Eircode discussion

plodder · 22-01-2016 11:46am #1

Thread for discussion of aspects of Eircode other than its implementation

Sorry about all the Eircode threads. Hopefully all the defunct ones will soon drop off the main page.

plodder · 22-01-2016 11:47am

There was an interesting discussion on morning Ireland about a drug testing service in Wales. Leaving aside the drug policy issues, I think it's an example of the kind of privacy issue that would cause problems with Eircode. The service allows illegal drug users to submit samples of substances to be tested anonymously. They don't ask for personally identifying information, but they do ask for a postcode for statistical purposes.

The equivalent approach with Eircode would be to ask for the full code and then map that to a small area or some other area code. The problem is of course that the person still had to provide their entire Eircode which reveals where they live. Would drug users really do that, given that their activities are illegal and the code could be used to trace them? Alternatively, the site could just ask for the routing key part, but its statistical value is very limited given the enormous variation in size of routing key areas. Obviously, if such a service is ever provided in this country, then they simply won't ask for people's Eircode.

http://www.bbc.co.uk/newsbeat/article/25951290/welsh-government-funds-drug-testing-service-for-public

23-01-2016 6:13am

"Eircode discussion"

Do we REALLY need another thread on this?

Sam Russell · 23-01-2016 10:39pm

Mod: Let the discussion run for a while. If it gets nowhere then it will be closed or fall away. I think a valid point has been made.

clewbays · 24-01-2016 11:28am

plodder wrote: »

They don't ask for personally identifying information, but they do ask for a postcode for statistical purposes.

the code could be used to trace them? [/url]

Used by who to trace them? You give the impression that the information is being given in confidence and the postcode is being collected solely for statistical purposes which would exclude it being used for other purposes. So who else do you think would be given access and on what legal basis?

plodder · 24-01-2016 5:11pm

clewbays wrote: »

Used by who to trace them? You give the impression that the information is being given in confidence and the postcode is being collected solely for statistical purposes which would exclude it being used for other purposes. So who else do you think would be given access and on what legal basis?

It is only collected for statistical purposes, which does preclude it from being used for other purposes (as far as I know). But, I can't imagine in this country illegal drug users being happy to provide what amounts to their full address, regardless of the theoretical protection afforded by data protection law. So, if a service like that were ever offered here, I doubt they would ask for people's Eircodes due to the risk to anonymity.

clewbays · 25-01-2016 8:55am

plodder wrote: »

... theoretical protection afforded by data protection law ...

I hadn't heard about that before, thanks for bringing it to our attention. Maybe let the DPC know?

plodder · 25-01-2016 12:14pm

clewbays wrote: »

I hadn't heard about that before, thanks for bringing it to our attention. Maybe let the DPC know?

You are looking at it from the point of view of the data controller, who we all know will mostly do the right thing. From the point of view of the user, the perception is different, especially when dealing with state agencies, who tend to be all lumped in together (whether that is right or not).

I've provided other similar examples before. Eg market research companies looking for people's opinions on particular questions or products. They can ask for your Eircode, and they might promise they will only use it to establish roughly where you live, but if the code was properly hierarchical you could only give them the first X number of characters from your code, and you would be sure that they only roughly know where you live. That's an important distinction and is bound to affect the usefulness of eircodes for these applications.

Sam Russell · 26-01-2016 1:47pm

A solution to the above problem would be for the Eircode finder to offer a 'statistical' Eircode (or anonimous code) that gives a code for the small area. For example, say my Eircode were K10 t6t6, then a statistical code could be given that applies to the small statistical area I live and is the same for all those living in the same area. Those areas are already defined, and is probably already there in the Geo-directory.

Second Toughest in_the Freshers · 26-01-2016 1:57pm

In the case of the drug testing above, unless they're being manufactured in your kitchen, I don't see how much more use the full code would be over the routing key.

In other cases I'd give the code of the nearest post office/ the local shop/ school/

my3cents · 26-01-2016 2:02pm

I really don't see the argument about drug testing. If a system devised for an area of the UK wouldn't work here what relevance has it?

It isn't a flaw in Eircode if it can't be used for something it was never intended to be used for.

plodder · 26-01-2016 2:15pm

my3cents wrote: »

I really don't see the argument about drug testing. If a system devised for an area of the UK wouldn't work here what relevance has it?

It isn't a flaw in Eircode if it can't be used for something it was never intended to be used for.

Postcodes are used for all kinds of different things in other countries.

But, can you point me to some document that describes what it was intended to be used for? It's certainly not needed for delivering post.

my3cents · 26-01-2016 4:17pm

plodder wrote: »

Postcodes are used for all kinds of different things in other countries.

But, can you point me to some document that describes what it was intended to be used for? It's certainly not needed for delivering post.

