Toby Stevens on ID Card contracts

Toby has a couple of good items today on ID card-related matters, here. One is on ID Card contracts, and the other is about false positives/threshold settings in the facial biometrics pilot at Manchester airport.

On the latter, I would take issue with Toby’s assertion that “[t]he scheme is still a trial, so there will be a great deal of useful data coming in through the systems, and this will improve performance in the future.” Toby also refers to the Iris scheme – and I think that’s the counter-example, right there. Whatever useful data has been gained from the initial years of the Iris project, it does not seem to me to have been used to improve performance of the system.

System performance, of course, is composed of many factors – such as the throughput and performance rates of the components (such as the Iris gates, the database, the link between the two, and so on), the competence of the users, the failure/fallback modes etc. etc.

In my experience, the component performance of the Iris gates themselves has remained pretty constant and, all things being equal, they have a maximum throughput rate roughly 25% of that of a human passport officer.

User competence (or lack of it) is the single biggest performance-killer. The worst thing that can happen to someone waiting in the queue for Iris is to find that the person in the gate cannot act on the commands “Please look into the mirror” and “Please move back a little”.

That is aggravated by the lack of a well-supported fallback mode. When someone manifestly cannot follow these instructions sufficiently well, there are no staff on hand to offer any additional guidance, so the user has no alternative (being locked in a glass booth) but to stand there while the gate exhausts its sequence of failure steps and finally spits the user out (air-side) to join the back of the non-Iris queue.

All this was true on Day One of the Iris scheme, it was true when I blogged about it in 2007, and it’s still true today. Sorry, Toby, but I think you may be being over-optimistic about the facial biometrics pilot…


2 thoughts on “Toby Stevens on ID Card contracts

  1. What’s the purpose of a trial? To improve a scheme or disprove it?

  2. Robin Wilton says:

    Good question. I remember going through “Project Management 101” at IBM, and being advised to avoid describing an early-phase project as a ‘pilot’, on the grounds that a ‘pilot’ is sometimes as far as it goes. Far better, we were told, to call it something like “Phase Zero”… because of the strong implication of subsequent non-zero phases.

Comments are closed.