Computers on the Witness Stand: Expert testimony that relies on data generated by computers in the age of Daubert.
Copyright 1997 / (c) 1997 by David Yale

David Yale

University of Connecticut Law School

Fall 1996

Forensics and the Law

Feedback to

Twenty years ago, Judge Van Graafeiland recognized a problem with computer evidence in his dissent in Perma Research & Development v. Singer, and said "[a]lthough the computer has tremendous potential for improving our system of justice by generating more meaningful evidence than was previously available, it presents a real danger of being the vehicle of introducing erroneous, misleading, or unreliable evidence." (1) Since that time the use of computers in various ways in litigation has grown tremendously, and the trend shows no signs of slowing down.

The Supreme Court has changed the playing field in the use of scientific evidence in the Daubert decision. (2) Since many uses of computers in trial constitute scientific evidence, Daubert has a direct effect on their use in federal courts, and state courts that follow the federal lead in their rules of evidence. In this paper I will differentiate between different types of computer evidence and point out some common misperceptions and discuss how Daubert affects these types of computer evidence, along with some suggestions for clarifying the current inconsistencies in the courts. Specifically I will look at how the Daubert decision could effect discovery issues when an expert witness relies on a computer to assist him in forming his opinion. I will limit myself to the Federal Rules of Evidence, but much of the information will also be applicable to courts still using the Frye test. (3)

Daubert Shakes up the Playing Field

Trial judges, being faced with similar scientific evidence, are reaching different conclusions on admissibility in different trials. (4) In remand the Ninth Circuit Court of Appeals in Daubert titled a section of the opinion "Brave New World" and said "Federal judges ruling on the admissibility of expert scientific testimony face a far more complex and daunting task in a post-Daubert world than before." (5) Some courts are applying Daubert only in situations presenting "novel" scientific evidence (which is not mentioned in the opinion) (6), some are applying it to all scientific, technical and expert witness evidence, (7) while others apply it to all scientific evidence but only to scientific evidence. (8) It very well may be some time before the new rules are finally settled out, but for now I'll limit my remarks to scientific evidence and assume that Daubert covers all scientific evidence, not just novel scientific evidence.

As to exactly what Daubert requires, one court said, "The role of the trial judge, properly following the Daubert mandate, is to ensure that the conclusions reached by the scientific experts have some minimal level of reliability and probative value. This determination is accomplished by establishing that the predicate principles and methodology relied upon by the experts are valid and that they can be applied to the facts at issue. The sufficiency of the evidence and the weight of the evidence, however, are beyond the scope of the Daubertanalysis." (9)

What is Scientific Evidence?

Daubert says of scientific knowledge, "[t]he adjective 'scientific' implies a grounding in the methods and procedures of science. Similarly, the word 'knowledge' connotates more than subjective belief or unsupported speculation." (10) The opinion limits itself to scientific knowledge and does not encompass technical or other specialized knowledge. (11) As to exactly what scientific knowledge is the court provides little helpful information. There are no stand alone factors, but the opinion does provide some hints to help judge "scientific knowledge that will assist the trier of fact." (12) Testability of the knowledge is said to be a key question. (13) Other considerations include whether the theory or technique has been published and subjected to peer review, the error rate, either known or predicated, and the general acceptance (a la Frye). (14) The approach is reemphasized as a flexible one. (15) It will take time before exactly what evidence is and what evidence is not covered by Daubert becomes clear.

Why Visual Evidence is so Important

Studies have shown that the average person retains 87% of information presented visually, but only 10% of information presented orally. (16) Another study showed 65% retention for visual information and 15% for oral information. (17) A third study showed juror retention increased 100% when using a visual presentation over an oral presentation. (18) The use of computer animation should result in jurors paying closer attention to the material being presented (19) For trial tactics, this almost mandates the use of animated visual evidence of some type when presenting facts to the jury whenever possible.

A lawyer thinking he can rely on tried and true charts and diagrams also has to face a new reality. A study comparing static visual presentations to computer animations of the same information found the computer animations to be much more persuasive. (20) This implicates not only retention, but the weight given to the evidence by the juror.

