In many areas of scientific expertise, computers are used to generate what will often be used as evidence in court.  Examples include video image clarifications, 2D and 3D modelling, collision simulations, crime scene diagrams, etc.  A common tactic used in court is for opposing counsel to grill the expert on his/her knowledge of the computer program that was used to generate the visual evidence.  Rarely will an expert be able to answer all of these questions.  But is that really a failure on the part of the expert?  How much does an expert have to know about the computer programs that are used to generate evidence?

The Supreme Court of Connecticut, in State of Connecticut v. Swinton, 268 Conn. 781; 2004 Conn. LEXIS 190 (Conn. 2004), issued an important ruling on the use of computer generated evidence.  In Swinton, part of the evidence against the defendant was the forensic comparison between bite marks found on the murder victim’s breast and dental impressions taken from the defendant.  The State tendered photographs of a bite mark that were enhanced using LUCIS computer software with Adobe Photoshop images of the defendant’s dental impressions superimposed on the LUCIS enhanced photographs.  A forensic odontologist concluded that the defendant inflicted the bite marks on the victim’s breast.  The defendant argued that the State did not present adequate foundation testimony on the adequacy of the two computer programs used for the forensic matching process.  He argued that the State experts had only an elementary familiarity with the computer programs and that as a result his constitutional right to confrontation was violated.

The photographic enhancements were introduced through a State forensic expert who, while not an expert in LUCIS software, provided a reasonably comprehensive explanation and demonstration of how the software works and what affect it had on the photographs in question.  The forensic odontologist, who merely observed the Photoshop overlays being created and had no skill or experience with the program, was able to provide very limited evidence about Photoshop and its workings.

In analyzing the issue of what is necessary to lay a proper foundation for the admission of computer generated evidence, the Court said at pages 812-814:

We agree that “reliability must be the watchword” in determining the admissibility of computer generated evidence; Nooner v. State, supra, 322 Ark. 104; and we conclude that these six factors adequately refine our requirement enunciated in American Oil Co. that, in order to lay a proper foundation for computer generated evidence, there must be “testimony by a person with some degree of computer expertise, who has sufficient knowledge to be examined and cross-examined about the functioning of the computer.” American Oil Co. v. Valenti, supra, 179 Conn. 359. In addition to the reliability of the evidence itself, what must be established is the reliability of the procedures involved, as defense counsel must have the opportunity to cross-examine the witness as to the methods used. We note that “reliability problems may arise through or in: (1) the underlying information itself; (2) entering the information into the computer; (3) the computer hardware; (4) the computer software (the programs or instructions that tell the computer what to do); (5) the execution of the instructions, which transforms the information in some way – for example, by calculating numbers, sorting names, or storing information and retrieving it later; (6) the output (the information as produced by the computer in a useful form, such as a printout of tax return information, a transcript of a recorded conversation, or an animated graphics simulation); (7) the security system that is used to control access to the computer; and (8) user errors, which may arise at any stage.” R. Garcia, “‘Garbage In, Gospel Out’: Criminal Discovery, Computer Reliability, and the Constitution,” 38 UCLA L. Rev. 1043, 1073 (1991); see also K. Butera, “Seeing is Believing: A Practitioner’s Guide to the Admissibility of Demonstrative Computer Evidence,” 46 Clev. St. L. Rev. 511, 525 (1998) (proper authentication requires that reliability of computer process and accuracy of results be subject to scrutiny).

We believe that these factors effectively address a witness’ familiarity with the type of evidence and with the method used to create it, and appropriately require that the witness be acquainted with the technology involved in the computer program that was used to generate the evidence. These factors also ensure that the hardware and software used to generate the evidence were adequate for that purpose and that the technology was reliable. As in our decision in Porter, we stress that these factors represent an approach to the admissibility of computer generated evidence, and not a mechanical, clearly defined test with a finite list of factors to consider. See State v. Porter, supra, 241 Conn. 79. “Trial courts must have considerable latitude in determining the admissibility of evidence in this area as in others.” American Oil Co. v. Valenti, supra, 179 Conn. 360. Although a trial court should weigh and balance these factors and decide whether they ultimately support the admissibility of the evidence, we offer these factors to serve as guideposts, and do not suggest that these factors necessarily are to be held in equipoise…

And at pages 829-830:

A witness must be able to testify, adequately and truthfully, as to exactly what the jury is looking at, and the defendant has a right to cross-examine the witness concerning the evidence.  Without a witness who satisfactorily can explain or analyze the data and the program, the effectiveness of cross-examination can be seriously undermined, particularly in light of the extent to which the evidence in the present case had been “created.”

The Court concluded that the bitemark photographic evidence that was generated by LUCIS software was admissible because the forensic expert that presented the evidence had a sufficient level of knowledge regarding LUCIS to provide foundation evidence.  However, the Photoshop evidence was ruled inadmissible because the forensic odontologist did not possess the requisite foundation knowledge.

Thus, an expert witness who presents computer generated evidence must have sufficient knowledge of the processes and equipment used in order to lay the appropriate foundation for the evidence.  It is not necessary for the witness to be an expert in the computer programs involved but certainly being able to push the buttons is not sufficient for admissibility.  An acceptable middle ground is to have a competent working knowledge of what the computer programs do and how to interpret the results.

The Swinton case was revisited by the Connecticut Supreme Court in the context of video evidence in State of Connecticut v. Melendez, 291 Conn. 693, 970 A.2d 64 (Conn. 2009).  At issue was whether digital video of drug buys was admissible at trial.  The State sought to introduce DVDs of drug buys.  Some of the video had been enhanced, slowed to 10% real time or both.  The State did not call sufficient evidence to comply with the requirements of Swinton and even though the trial court admitted this video evidence, the Supreme Court ruled that it should not have done so and excluded the video evidence.  Other video consisted of DVDs that had a copy of the original video that was not subjected to any modifications.  At pages 709-710, the Court stated:

We reach a different conclusion, however, with respect to the portions of the DVD containing the footage that Brunetti did not modify, that is, the two video clips that are exact copies of the footage originally captured on the eight millimeter videotape while the transactions were occurring. In Swinton we acknowledged the difficulty in establishing a precise definition of what constitutes “computer generated” evidence. We did, however, draw a distinction between technologies that may be characterized as merely presenting evidence and those that are more accurately described as creating evidence. With that fundamental distinction in mind, we conclude that the portions of the DVD containing the exact duplicates of the original, unenhanced footage played in real time, simply do not constitute computer generated evidence for purposes of Swinton. Thus, to the extent that Brunetti merely transferred a copy of the contents of the original eight millimeter videotape to the DVD, that process, which Rubinstein witnessed, does not implicate the foundational standard that we adopted in Swinton. Although it is true, of course, that generating such a copy required the use of technology, that technology, which is widely used and readily available, involves nothing more than the reproduction of video footage from one medium to another. Indeed, the defendant has provided no reason why the admissibility of copies that are produced by that process-copies that have not been enhanced, altered or changed in any way-should be subject to the more rigorous requirements of Swinton. We conclude, therefore, that compliance with Swinton was not a prerequisite for admission of the unmodified video clips contained on the DVD.

Therefore, Swinton is clearly applicable to digital video evidence but the detailed requirements set out by the Court do not apply to the process of copying video without modification.  To be safe though, it is recommended that sufficient evidence be called regarding this transfer process.

The Swinton ruling is not binding outside of Connecticut but the rules that it has created are sound and should be followed by all experts who use computer generated evidence as part of their courtroom testimony.