Introduction
We are living in what is usually described as an “information society,” and as the business community makes ever greater use of computers, the courts will find that increasingly, the disputes before them turn on evidence that has, at some stage, passed through or been processed by a computer. In order to keep in step with this practice, it is vital that the courts are able to take account of such evidence. As the Criminal Law Revision Committee recognized, “the increasing use of computers by the Post Office, local authorities, banks, and business firms to store information will make it more difficult to prove certain matters such as cheque card frauds, unless it is possible for this to be done from computers” (CLRC 1972, para 259).
Admissibility
The law of evidence is concerned with the means of proving the facts that are in issue, and this necessarily involves the adduction of evidence that is then presented to the court. The law admits evidence only if it complies with the rules governing admissibility. Computer output is only admissible in evidence where special conditions are satisfied. These conditions are set out in detail in section 69 of the Police and Criminal Evidence Act (PACE) 1984 (see further Nyssens 1993, Reed 1993, and Tapper 1993).
In general, the principles of admissibility are that the evidence must be relevant to the proof of a fact in issue, to the credibility of a witness, or to the reliability of other evidence, and the evidence must not be inadmissible by virtue of some particular rule of law (Keane 1994, pp. 15-20; Tapper 1990, pp. 51-61).
Real evidence usually takes the form of some material object (including computer output) produced for inspection so that the court may draw an inference from its own observation as to the existence, condition, or value of the object in question. Although real evidence may be extremely valuable as a means of proof, little if any weight attaches to it unless accompanied by testimony that identifies the object in question and explains its connection with, or significance in relation to, the facts in issue or relevant to the issue.
This is illustrated in the case of R v Wood (1982) 76 Cr App R 23 where the appellant was convicted of handling stolen metals. In order to prove that metal found in his possession and metal retained from the stolen consignment had the same chemical composition, cross-checking was undertaken, and the figures produced were subjected to a laborious mathematical process so that the percentage of the various metals in the samples could be stated as figures. This was done by a computer operated by chemists. At the trial, detailed evidence was given as to how the computer had been programmed and used. The computer printout was not treated as hearsay but rather as real evidence, the actual proof and relevance of which depended upon the evidence of the chemists, computer programmer, and other experts involved.
The difficulty in the application of this rule lies in its interaction with the hearsay rule. Evidence is hearsay where a statement in court repeats a statement made out of court in order to prove the truth of the content of the out of court statement (Sparks v R 1964 AC 964). Similarly, evidence contained in a document is hearsay if the document is produced to prove that statements made in court are true (Myers v DPP 1965 AC 1001). The evidence is excluded because the crucial aspect of the evidence, the truth of the out of court statement (oral or documentary), cannot be tested by cross-examination.
The problem, however, occurs because some statements, although in form assertive and inadmissible if they were to originate in the minds of human beings, in fact originate in some purely mechanical function of a machine and can be used circumstantially to prove what they appear to assert. The basis for this view was laid down in a case having little to do with computers. In the Statute of Liberty 1968 2 All ER 195, a collision occurred between two vessels on the Thames estuary. The estuary was monitored by radar, and a film of the radar traces was admitted into evidence. Simon P rejected the argument that the film was hearsay – he held that it constituted real evidence and not hearsay, and he placed it on a par with direct oral testimony. Where machines have replaced human beings, it makes no sense to insist upon rules devised to cater to human beings but rather, as Simon P said, “the law is bound these days to take cognizance of the fact that mechanical means replace human effort” (at p 196).
This useful distinction was apparently overlooked in R v Pettigrew (1980) 71 Cr App R 39, where the prosecution wished to prove that some banknotes found in the possession of the accused were part of a particular consignment despatched by the Bank of England. A computer printout was used to prove this, but the Court of Appeal held that such evidence was inadmissible under the statutory provision concerned (section 1 Criminal Evidence Act 1965 – now repealed). The Court took the view that the operator did not have the requisite personal knowledge of the numbers of the banknotes rejected from the machine since they were compiled completely automatically by the computer. This conclusion is quite accurate and a perfect application of the hearsay rule, but it failed to consider the use of the print-out as real evidence. This confusion between hearsay and real evidence is unfortunate and it may explain why it was necessary to create special rules for computer evidence.