An animation can also serve to make a complex subject more understandable. One animation reduced the traditional presentation of diagrams of a chemical plant, maps of city sewers, eyewitness accounts, and expert testimony on the nature of chemical explosions to a single visual story incorporating all the elements of an explosion from beginning to end. (21) Imagine the difference in an automobile accident case between an expert witness talking about weight shifts, tire traction marks, yaw, gouge marks, and coefficient of friction, possibly with some illuminating hand gestures, compared to an animation showing the tires sliding and the front end of the car digging into the roadway, and matching these actions with physical evidence found at the scene. In effect the animation brings together all the facts traditionally summarized by an expert witness into a more effective format.

Proponents value visual evidence because it demonstrates precisely the theory a lawyer is trying to present, avoiding the possibility of a juror misconstruing the witness's verbalizations into a concept far removed from the one intended. (22) Consider a witness testifying she heard a loud bang. One person hearing the word "bang" will think of a gunshot, another of a door slamming. Without further clarification different jurors could develop different mental images of what the witness is saying. A computer animation gives the attorney the advantage of each juror seeing the picture he is trying to portray. There are no problems with different interpretations of his story.

Using moving visual evidence allows a witness a host of new tools to help explain his opinion, including slow and fast motion, stop-action, and view from a variety of perspectives. Counsel used stop-action with dramatic results in the original Rodney King beating case. By taking the video apart frame by frame they were able to present a witness's theory that what was actually portrayed was an escalation/de-escalation response to a violent subject, and in conformance with the Los Angeles Police Department Training policies, causing the jury to return an acquittal, despite what most viewers of the video thought. (23) A computer animation could be produced that goes step by step through an automobile accident. Using a split screen it could show an external view of the vehicle as well as an internal view of the occupants. As the expert goes through his explanation of the accident, he could point out how every piece of physical evidence was left on the roadway, how each item of damage occurred on the vehicle, and how every injury suffered by an occupant was inflicted. Very little imagination is needed to see how powerful a tool that could be, fitting all the physical evidence into a neat story with no loose ends.

Costs on producing moving visual evidence have also dropped dramatically. While some articles talk about spending $150,000 (24), today prices have dropped significantly. (25) While in the past this type of evidence may have been restricted to well-financed cases, now it can be produced on a much more modest trial budget.

Given these facts it is no wonder that litigators are pressing to get their visual evidence into court and that so much has been written dealing with visual evidence. In the rest of this paper I will look at some of the dangers of computer generated evidence, discuss the different types of computer generated evidence, and define it based on how the computer is relied upon to produce it.

The Dangers of Computer Evidence

The saying that power corrupts, and absolute power corrupts absolutely, applies to computer generated evidence as well as anything else. A US District court, in excluding a computer animation that demonstrated an expert's theory of how an accident occurred said "[b]ecause the experts"s conclusion would be graphically depicted in a moving and animated form, the viewing of the computer simulation might more readily lead the jury to accept the data and premises underlying the defendant's expert's opinion, and, therefore, to give more weight to such opinion than it might if the jury were forced to evaluate the expert's conclusions in the light of the testimony of all the witnesses, as generally occurs in such cases." (26) Courts and commentators have recognized some of the potential problems with computer evidence and are attempting to bring attention to them.

Federal Rule of Evidence 403 says "[a]lthough relevant, evidence may be excluded if its probative value is substantially outweighed by the danger of unfair prejudice, . . . " (27) The supreme court quoted Judge Jack Weinstien saying "[e]xpert evidence can be both powerful and quite misleading because of the difficulty in evaluating it. Because of this risk, the judge in weighing possible prejudice against probative force under rule 403 of the present rules exercises more control over experts than over lay witnesses." (28) The Reference Manual for Scientific Evidence also recognizes that computer generated evidence may be so vivid and compelling that jurors may disregard its true value. (29) Added to the problem is the possibility of jurors misunderstanding judges' instructions when faced with new types of evidence. (30)