Its very clearly stated on the Eicode website that

Eircode, Ireland’s postcode system, will enable people and businesses to find every address in Ireland, ...

and you want a system that can't find an address

You can't talk about what other countries do without comparing other factors. In the UK most houses have a number or a house name displayed, roads even minor ones have road name, as a result there was no non-unique address problem to solve. iirc the UK Postcode system was introduced by the post office, in Ireland our postal service didn't want a post code.

plodder · 26-01-2016 4:39pm

my3cents wrote: »

Its very clearly stated on the Eicode website that

Eircode, Ireland’s postcode system, will enable people and businesses to find every address in Ireland, ...

and you want a system that can't find an address

I never said I don't want a system that can't find an address. I'm asking what else was it supposed to support. Eg It provides unique identifiers for the revenue. There should be some document we can look at to see what all the requirements were. So, we can judge whether they made sense, and whether the system delivered, met them.

You can't talk about what other countries do without comparing other factors. In the UK most houses have a number or a house name displayed, roads even minor ones have road name, as a result there was no non-unique address problem to solve. iirc the UK Postcode system was introduced by the post office, in Ireland our postal service didn't want a post code.

That isn't relevant to this discussion which is about one use of postcodes for anonymised statistics. In your previous post, you said Eircode was never intended for this use. That's why I'm asking where exactly were the requirements on Eircode set out.

The two previous posters suggested ways this type of anonymous statistic could be collected, which is fair enough. It answers the question I asked, but I do think those solutions show that Eircode was not really designed with that type of application in mind.

Second Toughest in_the Freshers · 26-01-2016 5:17pm

my3cents wrote: »

It isn't a flaw in Eircode if it can't be used for something it was never intended to be used for.

well then, you'd have to ask why wasn't it designed to be able to handle smaller areas for statistical purposes?

my3cents · 26-01-2016 5:29pm

Second Toughest in_the Freshers wrote: »

well then, you'd have to ask why wasn't it designed to be able to handle smaller areas for statistical purposes?

Because then it wouldn't handle the issue of non unique addresses that it was designed to solve.

At the far end of our road there are about 10 houses all the same non unique address and all would have the same postcode under a UK type postcode system. Between the 10 houses there are only 2 surnames. Eircode solves that the UK postcode wouldn't.

plodder · 26-01-2016 5:49pm

my3cents wrote: »

Because then it wouldn't handle the issue of non unique addresses that it was designed to solve.

At the far end of our road there are about 10 houses all the same non unique address and all would have the same postcode under a UK type postcode system. Between the 10 houses there are only 2 surnames. Eircode solves that the UK postcode wouldn't.

Why does everyone keep coming back to the UK system? Eircode could have been designed to allow for this, and have unique codes. You would just ask for the first five characters of the code in cases like this.

Sam Russell · 26-01-2016 8:58pm

Small areas? Talk to autoaddress.

https://www.autoaddress.ie/blog

https://www.autoaddress.ie/blog/autoaddressblog/2016/01/25/big-benefit-of-small-areas

plodder · 27-01-2016 10:34am

Sam Russell wrote: »

Small areas? Talk to autoaddress.

https://www.autoaddress.ie/blog

https://www.autoaddress.ie/blog/autoaddressblog/2016/01/25/big-benefit-of-small-areas

What's interesting about that is that they are clearly sold on the concept of small areas (maybe after it being pushed so hard here).

Small Areas are perfect as building blocks to define delivery areas and help group and plan deliveries.

But it raises the question as to why they weren't built into the code transparently instead of being hidden in a commercial database?

They even went as far as creating a new pseudo-postcode along the lines people like me have been suggesting as shown in image below. Wouldn't Eircode have been so much better if you could look at a map of postcode prefixes and see a logical order to them as shown in their own picture?

my3cents · 27-01-2016 11:17am

plodder wrote: »

...
But it raises the question as to why they weren't built into the code transparently instead of being hidden in a commercial database?
...

Isn't this just dragging up old arguments that we have been through before?

Weren't the groupings made at a high level to stop what in the UK gets called the postcode lottery - nothing to do with winning money.

If you do group houses at a lower level within the code then there are always going to be anomalies. Someone for example that gets a poor delivery service because they are on the edge of a rural grouping instead of being on the edge of a town grouping - happens to me due to my An Post address making me look further away from the nearest town than I really am.

Eircode identifies the house without providing any other meaning. So I can't tell from your Eircode if you live on the rough side of town in an area know for drug dealing or if you are living in a posh area ripe for burglary.

In reality you want a system that can't find people but can put them in ghettos so services can be apportioned accordingly?

plodder · 27-01-2016 12:09pm

my3cents wrote: »

Isn't this just dragging up old arguments that we have been through before?

I didn't write the blog. If they hadn't posted it, I wouldn't have commented. It's a new angle on an old argument perhaps.