Criminal Proceedings It is imperative that computer output should be readily used as evidence in criminal cases since otherwise, many cases, particularly those involving dishonesty, would be immune from prosecution. At the same time, one cannot be too complacent about the technology since computers are not infallible. It is widely acknowledged that “hacking” and “viruses” may affect information stored on a computer. These factors were obviously taken into consideration when enacting the provisions governing computer-generated evidence in criminal proceedings.
Section 69 of the Police and Criminal Evidence Act 1984 provides that: “(1) In any proceedings, a statement in a document produced by a computer shall not be admissible as evidence of any fact therein unless it is shown—(a) that there are no reasonable grounds for believing that the statement is inaccurate because of improper use of the computer and; (b) that at all material times the computer was operating properly or, if not, that any respect in which it was not operating properly or was out of operation was not such as to affect the production of the document or the accuracy of its contents.” In addition, any rules of court made under section 69(2) must also be satisfied (at the time of writing, no such rules have been made).
Real evidence and hearsay
So far, the discussion has focused on exceptions to the hearsay rule. However, evidence derived from a computer constitutes real or direct evidence when it is used circumstantially rather than testimonially. That is to say when the fact that it takes one form rather than another makes it relevant, rather than the truth of some assertion which it contains. (3)
Direct evidence produced by a computer is not subject to the hearsay rule. As we have already noted, in R v Wood, calculations were carried out by a computer specifically for the purpose of the trial to verify whether the composition of stolen metals matched original metals. Computer output was admissible as real evidence since it did not purport to reproduce any human assertion which had been entered into it. It was held that the machine was a tool, and that in the absence of any evidence that it was defective, the printout, the product of a mechanical device, fell into the category of real evidence. The court did recognise, however, that the dividing line between admissibility of computer-generated evidence as real evidence or hearsay would not always be easy to draw.
The same distinction and result were reached in Castle v Cross [1985] 1 All ER 87 and in R v Spiby (1990) 91 Cr App R 186, CA, an automatic telephone logging computer which logged the call details without human intervention was admitted as real evidence. The Court also held that, in the absence of evidence to the contrary, courts would presume that such a computer was in working order at the material time.
Thus, as far as the common law is concerned, the status of computer evidence as real or hearsay will depend, in each case, on the content of the computer record, the reason for using it in evidence, and the way in which it was compiled. Cases like R v Wood and R v Spiby, however, must now be read in light of the decisions in R v Shephard [1993] 1 All ER 225, HL and R v Cochrane [1993] Crim LR 48, CA.
In R v Shephard, the House of Lords held that section 69 of PACE 1984 imposes a duty on anyone who wishes to admit a statement in a document produced by a computer to produce evidence that will establish that it is safe to rely on the document. Such a duty cannot be discharged without evidence by the application of the presumption that the computer is working correctly, expressed in the maxim “omnia praesumuntur rite esse acta”. It makes no difference whether the statement is or is not hearsay. In R v Cochrane, it was held that before the judge can decide whether computer printouts are admissible, whether as real evidence or as hearsay, it is necessary to call appropriate authoritative evidence to describe the function and operation of the computer. In that case, the prosecution wanted to prove that certain cash withdrawals were made from a particular ‘cashpoint’.
The machine would only dispense money if the correct Personal Identity Number was entered. The matching was carried out by a mainframe computer, and evidence of its proper functioning was thus required by the court. The prosecution did not adduce this evidence, and the conviction was set aside on appeal.
As we have seen, a printout from a computer that has been used as a calculating device, or which records information automatically without human intervention, is admissible as real evidence and involves no question of hearsay. On the other hand, where the printout contains information supplied to the computer by a person, it is hearsay if tendered for the truth of what is asserted, but may be admissible under either sections 23 or 24 of the Criminal Justice Act 1988. A statement can only be admitted under sections 23 or s 24 if its maker (or the original supplier) had (or may reasonably be supposed to have had) personal knowledge of the matters dealt with. Furthermore, under section 24, the ‘creator’ of the document must have been acting in the course of a trade or business, etc.
