House of Lords - European Union - Twenty-First Report

13. We received evidence, some of it conflicting, on the accuracy and reliability of PNR data. Professor Elspeth Guild, giving evidence on behalf of the Centre for European Policy Studies, made the point that "the quality of data which is collected for commercial purposes, the standards which are applied, are very different from those which are required for law enforcement …" (Q 95) It followed that the larger the number of data elements which were included, the greater the risk of inaccuracy. (Q 97) We have no difficulty accepting that, if there is a given probability of inaccuracy in a single data element, then the more data elements collected, the greater the likelihood of an inaccuracy. But it seems to us that the greater the amount of data collected, the better the chance that between them they will identify the individual accurately, even if some of the data relating to him or her are inaccurate. This was confirmed by Jonathan Faull, the head of the Commission Directorate General on Justice, Freedom and Security (JLS): "They [the Americans] would say that the more PNR you have, the lower the risk of making mistakes."(Q 143)

14. Mr Joaquín Bayo Delgado, the Assistant European Data Protection Supervisor, agreed that the more data one had, the more accurate the identification, but in his view that was not the issue. The true purpose of gathering PNR data was not simply identification, but also for security reasons such as attempting to deduce the intentions of the passenger. (Q 201) We can see that if a traveller is correctly identified but, for example, has the wrong seat number attributed to him which appears to place him in the company of a suspected criminal, this can be as dangerous as attributing the right data to the wrong person.

Data profiling and data mining

15. Most of the controversy surrounding PNR data is about the use to which they are put. We received evidence in particular about "automated profiling based on passenger data",[1] and about data mining programmes to obtain computer-generated risks assessment scores which aim to identify passengers who may pose a risk but who are not on any Government watch list.[2]

16. Data profiling can be described as the determination of characteristics or combinations of characteristics which might identify someone or something as being potentially worth investigation. Data mining is the use of advanced algorithms to trawl through huge databases to discover someone or something matching that profile. The Home Office give as an example the use by banks of software to trawl through millions of transactions to identify those matching predetermined profiles which may identify the transactions as fraudulent.[3]

17. In the context of PNR the profile is those combinations of characteristics which might identify an individual as being potentially high risk. The fact that a passenger matches that profile in no way determines that he is a criminal or even a suspect; only that there is a case for further investigation to see whether other information supports or negates the initial impression.[4] Although data mining can reveal patterns and relationships, it cannot reveal the significance of these patterns; it can identify connections between patterns in the behaviour or conduct of individuals, but it cannot identify causal relationships. These are matters left to those interpreting the information.[5] We give an example in paragraphs 24 to 27 of what can happen when they jump to the wrong conclusions.

18. Mr Faull told us that in his view the use of PNR data for data mining was a lawful and legitimate use of PNR data. (Q 160) We have not heard evidence to the contrary, though Dr Gus Hosein, a visiting scholar of the American Civil Liberties Union (ACLU), told us that Congress prohibits the use of funds to develop or test risk-assessment and profiling systems on passengers. (QQ 8-12) The use of these techniques in connection with data on United States citizens is something Congress has never approved and, in the view of Dr Hosein, never would approve. (Q 34)

The positive value of PNR

19. The degree to which the collection, retention and transfer of PNR data is acceptable depends of course on its value in combating terrorism and other serious cross-border crime, but there is a major obstacle to the assessment of that value. The more serious the crime, the more reluctant the authorities are to disclose details of what information was used, and in what way, to prevent its commission or to arrest and bring to trial those suspected of committing it. Even when a case comes to court, the prosecuting authorities have to disclose only such evidence as is essential for them to prove their case or as the law requires them to disclose to the defence; and this will not necessarily include all the information about the data and methods that have led to the identification of the suspects.

20. In a letter to the Chairman of 3 May 2007 Baroness Ashton of Upholland, the Parliamentary Under-Secretary of State at the Department for Constitutional Affairs (DCA),[6] has given us a number of valuable examples of the benefits of PNR profiling in identifying and disrupting human trafficking, and also an example of the exposure of a drug smuggling operation by PNR profiling. (p 25) But no examples were given to us of the use of PNR data in the fight against terrorism. We were not surprised to be told by Mr Faull that examples given to him (sometimes only in outline) of the benefits of PNR in combating terrorism were very highly confidential. (Q 140)

21. At a tripartite meeting in Berlin on 5 April 2007 between the United States (led by Mr Michael Chertoff, the Secretary of Homeland Security), the Council (led by Dr Wolfgang Schäuble, the German Minister of the Interior), and the Commission (led by Vice-President Franco Frattini), the United States delegation undertook to make public on an anonymous basis some of the security achievements which resulted from data collected by PNR.[7] On 14 May 2007 Mr Chertoff addressed the European Parliament Committee on Civil Liberties, Justice and Home Affairs—the LIBE Committee—on the PNR Agreements,[8] and made public an open letter to Members of the European Parliament giving examples of how the analysis of PNR data had prevented dangerous individuals from entering the United States. We print this letter in Appendix 3 to this report. We note that of the eight examples given, seven relate to serious crimes but not to terrorism. The first example explains how the murder of 132 individuals by a suicide bomber in Iraq provided evidence confirming that inspectors at Chicago airport had been right to refuse him entry to the United States on an earlier occasion. Clearly the use of PNR data did not on this occasion prevent an act of terrorism, though it may have prevented such an act within the United States.

