Debunking the AutoAddress Debunking!

A Review of Autoaddress’s Attempted Debunking of the Independent Assessment of Eircode’s Significant Flaws

AutoAddress

“Tell me again how much you’re are blowing on this?”
“€27,000,000”
“…and it’s designed to fail”
“Yep”
“Ha ha ha ha ha ha ha ha ha taxpayer suckers!!!”

In April last, Professor Brian Lucey of TCD published an independent assessment of the major issues that are associated with the planned National Postcode, to be called Eircode. Almost immediately, an attempt at debunking these issues was published by a company called AutoAddress in Dublin. Of course, those watching expected a debunking of the debunking, but none was done because none was merited…until now, let us tell you why.

  1. Autoaddress is just a junior member of the postcode consortium which is contracted to deliver the National postcode.
  1. Members of Autoaddress are normally not used/allowed in public to present to potential industry users of Eircode or the media and all such presentations to date have been done by the senior members of the consortium;- Capita & Bearing Point.
  1. On its website, Autoaddress bravely claims that it designed Eircode but a design document released into the Public domain earlier this year under FOI clearly shows that all design considerations for Eircode were done by Capita & An Post together. Autoaddress is only involved in the consortium because it has been a reseller of An Post’s Geodirectory address database for several years and this is to be used as a basis for the address database that mistakenly has become the main focus of the National postcode development effort. Effectively, Autoaddress is the link between the senior players in the postcode consortium and An Post’s GeoDirectory agency staff.
  1. In April 2014, PA Consulting was contracted to oversee the design and implementation of Eircode on behalf of the State, with particular responsibility for identifying if Eircode is fit for purpose. For this reason, when a peripheral company such as Autoaddress puts its head above the parapet to defend Eircode, it highlights serious issues with the governance of the project. At this point PA Consulting should have published its fit-for purpose assessment for the Minister, DCENR, Capita or PA Consulting themselves to quote publicly, but instead, a company that claims it designed Eircode, but actually did not, finds itself having to defend the code because no-one else will;- especially not those who are paid to take responsibility on behalf of the Irish citizen. This alone highlights that Eircode, as a major National infrastructural project, has gone off the rails.
  1. Further evidence that compliance with legislation, governance, due diligence and best practice management of a major Government procurement project long since went out the window, comes from Autoaddress’s comment in its so called “debunking” that it itself “assessed alternative code technologies” for the postcode. Those that understand a procurement process will readily understand that a lowly partner in one bidding consortium does not normally assess the offerings of other potential bidders for the contracting authority. Where were DCENR and their helpers PA Consulting when this was going on and why had they abdicated their responsibilities to a bidder’s helper; thereby undermining all semblance of a fair & transparent procurement process. This alone makes it very clear that the “competitive dialogue” process for Eircode that was supposed to have happened, never did. It would appear, that there were no alternative bidders and those that might have had credible alternatives for the State were never considered by a fair and transparent process which should have used independent arbitration on competing designs.
  1. Knowing this background, the so called debunking by Autoaddress does not really merit attention but just in case the Minister, DCENR and PA Consulting are looking for some insights which they will not have received to date from contracted vested interests, let’s deal with  the main points:

In its so-called “Debunking”, Autoaddress claims that…

Loc8 isn’t free

DCENR claims in a letter to the Public Accounts Committee that a figure of €2m was paid for the design of Eircode. Loc8 Code was designed, developed and tested at Loc8’s own expense and has been offered to the State for free. Loc8 is tested and proven for multiple applications whilst Eircode has never been tested at all. Loc8 has been freely used by thousands in Ireland since 2010! Loc8 is available as a free and standard feature on all modern Garmin satnavs. The public of Ireland can look up and use as many Loc8 Codes as they wanted free of charge and without number restrictions through the Loc8 Code website. The same public can use Loc8 without relying on or maintaining up-to-date any database at all.

A competitive Dialogue Process was Run

As said already, when a bidder’s helper starts telling the public about the ins and out of a state procurement process as if they were part and parcel of it, when those who were supposed to be running it remain silent, then there is most obviously a serious lack of governance to be accounted for. Furthermore, having the same bidder’s helper assessing the offerings of potential competitors for the contracting authority just reminds us why the Moriarity Tribunal was necessary in the first place and why an 18 month long EC investigation concluded that the postcode tender which produced Eircode did not comply with EU legislation or International or National best practice. It is for these reasons that Eircode is the ongoing subject of interest for the Joint Oireachtas Committee on Transport and Communications, the Public Accounts Committee and the Comptroller & Auditor General.

No Trial of Eircode is Necessary as An Post is able to deliver post already

This comment indicates clearly why the FTAI has serious reservations about Eircode as their requirements were never considered or tested for. Equally, the requirements of those in the Emergency services have not been tested for either. It also makes it clear that An Post will not have much need to use Eircode even though they were the total focus of Capita’s design considerations. Every country in the world that has ever implemented a postcode has piloted it first before attempting a national rollout and this is even more important for a code that is claiming to be “historic” and a “world’s first”, especially in a country that already has had the PPARS and E-voting scandals! Even Nightline, which is conspicuously unique from the transport & logistics sector in voicing its support for any postcode that An Post supports, told an Oireachtas Committee that a pilot would be essential before rollout!