A statement in a computer printout that has satisfied the foundation requirements of sections 23 or 24 can only be admitted on satisfaction of the additional requirements contained in section 69. Section 69 is couched in negative terms, making it clear that evidence that does not satisfy its requirements is inadmissible. The object of section 69 is to impose a duty on anyone who wishes to introduce a document produced by a computer to show that it is safe to rely on that document, and it makes no difference whether the computer document has been produced with or without the input of information provided by the human mind and thus may or may not be hearsay (per Lord Griffiths in R v Shephard at p 228). The operation of section 69, therefore, is not limited to printouts that fall within sections 23 or 24 of the 1988 Act.
Reliability: If there is a dispute as to the admissibility of a computer printout in a criminal case involving a jury, the judge should hold a voir dire. A party seeking to admit a printout under section 24 (or section 23) must establish the foundation requirements of both that section and section 69. The judge, in deciding whether the prosecution has established these requirements, should apply the criminal standard of proof. Although, as we shall see, the additional requirements of section 69 can be proved by certificate, the foundation requirements of section 24 (or section 23) must be proved by evidence unless the other party makes admissions or allows the statement to be read.
There is also a third common law requirement before the judge can decide on admissibility, namely, that appropriate authoritative evidence must be adduced to describe the function and operation of the computer (e.g., R v Cochrane).
In R v Governor of Pentonville Prison ex p Osman 1989 3 All ER 701, it was argued that printouts were inadmissible because the prosecution had failed to prove the proper operation of the computers required by section 69. However, Lloyd J held that “where a lengthy computer output contains no internal evidence of malfunction…it may be legitimate to infer that the computer which made the record was functioning correctly” (p 727).
In R v Shephard, the House of Lords held that it will very rarely be necessary to call an expert to prove that the computer is reliable. The defendant was charged with theft from a store. A store detective gave evidence that she had examined all the till rolls for the relevant day from the tills, which were linked to a central computer, and that they contained no record of the unique product code for some goods found in the defendant’s possession. She also said that there had been no trouble with the central computer. On appeal, it was argued that the evidence did not satisfy section 69 since oral evidence that the computer was operating properly is not admissible unless given by a person qualified to sign the certificate under para 8(d) of Schedule 3, which provides that: “In any proceedings where it is desired to give a statement in evidence in accordance with section 69 above, a certificate -…. (d) purporting to be signed by a person occupying a responsible position in relation to the operation of the computer, …..shall be evidence of anything stated in it; and for the purposes of this paragraph, it shall be sufficient for a matter to be stated to the best of the knowledge and the belief of the person stating it.”
Dismissing the appeal, it was held that section 69 can be satisfied by the oral evidence of a person familiar with the operation of the computer who can give evidence of its reliability and need not be a computer expert. Lord Griffiths said that: “Computers vary immensely in their complexity and in the operations they perform. The nature of the evidence to discharge the burden of showing that there has been no improper use of the computer and that it was operating properly will inevitably vary from case to case. I suspect that it will very rarely be necessary to call an expert, and…in the vast majority of cases, it will be possible to discharge the burden by calling a witness who is familiar with the operation of the computer in the sense of knowing what the computer is required to do and who can say that it is doing it properly.”
This approach was adopted in Darby v DPP The Times 4 November 1994. The appellant had driven her car into an area of road subject to a 30 mph speed limit. At that point, a police speed trap was being operated. A police officer was operating a device known as a GR Speedman, and he concluded that the appellant had exceeded the speed limit by driving at 43 mph. It was submitted that the evidence of the reading of the GR Speedman was inadmissible if it was held to constitute a document. It was also contended that the evidence of the read-out required certification and that, while oral evidence of certification would be admissible, the police officer could not give such evidence as he was not an expert in the workings of the machine, only its operation.
Potts J adopted the approach of Lloyd LJ in the Shepard case and assumed that the machine was a computer and that the visual image was a document produced by a computer. He also referred to the principle enunciated in Lord Griffiths’s speech above and accordingly found no problem in holding that, on the basis of the evidence of the police officer, who was a trained and experienced operator of the device, the machine was working correctly. The appeal was consequently dismissed.
Thus, it seems that the provisions in section 69 are capable of being applied without undue difficulty. However, it is interesting to note that Rose LJ pointed out that if the GR Speedman had been central to this case and had produced a printout on which the prosecution had relied, then it may well have been caught by section 10(1)(c) (8) of the Civil Evidence Act 1968 (section 118(1) of PACE 1984 provides that a ‘document’ within that Act has the same meaning as in Part I of the CEA 1968). This would have meant that, as a document within the meaning of section 10(1)(c), it would have constituted a document requiring certification within the meaning of section 69 and the terms of para 8 of Sch 3. But it was the police officer’s opinion evidence which was central to the case, and that was capable of being corroborated by a technical device, the accuracy of which had been established. Thus, it appears that the conditions for admissibility for computer output in a criminal case are less demanding if the evidence provided by the machine is merely corroborative.