Because of the powerful effect a computer animation can have on a juror, the court is forced to ensure that the jury understands the expert's testimony and consider the cross examination, and don't rely simply on the animation. (31) The very fact that computers perform daunting amounts of number crunching, allowing experts to do things that otherwise would not be possible, tend to overawe judges and juries. (32) Coupled with a "gee whiz" quality often associated with computers, the result is that opposing attorneys, juries, and judges often do not regard them with the skepticism that they might deserve. (33) Additionally, many attorneys are not very well informed about computers and lack sufficient understanding to critic an opponent's use of computer evidence. (34)

There are some aspects of computer simulations that many people are not aware of. A simulation is a simplified version of the truth and generally does not include all the factors that would have an effect in the real world. (35) Simulations may have known mistakes in them because the programmer feels attempting to correct the problem will create bigger problems. (36) As such, many simulations are not trying to portray actual truth, merely an estimation of it, a fact that many people are not aware of.

The very nature of visual evidence as opposed to oral evidence can lead the jury to give more weight to it. The form of the exhibit can have a greater weight than the message, with color, size, depth, movement and pattern being a bigger component in the jury's perception of the evidence than the message it portrays. (37) Above I described a hypothetical visual exhibit where a reconstructionist made a stop-action animation of a vehicle accident which included the movement of the passengers inside the vehicle. Typically a medical expert can testify to the nature of injuries using photographs at best. It is much more visceral to have the expert use the stop-action graphic and give a blow by blow description of the victim being injured or killed. (38)

There is also the issue of bias. Because an animation is produced by a party to a lawsuit there can be an included bias, even if unintentional, to shade the graphic in a light more favorable to the proponent of the evidence. (39) If an outside company is hired to produce an animation for litigation, they are naturally going to want to produce a product that will be pleasing to their customer, and their customer is an attorney at trial, not the truth.

In complex cases there is a danger of overwhelming jurors. There are already studies that show that complex cases are already throwing so much at juries that they are not able to deal with it and have trouble remembering facts and issues. (40) If one side presents a clear computer graphic of their theory of the case, and, as mentioned above, animated evidence makes a greater impression on the jury, there is a real danger that more persuasive facts will be brushed aside because of the less impressive method of presentation.

Indeed, video evidence can be a two edged sword. I already mentioned the amateur video shot by George Holiday that was used in California v. Powell, commonly referred to as the "Rodney King Case." Although this evidence was produced by a video tape, and not a computer, it helps illustrate some important themes. In his opening remarks, prosecutor Terry White said "" . . . what Mr. Powell did next, as you're going to see on the video, was unreasonable." (41) In closing he said "This videotape is the central piece of evidence in this case." (42) He further asserted it could not be rebutted. (43) Ira Reiner, L.A. District Attorney said that it was unnecessary for Rodney King to testify because the video tape showed the story more clearly than King could have done. (44) The defense, through witnesses Charles Duke and Robert Michael, dissect the video and portray it as reasonable reactions on the part of the defendants to King's aggressive behavior. (45)Captain Michael Bostic, a training captain with the Los Angeles Police Department at the time, used stop-action of the video to establish exactly where he thought the police actions became unnecessary force. (46) This was a case where an 81-second video expanded into days of testimony and evidence and each side attempted to twist it to their own purpose. More important, because it existed, each side felt it was necessary to use it and make it central to their case.

The prosecution and defense argued that the tape showed different truths. This case illustrated a situation where an 81-second piece of visual evidence dominated everything else in the trail. In a similar way, courts need to safeguard that computer generated visual evidence does not become more important than it actually is, and does not rise to the importance of the video in the Rodney King case where the video showed an actual event. It must always be remembered that with computer evidence, it almost always shows on party's version of the facts, it is not impartial evidence.

Different Flavors of Computer Generated Evidence

As a preliminary matter the legal profession must be very clear as to the classification of different types of computer output. Literature uses various terms including "Computer Generated Visual Evidence," (47) "Computer Animations," (48) and "Computer Simulations" (49) among others. Different authors use the same terms to define different concepts (50) and some fail to recognize the very real differences between different types of computer evidence. (51) Courts also fall into the same trap. (52) One opinion uses three different words to describe the same computer evidence in one paragraph. (53) There is no consensus on definitions so I will advance the following in order to clarify the differences.