The Eircode Database is insignificant

This point is made by Autoaddress in the context of making the database available on navigation devices. They say the file size is comparable to that of Loc8 Code but Loc8 Code needs no database to function;- with Loc8 Code, the software extracts a location directly from the code itself. Obviously therefore, Autoaddress’s claimed assessment of Loc8 Code was not very comprehensive. Loc8 insists that nobody from Autoaddress ever spoke to them on the detail of how Loc8 is currently implemented as an elective address code or how it would be implemented as a National postcode. This of course gets away from the fact that consumer navigation devices have their limited memory well allocated already and suggesting that they would readily give over any percentage to an unproven and unpopular Irish postcode without a significant cost, displays the naivety of a company that may well never have really consulted the navigation industry on the matter in the first place. The very fact that no marketing for Eircode has ever claimed support from the big navigation/GPS companies says a lot.

For €60 you can purchase an Entire copy of Eircodes & addresses

In recent pricing for Eircode released by Capita, a €60 cost does not feature at all. However, most would be aware that you can purchase a navigation app from a major name in navigation with all of Ireland and UK mapping, including UK postcodes, for a lot less than that!

Grid-based postcodes are incapable of providing this functionality (grouping), because they ignore accessibility

Grid based codes where those which proved most popular by all industries in the small bit of consultancy that was done by PA Consulting on a Postcode back in 2005. As geographically adjacent Loc8 Codes look similar, visual grouping, and indeed grouping by Excel, are more than possible. Furthermore, Eircode is only for buildings where An Post delivers mail and then only for the geometric centre of those buildings, ignoring road access completely, whilst a system like Loc8 can support a code which actually identifies the road access of any delivery point instead. Again this undermines claims that Autoaddress fully assessed alternative codes such as Loc8 Code. In making this statement, Autoaddress also ignores the fact that Eircode can only be grouped using the routing key which may cover a whole city area; – making rational grouping impossible for any normal logistic process. The table below compares sample Eircodes (published to the web by Gamma which is a sister company of Autoaddress) & the related Loc8 Codes:

House Number Eircode Loc8 Code*
104 D12 Y7F5 NP5-61-TL7
102 D12 W97N NP5-60-TL7
100 D12 TQ45 NP5-51-TL7
98 D12 V4H8 NP5-50-TL7
96 D12 BHU7 NP5-41-TL7
94 D12 987KF NP5-40-TL7
92 D12 X36M NP5-31-TL7
90 D12 AQ45 NP5-30-TL7
88 D12 M8D3 NP5-21-TL7
86 D12 H8D9 NP5-20-TL7
84 D12 5FG8 NP5-11-TL7

* Yes the Loc8 codes are valid codes and the links above are working hyper-links.

Firstly, it will be noted that there is nothing visually conspicuous in the Eircodes shown to indicate that they are all on the same street and adjacent to each other and that some are actually physically adjoined! It will also be noted that the only grouping visually identifiable in the Eircodes is for the first 3 characters (the Routing key) which, in this case, is the well-established Dublin 12 (D12) postal area. This area extends over a large part of Dublin City and many thousands of properties. Whilst details of routing keys for the rest of the country have never been published, the design document produced by Capita & An Post suggests that all of Limerick City will have the same routing key. This means that the only visual grouping that can be done for an Eircode will be for vast areas which have no practical application other than for An Post. It should also be noted that according to Autoaddress, 100,000 addresses have an An Post postal address which indicates the wrong county. The Eircode routing key will maintain and reinforce this misleading information so that even simple visual validation and grouping by relating the routing key to the actual county will not be reliably possible either.

On the other hand, however, it is immediately visually conspicuous that there are two grouping possibilities for the Loc8 Codes shown;- firstly, in the zone of NP5 (roughly 3.5km wide) and then in the locality of TLF (only 120 metres wide). Furthermore, and unlike Eircode, NP6 is adjacent to NP5 and TM7 is adjacent to TL7, thereby allowing a user to iterate and approximate the general proximity of one code to another. Furthermore, the location ID number in the middle of the Loc8 Code is geometrically sequential so that if a property was built between NP5-60-TLF and NP5-51-TLF there is natural geometric redundancy to accommodate that; – once again undermining the myth put about that sequential codes cannot be used for a postcode. It will also be noted that absolute sequential numbers are not required;- just an indication whether you are getting closer or further away. This means that Loc8 Codes can be learned and iterated and used in a non-technological and practical way in the same way as the UK postcodes have been used for over 50 years. However, Eircode has none of these useful real-world, “Sean & Sinead” or the “Man in the Van” type features. Incidentally, one of the reasons why the location ID numeric element of the Loc8 Code is in the middle is because research done for the UK postcode indicated that breaking up strings with numbers helped memorability and helped with the natural pentameter of the code. Loc8 also uses the benefit of research on the idea of “chunking”, i.e. breaking up character strings into chunks to aid memorability, with the ideal maximum number of characters in any one chunk being three. Eircode has apparently not followed any published research in relation to these important design considerations and none are sited in the published design document.

