|
Mr. Mole: The hon. Gentleman referred to the number of biometrics. Does he accept that the probability of mismatches decreases exponentially as the number of biometrics used goes up?
Mr. Allan: The increased number of biometrics allows double checking to take place, but that opens up other questions, too. I think that everyone accepts that a biometric system based purely on facial recognition would be not entirely useless; in fact, it would be a significant advance on where we are now. We use one of the best computing systems available to mankind; it lives in the head. Passport officers perform facial recognition to a very high degree of accuracy using their cerebral computing devices, and for many years we have accepted that that is a very good safeguard.
Similarly, if someone goes into a police control station, as I am sure most hon. Members here have done, they will see that the police have a wall of villains—pictures of the people whom they know are the villains operating in their area. The police memorise those, go out into the streets and do a pretty good job based on their human form of facial recognition technology. Biometrically recorded facial recognition added to our method of facial recognition is actually a significant advance. But if we are going for the levels of accuracy that the Government want, and if the Government want the results to be computationally accurate—that is important—they will go for iris recognition and fingerprints, too, as that will allow those things.
Using multiple biometrics opens up questions about the algorithm used for trying to deal with competing claims of those different biometrics. It is quite possible that, when someone goes through passport control, their face will be recognised but their irises will not, or that their irises will be recognised but not their face. There has to be a system to establish whether someone is allowed to pass on the basis of having gotten through one or two of the biometric tests but not the rest. Exploring those algorithms is important.
There are levels at which the systems can be deemed non-functional—in other words, levels at which they would be so much hassle that they would not work at all. If a passport control system based purely on facial recognition were set up, and the number of failures was very high and each failure required somebody to go off into a separate cubicle and check a person and their documentation and to go through a great bureaucratic process, that system would be unworkable. The queues at Heathrow would be unmanageable, and the system would grind to a halt.
There are levels at which systems are effectively unworkable; they are computed by the people who study such things. The biometric feasibility study establishes that a false match rate needs to be less than one in 1,000. It states that
''a false alarm rate of much above 1 per cent. would probably make the system unworkable.''
We need to find the right compromise between cost, hassle, accuracy and a system that is genuinely
Column Number: 141
workable. In order to avoid false alarms, one goes for more biometrics; if one goes for more biometrics, the time required to check them increases; and thus we return to the search for that delicate balance.
We should not necessarily accept as a given the Government's proposed scheme. It is important that they flesh out precisely how they intend to use biometrics. It is a key issue, and it must remain on the table as we discuss the Bill. Biometrics are potentially very intrusive and costly. They carry particular implications for people with disabilities, and if we are going to implement them, we must ensure that that is done in the right way.
I can envisage circumstances in which we get it wrong. We might have invested in all the costly biometrics, but because things have not been done in the right way they might not get checked. I can also imagine a situation arising—at Heathrow, for example—where there is an inflow that cannot be stopped; people will keep coming, and if the systems do not work, at some point the decision will have to be made just to let them through. The holy grail of a biometric system is to get to a situation where the necessary checks are performed but excessive checks are not, and people can be kept flowing through the system.
A common theme that the Committee will return to throughout its scrutiny of the Bill is whether the Government have gone for an overkill solution for something that I described as a ''nut.'' The hon. Member for Reading, West (Mr. Salter) took offence to that, and asked whether I was describing illegal working and other things as nuts. I was describing the checking of identity as the nut. In many contexts, it is quite a minor problem, and other problems are more serious.
There are different solutions for certain services. Even if one has a national identity register, it is not necessary to use the gold-plated, belt-and-braces, all-singing, all-dancing biometric system for everything. It would be perfectly possible to implement something like a chip and PIN system for some services, which is much easier and quicker to operate. Biometrics only where necessary is an important principle if one is seeking to reduce costs.
That was a canter around the biometrics issue. It needs to be seriously debated. The amendments knock out all the biometric bits to give us a debating peg to hang this discussion on. I hope the Minister will be able to clarify the Government's intentions and flesh out their thinking. In particular, I would like to learn of their response to the biometric feasibility study. It was a useful piece of work that informed all of us about how these complex technologies work. In general, it is written in English rather than techy speak, and it poses important questions that the Government must answer.
The Chairman: I shall use my cerebral recognition system to call Mr. Malins.
Mr. Malins: And what a great system it is, Mr. Conway.