There are three broad types of computer evidence. In one type the computer is used simply as a drawing tool. A moving illustration is prepared on a computer for display to a jury. The illustrator determines where the objects are placed in the scene, how they look and how they move. Commonly a three-dimensional graphics program is used to assist in creating texture, shading, and movement, but it does so solely at the direction of the computer operator. In it's simplest form this would be a cartoon drawn on a computer instead of on paper. This type of computer generated evidence is a Computer Animation. (54) The Disney movie Toy Story is an example of computer animation. Although it was amazing life-like, no one is arguing it was real.

The second general category consists of the output of a computer program that is designed to take a set of input data and model what would happen in an actual occurrence of the event. Although all the action is taking place inside the computer, the results are offered as a prediction of what would happen if the event actually took place. This might be as simple as a financial spreadsheet program being offered to show what a tort victims future earnings will be, where the underlying data and equations used to manipulate them are fairly easily learnable, (55) to a program designed to determine future energy use in a building, where the equations are sophisticated and included inside the computer program so as not to be readily available to the user of the program. (56) This type of computer evidence is a Computer Simulation. (57)

The third general category is when a computer is used to determine how an event occurred given an end result. It is used to go back in time to take an event apart and see what individual actions and reactions caused the end result. Most commonly this type of program is used to investigate a motor vehicle or aviation accident. This is a Computer Reconstruction. (58)

Other types of computer evidence can blur the distinctions because they may be a combination of these three types, or a program designed to do one may be used for a different function. Undoubtedly skillful and imaginative litigators will discover new ways to use computers in the courtroom, but this offers a start.

It is very easy to confuse the different types of computer generated evidence because what is seen on the computer screen might look very similar. Consider a hypothetical motor vehicle accident case that includes testimony from three different expert witnesses.

The first witnesses gathers data and sits down with a paper, pencil and calculator and reconstructs what he believes happened in the accident based on physical damage to the vehicles, skid marks and witness statements. He has an animation produced where he tells a computer artist what he wants the animation to show, based on his reconstruction.

The second witness starts with what the plaintiff says were actions prior to the accident. He investigates the scene od the accident and enters information about the accident scene and the plaintiff's vehicle in to a simulation to see what the vehicle should do and either supports, or attacks the plaintiff's testimony. His simulation generates a display of what the plaintiff's car would do based on the information entered.

The third expert investigates the accident similarly to the first, using the same types of evidence, but he uses his computer to do the math instead of paper and calculator. The computer creates a display of a third version of the accident based on the formulas and instructions in the reconstruction program.

It is easy to see why so many people get confused over the different types of computer evidence. In each case the visual display created by the computer and shown to a judge and jury attempt to show what happened in the accident and they may look very similar, but they arrived at the end result through very different paths, and because of fundamental differences they must be treated differently by the courts.

The Hearsay Issue

Some commentators consider computer generated evidence hearsay, feeling that the jurors are relying on the truth and accuracy of the computer and software used to generate the evidence to prove a factual issue, as such the program is being admitted to prove the truth of the matter asserted. (59) Most authors rely on the Federal Rules of Evidence for this theory, and do not cite any case law. U.S. v. Trenkler (60) deals with a computerized database and the hearsay issue caused the computer information to be excluded. This ruling in a criminal appeal centered on the fact that the prosecution presented a summary of a database of bombs without presenting witnesses who had first hand knowledge of the truth of the information provided. (61) This was not a Daubert analysis, and did not deal with scientific evidence.

If the computer evidence is part of an expert witnesses testimony, Federal Rule 703 allows the testimony even if the underlying basis is not admissible it is of the type normally relied upon by experts in the field. (62) There seems to be a mini-Frye test here, applicable to the admission of otherwise hearsay evidence. Although the literature shows some concern over the issue, it does not appear to have been a factor in many situations

In Commercial Union Insurance Co. v. Boston Edison Co. (63) the Supreme Judicial Court of Massachusetts answered a hearsay objection raised against a computer simulation by pointing out "We permit experts to base their testimony on calculations performed by hand. There is no reason to prevent them from performing the same calculations, with far greater rapidity and accuracy, on a computer." (64)

