Feeds

Clarke's ID card cost laundry starts to break surface

Hide the billions, levy transaction fees and voila! It only costs £30

  • alert
  • submit to reddit

Security for virtualized datacentres

The ever-watchful Spyblog recently unearthed some signposts to true reader costs, flagging a piece of scheme cost laundering while it was about it. The RIA for the Immigration, Asylum and Nationality Bill puts the cost of the initial deployment of biometric passport and visa readers as £3-5,000 per reader plus the cost of a PC, plus £21,000 for the computer network cabling.

The latter is clearly a one-off cost per location, so it's an initial deployment hit the 47 main airports and ports will only have to bear once, but it's also a cost that is likely to be incurred at most of the other sites where an online verification capability is required. As much of this expenditure will be absorbed by other departments via their IT budgets, or incurred by major financial institutions and ID verification third parties, little if any of it will ever be accounted for under ID scheme costs. One might also speculate about the costs of whatever it is the new cabling at ports connects to, and who's paying for that - unhelpfully, the RIA seems not to mention this bit.

The actual costs incurred in association with the Immigration, Asylum and Nationality Bill will be shouldered by the Immigration and Nationality Directorate, and will ultimately be a lot higher than those specified in the RIA, which states:
"Currently there are 47 major ports of entry and an average of 20 desks per location. Due to the staged implementation of biometric identifiers in passports, ports will only have a relatively small percentage of arrivals with biometrically enabled passports. Initially, we may only provide one reader per port. However, as biometrically enabled passports become more common we will increase the numbers of readers per port accordingly. If every desk at every port were to have a reader, Border Control would have to deploy the biometric solution at 940 desks at airports, seaports and the Juxtaposed Control."

It's difficult to get your head around the logistics of the planned 'one per port' initial deployment. Initially the RIA anticipates there may be two readers per port, one handling biometric passports and one handling biometric visas. It is true that numbers for passports will initially be quite small, but once the major economies are starting to ship biometric passports, the number will be ramping fast as old passports expire. European travellers whose passports could be read biometrically will therefore be coming into the UK in fairly large volumes in fairly short order. Biometric visas are intended to be issued to all visitors requiring visas within the next couple of years, and they will therefore constitute a very large volume, very soon.

So what on earth do the loves propose to do with the one appropriate reader available? Clearly it will not be a case of scanning all of the people with biometric passports or visas (try this at Waterloo or when the morning flight from Frankfurt comes into Heathrow), and the machines will be used purely to deal with those 'randomly stopped' or who have aroused the suspicions of immigration staff.* This does not differ greatly from the system as it currently stands. Also note that the RIA appears ("one [reader] for biometrically enabled ID cards") to view visa and ID card readers as the same thing, and therefore to envisage reading ID cards at ports of entry. So perhaps UK citizens could, like other EU citizens, travel within the EU on an ID card, no passport required. The Register has floated this notion before, and soon we may be told.

More broadly, the LSE attempts to nail down major areas of Government cost underestimation in the Cost Projections (Chapter 17, page 241) of its report. Aside from areas we've covered here already, the cost of the National Identity Register and integration costs are likely to be the most substantial additions. The LSE points out that there are clear parallels between the proposed NIR and the NHS spine, but that the former (for obvious reasons, we are continuing to avoid, with a growing sense of futility, calling it "The Register") will involve "greater complexity and must embrace more rigorous security measures. It must also incorporate biometrics - something that we believe will be a technological challenge far greater than the Government has anticipated." The LSE group has therefore put the cost of the NIR at between two and four times the contract price of the NHS spine.

This is one of the most dramatic and potentially contentious variations between the LSE costing and the Government ones. The LSE's reasoning however seems sound; according what's written on the tin, it is more complex and challenging than the NHS spine. So, if this is not reflected in its claimed total tab for ID scheme, the Government needs to explain either why the cost is lower, or where in the Government's IT budgets it is reflected.

Oddly enough, although the LSE's minimum cost estimate is approximately double the Government's most recent estimates, given the existence of some kind of cost-laundering iceberg beneath the visible aspects of the Government estimate, the real numbers might not be that different. If, somewhere within the Government, someone is tallying up the total cost of the scheme and all its related components, then the big number would quite probably fall within the LSE range of £10.6-£19.2 billion).