22. We did not receive, and did not expect to receive, details of counter-terrorism operations in which PNR data were relied on successfully. We would however have hoped to receive sufficient evidence of the use of PNR to enable us to assess for ourselves the value of such data. We would not have expected such evidence to be given in public, nor would we have referred to it in this report. But it is an important principle of democratic accountability that Parliament should be able to reach its own conclusions, and not have to rely on statements from the executive. This would help to secure public confidence.

23. Nonetheless, having received no evidence to the contrary, we are prepared to accept that PNR data constitute a valuable weapon in the fight against terrorism and serious crime, and that their continued use is both necessary and justified.

What can go wrong

24. This assessment is based on the assumption that data are accurately collected and correctly analysed. Plainly, inaccurate PNR data can produce a false identification, and so attribute to an individual conduct and behaviour which is not his. A notorious example frequently given[9] is that of Senator Edward Kennedy, who was once forbidden to land in the United States because he shared a name with an individual on a watch list.

25. Even where correspondence of information produces an accurate identification, inaccurate use of PNR data can still wrongly attribute to an individual behaviour or conduct which is not his. The case of Maher Arar is an illustration of this.

BOX 1

Maher Arar

Maher Arar, a 34-year-old wireless technology consultant, was born in Syria and came to Canada with his family at the age of 17. He became a Canadian citizen in 1991. On Sept. 26, 2002, while in transit in New York's JFK airport when flying to Montreal, Arar was detained by US officials and, on the basis of information supplied to them by the Royal Canadian Mounted Police, interrogated about alleged links to al-Qaeda. Twelve days later, he was chained, shackled and flown to Syria, where he was held in a tiny
"grave-like" cell for ten months before he was moved to a better cell in a different prison. He was beaten, tortured and forced to make a false confession. He was eventually returned to Canada in October 2003. On 28 January 2004, under pressure from Canadian human rights organisations, the Government of Canada announced a Commission of Inquiry into the actions of Canadian officials. On 18 September 2006 Justice Dennis O'Connor cleared Arar of all terrorism allegations, stating he was "able to say categorically that there is no evidence to indicate that Mr. Arar has committed any offence or that his activities constitute a threat to the security of Canada".[10]

26. The authorities at JFK airport correctly identified Mr Arar, and accurately identified him as knowing a person being investigated by the Royal Canadian Mounted Police. They failed to investigate the degree of this acquaintance further, they made assumptions from it which were unjustified, and they took action which would have been unjustified even if he had been guilty of the most serious crimes.[11] On 26 January 2007 the Canadian Prime Minister in a statement in Parliament apologised formally to Mr Arar, and the Canadian Government has awarded him C$10.5m (£4.4m) in compensation, the highest settlement by the Canadian Government in an individual human rights case. The US authorities refuse to accept that he is innocent; to this day he is still on their no-fly list. (Q 47)[12]

27. This of course is an extreme case, but it is an example of what can happen when the right data are wrongly used. The principal risk of error in using PNR data seems to us to arise, not from the quality of the data, but from the erroneous interpretation of the data, even if accurate.

1 Letter to the Chairman of 30 March 2007 from Ms Joan Ryan MP, Parliamentary Under-Secretary of State, Home Office, p 19. Back

2 Letter of 9 January 2007 from Privacy International to Vice-President Frattini. Back

3 New Powers Against Organised and Financial Crime, Cm 6875, July 2006. Back

4 Cf Faull Q 160. Back

5 Report for Congress by the Congressional Research Service: Data Mining and Homeland Security: an Overview, 18 January 2007. Back

6 Since 9 May 2007 the Ministry of Justice. Back

7 Council Document No 8282/07. Back

8 Dr Schäuble and Vice-President Frattini were also present. Back

9 By (among others) Dr Gus Hosein (Q 34). Back

10 See www.maherarar.ca Back

11 See the official website of the Commission of Inquiry at www.ararcommission.ca for full reports of the events relating to Maher Arar, including the Commission"s conclusions and recommendations. Back

12 The Guardian, 27 January 2007. Back