Computer Animations as Demonstrative Evidence

Computer Animations are frequently used as demonstrative evidence. (65) Our first hypothetical expert mentioned above would be such a situation. First, the expert would present his investigation. After the foundation was laid as to how he came up with his theory of the accident, and what his theory is, the animation would be presented to make his theory more understandable to the jury. In this situation the computer has made no decisions and produced no facts that the expert relied on as to what occurred. The animation came into being after the expert's opinion was already formed. At a simplified level ask if the picture the computer generated is merely what I would see if I closed my eyes and listened carefully to the witness testify as to what happened? Could this visual be presented with models, or a cartoon without the interaction of the computer? If so then it is probably demonstrative evidence, merely a more effective type. The basic evidence presented to the jury as factual would still exist without the computer.

A short time ago the District Court of Appeals of Florida looked at a computer animation offered as demonstrative evidence. Reviewing a lower court's admission of the evidence, it said "the computer animation in the instant case was admitted solely as an illustration of Detective Babcock's opinion of how the accident occurred.... The trial court ... determined that the demonstrative exhibit was not subject to Frye analysis. We agree..." (66) The court went on to review the standards for admission of demonstrative evidence and said that "the proponent must establish that the facts or data on which the expert relied in forming the opinion expressed by the computer animation are of a type reasonably relied upon by experts in the subject area." (67) Note particularly that the court focused on the opinion formed by the expert, not on the animation. The opinion came first, and the animation relies on the opinion which is very different from a situation where the expert relies on the computer to form an opinion. They finished by requiring a computer animation pass the same standards as any other photograph, motion picture, video tape or other type of pictorial representation. (68) Although Florida still uses the Frye standard, the analysis would be the same under the Federal Rules of Evidence, because Daubert did not deal with demonstrative evidence.

If the animation fits the criteria described above then it is not scientific evidence, merely a very effective form of visual evidence, and should not be scrutinized under a Daubert analysis. Instead a computer animation should be judged under the same rules as other visual demonstrative evidence. It's accuracy is totally dependant on what it looks like. The United States District Court for New Jersey spoke about some computer generated visual evidence in Exxon v. Halcon Shipping Co, Ltd. (69) The court considers what appears to be a computer animation of Exxon's version of how some damage occurred to an underwater pipeline, showing the passage of a tugboat over the pipeline which the court considered inaccurate. (70) The computer was not programmed with a set of input data and asked to determine where the tugboat went, nor was it used to try and backtrack the tugboat path from an end point. The animation was prepared like a cartoon, the computer being merely an animated sketch pad for an artist.

Imagine that instead of a computer animation, Exxon had produced a video using scale models set in a scale representation of the area of the damaged pipeline. By moving the models very slowly, and exposing one frame of film at a time, they could have produced a stop action movie of what their version of the disputed facts were. This movie would be based on Exxon's expert's opinions of how the tugboat moved, the path it took, and how it hit the pipeline. The end product would not be substantially different that the computer animation they prepared, and it could have been offered as the same type of evidence.

The opinion states, however, "The Court need not spend much time in comment on this exhibit because it is of no evidentiary value as a demonstrative exhibit. This is the type of exhibit that Daubert v. Merrell Dow Pharmaceuticals, Inc. had in mind when it designated the judge as the gatekeeper to eliminate junk science from the judicial arena." (71) The court may have made the right decision, barring demonstrative evidence that was not supported by the facts, but the quote shows faulty reasoning. There was no claim that the animation was scientific evidence, and not all output from a computer is scientific merely because it comes from a computer. Daubertlimits itself to scientific evidence from expert witnesses. (72) There is simply nothing in the Daubertdecision to allow it's application to what is merely a visual representation of an expert's opinion.

A computer animation does not lend itself to a Daubert analysis. It would be possible to bring in the computer animator and ask him if the program used has been reviewed in the appropriate professional publications, and if the program is accepted in the computer animation field. You could also bring in an artist who prepared an ink and paper diagram and ask if the pen he used has been written up in art journals and if the paper is commonly accepted among other artists. Both set of questions are equally worthless. The product, as demonstrative evidence, must be viewed as an end result, not a process.