Weren't the groupings made at a high level to stop what in the UK gets called the postcode lottery - nothing to do with winning money.

That was the post hoc justification all right, but it was never seriously considered in terms of pros and cons.

If you do group houses at a lower level within the code then there are always going to be anomalies. Someone for example that gets a poor delivery service because they are on the edge of a rural grouping instead of being on the edge of a town grouping - happens to me due to my An Post address making me look further away from the nearest town than I really am.

Eircode identifies the house without providing any other meaning. So I can't tell from your Eircode if you live on the rough side of town in an area know for drug dealing or if you are living in a posh area ripe for burglary.

But, businesses who license Eircode and use this small area structure will know it and use it. Maybe you think the public are better off being happy in their ignorance. My theory is that they avoided making this structural information visible, partly to make life easier for themselves, and partly to make it easier to monetize the data. But, those are not valid design criterion for important public infrastructure (imo).

In reality you want a system that can't find people but can put them in ghettos so services can be apportioned accordingly?

There you go again with that claim :rolleyes:. I think the ghettos claim is a bit hysterical too.

Sam Russell · 27-01-2016 4:22pm

plodder wrote: »

They even went as far as creating a new pseudo-postcode along the lines people like me have been suggesting as shown in image below. Wouldn't Eircode have been so much better if you could look at a map of postcode prefixes and see a logical order to them as shown in their own picture?

That map looks like a post code map, and looks useful. Can we see more of it please?

MBSnr · 28-01-2016 11:09am

Where did it all go wrong? PAC aims to get to root of Eircode failures

marmurr1916 · 28-01-2016 1:46pm

This is the reason why any attempt to rationalise postal addresses in Ireland, to remove 'postal county' anomalies, would have been doomed to failure:

On a map, the 30 sq km of Roscommon that is being proposed as an addition to Westmeath looks like a square wedge marked off by one of those colonial lines used by 19th-century Europeans to carve up Africa.

Back then, imperialist decisions were taken without reference to the ties, customs and loyalties of indigenous populations.

The result, in some instances, was tribal warfare.

The language in Roscommon is no less incendiary. Independent TD Denis Naughten has spoken of a “set of ‘invaders’ marching from Mullingar . . . to take over the economic heart of Co Roscommon – and we have to stop them.”

Following the decision of Westmeath County Council last Monday to back this extension of its boundary into Roscommon, ill feeling, at least, seems inevitable. Rarely has the mood in Roscommon been as unanimous.

Even more rarely, politicians are united. Even the explosive county hospital issue did not bring people and politicians in the county together in the way that the proposed boundary revision has managed.

The Department of the Environment has recommended that the southern end of Roscommon should lose 30 sq km to Westmeath, arguing this would make it easier to manage the needs of a growing Athlone.

http://www.irishtimes.com/news/social-affairs/roscommon-unites-against-westmeath-invaders-1.2512859

plodder · 28-01-2016 1:46pm

At the PAC, John McGuinness asked Mark Griffin (secretary general) of Dept of Communications about the random codes and his answer was that housing policy is to use brownfield sites for new housing development and all neighbouring codes would have to be renumbered in such cases. It's debatable how much that would happen in reality, but it's a pity the question wasn't about the lack of hierarchy in the code rather than it not being sequential, which is what people are really looking for.

gctest50 · 28-01-2016 1:58pm

Sam Russell wrote: »

That map looks like a post code map, and looks useful. Can we see more of it please?

https://data.gov.ie/dataset/census-2011-boundary-files

plodder · 07-02-2016 12:18pm

Boatmad wrote:

I see great advantage to a system that has no meaning outside a database lookup. The fact is that any computer system can easily resolve eircodes to geo locations and go on to produce geo-orientated statistics.

designing a machine oriented code that humans can directly interpret , is 1970s technology in an increasing online world

personally I want my eircode to mean nothing to the casual observer

You're entitled to your opinion obviously, but I doubt that most people would share it. It also contradicts the recommendations made in the various reports that led up to Eircode. Eg, this report by Comreg (http://www.comreg.ie/_fileupload/publications/ComReg0507.pdf

Number 1 clearly says the code shouldn't have been a 'hidden or technical code'. That rules out the database lookup key idea clearly. In fact, Eircode fails on all of the other criteria as well except for 4. and 6 (unique addresses and funding).

It is a public postcode that is proposed, not a “hidden” or technical code.

It should be structured, at least to the level of specific areas within each county.

It must be easily memorised so that it will gain maximum usage.

It must solve the issue of non unique addresses without asking people to change the name of their townland, parish or county.

It must be neutral as between operators. In particular it must enable the postcodes to be aggregated for operational purposes in whatever way each operator desires. Just because An Post has decided to deliver mail for East Clare from Limerick doesn’t mean that another operator cannot decide to use Ennis as its base for the whole of Clare – and the postcode shouldn’t ask the addressee whose affinity is with Clare to use a different county name in the address.