The ambiguities and illogicality arising from the complex conditions for admissibility of computer evidence can clearly be seen in the recent case of McKeown v DPP 1995 Crim LR 69, where the Divisional Court held that if it cannot be proved that the computer was operating properly, the computer evidence will be inadmissible. This flies in the face of Lloyd LJ’s dictum in the Osman case, since the conclusion was reached despite the fact that evidence showed that the malfunction did not affect the accuracy of the information. The case concerned an appeal by Miss McKeown against her conviction for driving after having consumed so much alcohol that she was over the legal limit, contrary to s5(i)(a) of the Road Traffic Act 1988 and Sch 2 of the Road Traffic Offenders Act 1988.
The appellant underwent a breath test using the Lion Intoximeter 3000 breath testing device. This machine has a visual display and a memory which stores a number of results. Four printouts were produced by the machine, and these were certified by the officer in charge in accordance with s 69 PACE. In his statement, the officer recorded the time shown on his watch as the machine was thirteen minutes out. The submission of the appellant was that the visual displays and printouts were inadmissible on the basis that since the timing device was thirteen minutes slow, it could not be shown according to s 69(1)(b) “that at all material times the computer was operating properly, or if not, that any respect in which it was not operating properly or was out of operation was not such as to affect the production of the document or the accuracy of its contents.” On behalf of the respondent, it was argued that the words “to a material degree” should be read into the statutory provision and that the incorrect timing did not, in itself, render the evidence from the machine de facto inadmissible.
Dr Williams, a director of the laboratories that supplied the breath testing machine, had been called as an expert witness on behalf of the prosecution. It was held that although he was not an electronics expert, his qualifications and experience entitled him to give evidence regarding the machine’s operation. The court accepted his evidence that the breathalyzer’s working and accuracy were not affected in any way by the clock. However, despite these findings, the court took the view that there was substance in the appellant’s submission that, based on the wording of s 69(1)(b), the inaccurate timing mechanism on the machine rendered the printouts produced by it inadmissible. The appeal was allowed, and the conviction was quashed entirely on the basis that, despite the evidence, the prosecution could not prove that the machine was working correctly. The outcome, although in line with the statutory requirements of section 69(1)(b), is quite absurd since there was no question as to the reliability of the evidence.
The McKeown case also gives rise to concern in that the defense raised the smokescreen of concentrating on the fallibility of the computer evidence rather than its reliability. This point was raised by Dr. Castell when he delivered The VERDICT Report to the Treasury in 1987 (9). He was perturbed that the current law could be effectively exploited by defense counsel to undermine a prosecution. The Law Commission, in its Consultation Paper (Law Com CP No 138), claims that judges commented on the lengthy cross-examination of the prosecution’s computer experts, supporting this contention. It will be recalled that the standard of proof in a criminal case for evidence tendered by the prosecution is “beyond all reasonable doubt.” The intricacy and complexity of many modern computer systems may make it relatively easy to establish a reasonable doubt in the juror’s mind as to the proper functioning of the computer. Using the example of the McKeown case, it appears that in the absence of a presumption that the computer is working, it may be quite easy to raise such a smokescreen. It would seem perfectly feasible that where there are doubts as to the reliability of computer-generated evidence, these doubts should not go to the issue of admissibility but rather to the weight of the evidence.
As we have seen in Shephard, s 69 only applies where computer-generated documents are tendered in evidence, and there is an affirmative duty on those introducing computer evidence to show that it is safe to rely on it at all times. Thus, when applying a literal interpretation of the statutory provision, illogicality and confusion reign as demonstrated by the McKeown case. Furthermore, it has been held that s 69 does not apply where a witness uses computer-generated evidence to refresh their memory or where it is used by an expert to reach a conclusion.
In Sophocleous v Ringer 1988 RTR 52, another driving with excess alcohol case, evidence was given against the accused by an analyst who had used a computer to produce a graph illustrating the levels of alcohol in the bloodstream. The graph was not put in evidence, but the analyst was allowed to look at it to refresh her memory. As the graph had not been put in evidence, the court held that s 69 did not apply.