Given how career-threatening such a tallying exercise might be, we very much doubt that anybody's doing it. But if you think about it, it's exactly what the Government should be doing, then putting the facts before the country and Parliament before embarking on such a scheme. As opposed to the current approach of "capping" card cost at "only" £30, and avoiding telling anybody, probably including themselves, what it will all really cost. ®

* Borderwatch We at The Register take an understandable interest in developments in what really happens at UK border checkpoints, the Eurostar London to Paris run being particularly fascinating, given the quantity of shouting about illegal immigrants on this route there was a few years back. Recently we observed at Paris that the UK checkpoint had started putting the machine readable section of the passport into a reader (well done chaps, even if it has taken you nigh-on 20 years to start), but that only the French post used a forgery detector.

At Waterloo incoming practically everyone on the train was waved through without a check, so we're clearly banking on nobody screwing up at the Paris end. In the other direction, no UK official at all even asked for a passport, but there was a nice new checkpoint checking them, again using a forgery detector. It was operated by... the French. No doubt they're telling us how many of our terrorists are leaving the country.

Public contributions There's more than one way for the general public to contribute to the cost of the ID scheme. You can shut up and pay your taxes, thus helping meet the cost of the scheme, and you can contribute to the cost of the scheme by making it higher. The LSE (hinting, perhaps, at the institution's glorious past) deems non-cooperation as a potential cost, and suggests that one "dedicated non-cooperator", working "strategically and systematically can, quite feasibly, exhaust 200 hours of administration time through the generation of queries, appeals, access requests, database modifications and general civil disobedience." At time of writing the No2ID pledge to refuse an ID card was closing fast on its target of 10,000 signatories, which could mean an awful lot of time-consuming civil disobedience. Click below to nudge it over the 10k, of you haven't signed up already.

Click here to sign the no2id pledge

Related links:

LSE ID card report
Biometrics won't deter passport fraudsters, chief admits
Soaring card cost headlines threaten UK ID scheme
UK to outsource biometric visa checks to Mumbai
Cost of ID Cards could triple, plan could breach DRA

Choosing a cloud hosting partner with confidence

More from The Register

next story
Ex-US Navy fighter pilot MIT prof: Drones beat humans - I should know
'Missy' Cummings on UAVs, smartcars and dying from boredom
Facebook, Apple: LADIES! Why not FREEZE your EGGS? It's on the company!
No biological clockwatching when you work in Silicon Valley
The 'fun-nification' of computer education – good idea?
Compulsory code schools, luvvies love it, but what about Maths and Physics?
Doctor Who's Flatline: Cool monsters, yes, but utterly limp subplots
We know what the Doctor does, stop going on about it already
'Cowardly, venomous trolls' threatened with TWO-YEAR sentences for menacing posts
UK government: 'Taking a stand against a baying cyber-mob'
Happiness economics is bollocks. Oh, UK.gov just adopted it? Er ...
Opportunity doesn't knock; it costs us instead
Sysadmin with EBOLA? Gartner's issued advice to debug your biz
Start hoarding cleaning supplies, analyst firm says, and assume your team will scatter
Don't bother telling people if you lose their data, say Euro bods
You read that right – with the proviso that it's encrypted
prev story

Whitepapers

Forging a new future with identity relationship management
Learn about ForgeRock's next generation IRM platform and how it is designed to empower CEOS's and enterprises to engage with consumers.
Cloud and hybrid-cloud data protection for VMware
Learn how quick and easy it is to configure backups and perform restores for VMware environments.
Three 1TB solid state scorchers up for grabs
Big SSDs can be expensive but think big and think free because you could be the lucky winner of one of three 1TB Samsung SSD 840 EVO drives that we’re giving away worth over £300 apiece.
Reg Reader Research: SaaS based Email and Office Productivity Tools
Read this Reg reader report which provides advice and guidance for SMBs towards the use of SaaS based email and Office productivity tools.
Security for virtualized datacentres
Legacy security solutions are inefficient due to the architectural differences between physical and virtual environments.