The approach to funding should ideally be self financing /minimal cost to operators and government

my3cents · 07-02-2016 12:58pm

plodder wrote: »

You're entitled to your opinion obviously, but I doubt that most people would share it. It also contradicts the recommendations made in the various reports that led up to Eircode. Eg, this report by Comreg (http://www.comreg.ie/_fileupload/publications/ComReg0507.pdf

Number 1 clearly says the code shouldn't have been a 'hidden or technical code'. That rules out the database lookup key idea clearly. In fact, Eircode fails on all of the other criteria as well except for 4. and 6 (unique addresses and funding).

Which refers to An Posts use of "an advanced technical (hidden) postcode" as noted in section 6.2.

It doesn't back up your argument at all.

plodder · 07-02-2016 1:08pm

my3cents wrote: »

Which refers to An Posts use of "an advanced technical (hidden) postcode" as noted in section 6.2.

It doesn't back up your argument at all.

It refers to any hidden or technical code, not just An Post's. Why would it matter whose code it was if it were hidden or technical?

Granted the routing key is public, but it has been designed to discourage such use (variation in sizes, the lack of an official map etc)

The post I was responding to was making a virtue out of this 'feature' by arguing in favour of a completely meaningless code.

plodder · 11-02-2016 4:07pm

BoatMad wrote: »

There is no " other " side of the debate, Theres a sort of " down with that " without any clear reasons being advanced as to why Eircode is so unsuitable. Then we have the "crank" type claims , no GPO eircode, etc, well UPS " mightn't use it ", so its a complete failure.

The fact is you can advance certain technical criticisms of Eircode, but in the absence of an alternative system, its hard to make comparisons.

The UK system is very poor from a computer perspective as it was designed for humans , and the PAF was added later. Postcodes in a computerised society should be machine centric. ( which eircode is )

Ive covered a lot of info on eircode implementation , including issues around building courier custom sort lists and routes etc

It has been argued before, but maybe you joined in the discussion later.
Here is what I think Eircode should have been and then we can discuss the relative merits.

It's actually not all that different to what it is and is still based on database generated from geodirectory.

First two characters could have been a high level area (maybe a county, using existing car reg codes and existing Dublin postcodes)

Next three characters identify a CSO small area in the high level area.
Last two characters would identify individual properties in the small area, same as Eircode does. Could be random or not. Doesn't matter either way.

A free of charge database would provide geo-coordinates for centroid of each small area (five char code). This would be similar to what is provided free of charge in UK.

Paid for database is what ECAD is today.

Advantages

- does everything Eircode does today (ie identifies individual units) enables software based applications based on any possible grouping of codes.
- codes dramatically easier to remember (first five chars relate to real areas). Would have resulted in much faster adoption.
- provides anonymous code similar to other countries postcodes (first five characters) for situations where privacy demands it. Five chars could be marketed as postcode. Seven chars as location code. Eg 5 char code adequate for locating stores, services etc. This would have dealt with previous data commissioner's objection to Eircode.
- For simple cases of businesses and voluntary orgs the structure defined by large and small areas would be usable directly without needing to license special software or Eircode database. Small businesses could use simple maps for sorting and routing deliveries/service calls. If licensed software adds value to this, then all the better, just don't force everyone to license something.

Disadvantages

- if small area boundaries change then codes would have to change. This has been discussed at length and wouldn't happen anything like as frequently as people say. In general, new developments result in new postcode (small) areas, not affecting existing postcodes. Doesn't seem to be a major problem in UK.
- postcode snobbery. Also discussed extensively, and again this problem is overstated and most people in UK don't care about it. At end of day, it shouldn't have been the postcode contractor who decided on such an important architectural issue. It should have involved public consultation.
- doesn't follow An Post structure. But An Post never needed a postcode. The postcode working group originally recommended that it shouldn't follow An Post's structure. The absurd disparity in shape and size of Eircode routing keys only proves this point. (population varies from hundreds to probably over a hundred thousand)

Sam Russell · 11-02-2016 8:16pm

I think that design is workable and better than Eircodes design.

However, I would prefer a purely numerical postcode - for at least the first 5 characters (the public bit that ). I would use the Telecom STD codes as the base for the numbers, dropping the leading zero. It might be worth tidying them up a bit, but it is a good basis and would feed on peoples existing knowledge.

However, the biggest failure of Eircode was the lack of any attempt to solve the non-unique address problem. It is the combination of gobbledygook post code coupled with non-existent identifying markings for 30% of addresses - even townlands are not currently signposted in most counties.

What is needed is for official addresses to be the responsibility of some arm of the Government - either local or national. Then An Post and others will have to live with it.

Eircode discussion

Comments