LocateMe112 replaces the need for a National postcode which is useful for public safety

It is clear that Autoaddress does not understand that it is not for it to determine national public safety communications policy. This role is undertaken by COMREG who have ongoing consultation on the subject of Caller Location ID, some of which is published at this  link: http://www.comreg.ie/_fileupload/publications/ComReg14110s.pdf

Interestingly in this documentation, key service providers for Emergency Response in Ireland state that they know every little about Eircode; – yet again belying claims that all relevant industries were widely consulted in the design of Eircode. Also in this document the point is well made that In the complete emergency management scenario, automated transfer of information is not always possible and therefore the means by which information, including location information in this case, is managed must be carefully specified and standardised”, meaning that electronic solutions such as Locateme112 are only one part of the emergency management picture and to complete the picture, a means by which location can be unambiguously manually & verbally communicated for all possible locations must also be specified. This further emphasises the very narrow and blinkered perceptions, without the enlightenment of a meaningful consultative process, that have guided the errant design (or post design justification) of Eircode.

Loc8 “checker” does not check the two digits in the middle of a Loc8 code, so if digits were reversed, e.g. 19 instead of 91, then you could easily end up the wrong side of a river

Loc8 has a “checker Character” which not only immediately indicates a likely error but also suggests a solution. Eircode has no such capability and relies on a cross reference with a database to help verifications instead. And, that database recognises the possibility that any one address can have multiple aliases (variations) some of which are in ECAD but not ECAF;- thereby rendering ECAF useless for any kind of code validation. With ECAD and all its aliases, once you recognise an alias at all, you recognise the possibility for multiple possible variations for the same address (including new ones not yet in the database), thus complicating further any validation process. Encouraging random address aliases only exacerbates the problem and with a visually random code, neither a code nor an address validation may ever be possible. Autoaddress suggests that the number 19 is easily mixed up with 91. There is no obvious research evidence suggesting that this would be a normal occurrence and most would be aware that it rarely happens even with house numbers. However, if it did, so also would it happen with D02 and D20, which are Eircode routing keys, and such an error would cause a serious location error of at least 10km whilst getting the centre numbers of a Loc8 Code wrong would only have you at worst 120 metres away from where you should be!

Eircode is a public database, accessible to all

And therein lies the problem. Autoaddress tells us Eircode is a database whilst the Irish public thought they were getting a publicly useable postcode that would make some sense to them without looking up some obscure and technical database.

Rock of Cashel, Newgrange, Hill of Tara, etc. will all have
an Eircode…they have visitor centres, etc. that receive post

So if you are in one of the thousands of buildings in agricultural, forestry, health, educational, industrial, sporting, tourist, heritage, religious, outdoor activity or administrative  centres, campuses or sites around the country that do not receive mail from An Post, then you now know how to get yourself an Eircode;- just build an interpretive centre for yourself! In Northern Ireland and everywhere else in the UK, even if a site does not receive mail, it still has a useful postcode and millions of Euros are never needed for an interpretive centre to connect the postcode to the public!

A successful postcode design must enable efficient manual sorting of post, otherwise it isn’t a postcode

Yes, An Post told the Oireachtas Committee hearing that Eircode will allow manual sorting of mail at Christmas for temporary staff but could not identify any other specific use for the code. The fact that no temporary or permanent staff member could manually interpret the visually random unique identifier element of Eircode (see the bale of codes above again – the random last 4 characters are the unique identifier) was completely overlooked in the testimony given to the Oireachtas Committee and in Autoaddress’s more recent desperately flawed debunking efforts.

Loc8 cannot be modified if it inadvertently assigns an offensive code to a dwelling

Loc8 does not assign offensive codes to any location. It is in use for 5 years without any credible examples or complaints. Loc8’s structure was engineered so that two 3 character alphanumeric strings are separated by a 2 number string and characters are restricted so that key components of actual dictionary words cannot appear together.

Conclusion

All of these points clearly highlight that Autoaddress has a very poor understanding of the necessary design characteristics of a modern postcode and the actual design characteristics and features of Loc8 Code. They also identify why Autoaddress’s attempted debunking of the published assessment of Eircode’s flaws does not merit real consideration. It is also of note that Autoaddress continually compares Eircode to Loc8. Eircode has benefitted from the full political, financial, legislative and administrative support of the State but even after 17 months of effort has repeatedly missed all deadlines and has not yet demonstrated a working and, independently declared, fit-for-purpose postcode. Whilst Loc8, on the other hand, is a start-up SME which, using its own limited resources, delivered a working address code which has proven itself over 5 years already. Loc8 was never implemented as a National postcode and its plans as to how it would achieve that kind of implementation have never been sought or heard!

1 comment on “Debunking the AutoAddress Debunking!”

  1. Pingback: Debunking the AutoAddress Debunking! | Get Lost Eircodes | Brian M. Lucey

Leave A Reply

Your email address will not be published. Required fields are marked *