

plodder

An Post's delivery performance has been improving consistently for the last ten years, independent of Eircode, and the latest figures show 91% next day delivery (2015)

gctest50

Curly Judge wrote: »

I'd like to thank you for bringing in a system that An Post didn't ask for, doesn't use and has no intention of using in the future.
Could you inform us of how An Post 's deliveries can be effected by a system it doesn't even use?

An Post said Eircodes have been fully integrated into its automated sorting systems and it "fully supported and continues to support the introduction of the new Eircode system".

PDVerse

Curly Judge wrote: »

I'd like to thank you for bringing in a system that An Post didn't ask for, doesn't use and has no intention of using in the future.
Could you inform us of how An Post 's deliveries can be effected by a system it doesn't even use?

The more Eircode is used on post, the more An Post will use Eircode.

Sam Russell

PDVerse wrote: »

The short version is that Routing Keys are not geographic areas, so the fact that it appears as two discrete "areas" has zero impact on sortation.

The following is a gross simplification of the delivery process for An Post or any large courier/distribution company.

1. Get items from many locations and bring them to one or a few central locations.
2. Group items and send to dispersed delivery centres
3. Group items for delivery personnel to deliver from these centres
4. Deliver

If you can't find the address, this causes a delivery delay at step 4.
What impacts most delays however is step 2. The stickers that An Post put on mail to complain about the wrong postal address being used is because it caused step 2 to go wrong, which causes delay.

Most commentary along the lines of "I have a unique address which no one has any difficulty finding so I don't need to use Eircode" ignores the delay caused by step 2. An Eircode dramatically improves the optimisation of step 2.

The Routing Key is designed to optimise step 2 process for An Post for items that require manual sortation.

What difference for An Post would renumbering P51 (the bit east of the rest of P51) as P52?

Surely it would make no difference as the software would be programmed so the [P52 = P51]. However, it would be clear to anyone that P52 is to the east of P51, which could be quite useful.

Continue that argument for the huge area of H91 and divide into three new routing keys : Galway City; North of Galway City; and South of Galway City.

Continue that for all the huge ones and it begins to look more promising - maybe when there are about 500 routing keys that are arranged in logical areas like Urban vs rural, align as much as can be with county boundaries, etc.

You never know, people may even start using them.