The same outcome is illustrated in a recent Court of Appeal case, R v Golizadeh (1995 Crim LR 232). In this case, a brown substance was found in the possession of the appellant, which turned out to be a Class A drug (opium). The substance was analyzed through a machine that produces a printout in the form of a pattern; this pattern is then interpreted by an expert to determine the chemical constituents of the substance. In arriving at his conclusion that the substance was indeed opium, the expert witness relied on his own interpretation of the printout and the opinion of another expert called to give evidence.
One ground of appeal was that under s 69 PACE, the evidence should have been excluded on the basis that it was based on the computer printouts and was therefore inadmissible. The Court of Appeal rejected this argument and held that s 69 did not apply. Morland J reiterated Lord Griffith’s speech in the Shephard case whereby he stated that the object of s 69 “requires anyone who wishes to introduce computer evidence to produce evidence that will establish that it is safe to rely on the documents produced by the computer”. Thus, it is clear that s 69 will only apply where computer printouts are actually put in evidence. Since in the present case, the printouts had merely been used by the experts in reaching their findings as to the chemical constituents of the substance, s 69 had no application on the facts of the case. In the words of the Law Commission in its recent Consultation Paper, “if it is safe to admit evidence which relies on and incorporates the output from the computer, it is hard to see why that output should not itself be admissible” (Law Com CP No 138, para 14.13).
The irony of the situation is that it appears perfectly acceptable for evidence to be adduced, which is based on computer-generated printouts, but at the same time, if the computer evidence itself were to be presented to the court, then the hurdle of complying with s 69 would have to be surmounted.
Are the special provisions necessary? As we have seen, the statutory provisions impose special conditions on the admissibility of computer output. Are these justified? What is it that is special about computer-generated documents and that distinguishes them from their paper equivalents? It is obvious from examination of the admissibility requirements that computer evidence is regarded as suspicious in several respects. The main problem is concerned with the authentication and accuracy of computer records. It is almost as if the technology is believed to be inherently inaccurate. Section 69 PACE requires some minimum proof of accuracy before the document is admissible. The court must be satisfied of the reliability of the statement as a true record of what the witness observed and also of its authenticity as an accurate record of what was intended to be recorded. As a result, it is necessary to show that at all material times the computer had been functioning properly, or at least that any malfunction had not affected the accuracy of the information.
It was envisaged by the Criminal Law Revision Committee (CLRC 1972, para 259) that there would be many cases where the document might have become corrupted by software errors or hardware malfunctions. It is the contention of this article that this suspicion was probably unfounded. There has been no tangible evidence to date illustrating why computer records are likely to be less accurate than those contained on paper. Paper-based records are also susceptible to alteration and deterioration. Yet, where it is alleged that such alteration has taken place, the paper document remains admissible. The challenge goes to the question of its weight as evidence, to be decided on the basis of the evidence called to prove falsification or authentication.
Regarding documentary evidence, para 3 of Schedule 2 to the Criminal Justice Act 1988 provides that: “In estimating the weight, if any, to be attached to… a statement given in evidence under section 23 or section 24, regard shall be had to all the circumstances from which any inference can reasonably be drawn as to its accuracy or otherwise.”
Although no particular circumstances are specified, it seems safe to assume that regard may be had, for example, to the following matters: whether the person who made the statement in a document did so contemporaneously with the occurrence or existence of the facts dealt with in the statement; whether any person who supplied the information did so contemporaneously with the occurrence or existence of the facts dealt with in that information; and whether or not such persons or the ‘creator’ of the document containing the information had any incentive to conceal or misrepresent the facts.
In stark contrast to this, unless it can be shown that there is no chance of unauthorized use of a computer system, or of system failure, the same document stored on a computer is inadmissible under the additional requirements of section 69 PACE (e.g., McKeown v DPP). Doubts concerning the accuracy of information recorded on computers apply equally to paper-based systems, as do those concerning authentication. As with paper records, the necessary degree of authentication can be proved through oral and circumstantial evidence, if available, or via technological features of the system or record. Although a paper document can be authenticated by its author appending a signature, various technical ways of authenticating computer-based documents are available.