Column Number: 142
I rise to speak on amendment No. 152, tabled in my name and that of my hon. Friend the Member for Newark. It relates to clause 9, but it has helpfully been grouped with amendments to clause 5. It would insert the words ''all of'' before ''his fingerprints''.
I pay tribute to the hon. Member for Sheffield, Hallam for his knowledge of and expertise in biometrics, which I do not share. As he rightly said, this gives us an opportunity today to probe the Government on this and related issues. I refer to the words of my right hon. Friend the shadow Home Secretary, the Member for Haltemprice and Howden (David Davis), on Second Reading before Christmas, when he referred to a number of tests or hurdles on which the Government would need to satisfy us as the Bill proceeded through Committee. He said that
''we must be clear about the technology's capabilities. Biometrics technology is not infallible.''
I pause only to say that I imagine that there are huge numbers of people in Parliament who are simply not familiar with the world of biometrics. They will find it difficult to understand the technicalities and would welcome a careful and detailed explanation from the Minister.
My right hon. Friend went on to another of the important tests that the Opposition are putting forward. He said:
''The question is whether the Government have in place the organisation necessary to introduce the scheme. We know that what they are proposing is one of the most ambitious technology projects that the country has ever seen. We also know that the Government have an abysmal record—I am not making a political point here, because most Governments have an abysmal record—in overseeing large-scale projects.''—[Official Report, 20 December 2004; Vol. 428, c. 1961–63.]
So there we have it. The Minister's task in this debate, as elsewhere, is to persuade those on the Opposition Benches of the twin issues of the effectiveness of the technology and of the capability of the scheme operators and the Government Departments involved. That will mean telling us a lot more than we know already about the details and certainty of the technology, about the companies, industries and experts who will forward the work and, incidentally, about the record of effectiveness and efficiency of some of those companies—subject to commercial confidentiality.
I was prompted to table this amendment by an article by Philip Johnston of The Daily Telegraph, shortly before Christmas, I think, who pointed out:
''The biometric technology is not foolproof and runs a high risk of false positive and false negative identifications. A feasibility study of the National Physical Laboratory says that in order to avoid false matches in a population of 50 million, a fingerprint system should use at least four fingers per person, preferably eight. An iris system should use both eyes and facial recognition could not on its own provide a sufficient accuracy of identification.''
During the past days and weeks a number of us on this Committee have been briefed by outside and interested organisations. Almost invariably, they have referred to the importance of biometrics—will they be effective, safe and secure? Can they be foolproof? The Law Society is one body raising doubts about whether biometrics will work accurately. There are others. The Minister may well have seen their papers, which
Column Number: 143
suggest that there is a level of unreliability in the most sophisticated current biometric checking procedures.
I hope that the Minister has a chance to refer to the Cabinet Office document to which I was referred. I am told that the Cabinet Office found that 10 to 15 per cent. of genuine people failed biometric tests set at the highest level of corroboration, which is really quite high. That percentage, which comes from page 61 of the Cabinet Office document from July 2002 entitled ''Identity fraud: a study'', is really quite high. Accuracy rates will become crucial when magnified by the intention to include the entire population on the register.
10.45 am
The Government, perhaps properly, claim that some biometric methods are extremely secure and reliable forms of identification. They are plainly keen on fingerprints and iris scans to establish one's identity, or at least one's uniqueness. Will the Minister be able to say whether biometrics are really secure?
It is said—I refer in part to the report by Privacy International—that all biometrics have been successfully spoofed or attacked by researchers. A great deal of work has been undertaken to establish the technique of forging or counterfeiting fingerprints, and researchers in Germany have established that iris recognition is vulnerable to simple forgery. I have read that proposition in several places, and I hope that the Minister will dwell on it at length in his response.
It is, of course, worth considering other schemes around the world to ascertain what they have done. Has a scheme similar to ours been proposed or adopted in other countries? Yes; Malaysia, Singapore and Thailand have very similar card systems. Significantly, I am told that China is moving rapidly in the same direction; it has developed a compulsory ID database and card system, but—and here is the rub—I understand that it abandoned the biometric element after concluding that the technology was unworkable with large populations.
Does the Minister know about that? Can he comment on it? I have summarised what was said to me in a report. I do not know what the population of China is; clearly, it is much more extensive than ours, but none the less, we have a very large population here. Does he know anything about how things went wrong in China and whether we face any of the problems that it had?
| 