If Exxon had offered the proposed video using models as proposed above and offered it as demonstrative evidence, no one would have looked at Daubert for guidance. The animation offered was the same type of visual evidence as the model video would have been, and any analysis of the two types of evidence should encompass the same principles. Any Daubertanalysis is misplaced.

The Court of Appeals of Arizona makes a similar mistake in reasoning in Bledsoe v. Salt River Valley Water Users' Association. (73) Although their are some proponents of using computer animations in closing arguments (74) the court rejected that approach in this case revolving about a bicycle rider who ran into a cable across an access road while riding in the dark with no lights. (75)The proponent of the animation was attempting to introduce it for the first time in his closing argument. He represented it as a synthesis of what various witnesses had testified to in the trail. It was not represented as a scientific investigation or analysis of the evidence, again, it was merely a visualization of an opinion of what the facts were. Using the same technique as above, the attorney could have prepared a stop action movie using scale models and surroundings. This movie would have been similar to the animation in question and served the same purpose. The animation was not represented as a computerized analysis of the true course of events, or as any type of investigative tool.

In analyzing the animation the court starts out saying, "[a]t a minimum, the proponent must show that the computer simulation fairly and accurately depicts what it represents, whether through the computer expert who prepared it or some other witness who is qualified to so testify, and the opposing party must be afforded an opportunity for cross-examination." (76) This appears to be fairly standard for demonstrative evidence. It focuses on what the jury sees and questions the fairness of the jury being allowed to see it. The opinion goes on to state that the animation "depicts a computer expert's opinion of, among other things, how the accident happened,..." (77)This is the point where the decision starts to go off track. It may be that the program used to produce the animation was in fact a simulation, but the published decision does not bring that out if it was. The case is further confused by the offering party's apparent reliance on precedent cases which used simulations created by and offered by expert witnesses during testimony as authority to use an animation during closing argument despite arguing that the animation was just like a chart, diagram or any other pedagogical device. (78)

The court further says that the offering party may also be required to show that the computer was functioning correctly, that the underlying equations were correct, and that the program was generally accepted by the appropriate community of scientists. (79) These factors focus on the means and not the ends. This test is not appropriate for this circumstance. It does not matter if the animation was made by a high school student on an IBM pentium using Windows and a barebones 3d animation program or by Lucas Films on a $50,000 Sun Graphics workstation running Unix and a proprietary cinema quality animation program. Each of these tools is capable of producing a similar animation. As such how it was produced is not important, the focus should be on the end product. Does it accurately portray evidence that was admitted in the trial? Is there a basis for what it depicts? Is it merely a visualization of the closing argument?

Perhaps the court could conclude that it is such a strong presentation that it is too prejudicial to be shown for the first time in a closing argument, but the analysis should not rest on a scientific basis, because science did not determine the end result. The animation looked like it did because it visualized for the jury the picture that was in the proponent's mind. It was what he wanted in the jury's mind when they thought of the accident. The fact it was produced on a computer had nothing to do with what it showed, and an detailed examination of means of production is inappropriate. It was the factual determination of the nature of the animation that was in error, leading the court to apply the wrong standards.

Are Computer Simulations Scientific Evidence?

Simulations used in the investigation of a motor vehicle accident use physics to predict how a vehicle will behave under certain circumstances. (80) Vehicle simulations are the subject of intensive studies at university graduate level, and practitioners are constantly attempting to improve the field through validation techniques. In a simulation of a Ford Taurus, twenty nine thousand elements were modeled. (81) In a recent simulation of a pick-up truck impact into a jersey barrier at an angle took 152 hours of CPU time to simulate one half of a second's worth of impact. (82) This model was compared to an actual crash test of a very similar truck and analyzed to see how to better model the actual event. (83) On a less sophisticated level, the University of Michigan has made available a simplified simulation program to model vehicle handling and braking that can be run on a home computer. (84)

Modern vehicle simulation programs are designed to model many different vehicles, but some authors maintain that these modern simulations are sometimes less accurate than the old, hand written models of 40 years ago. (85) As interesting as that is in itself, it shows that the research being done is being presented to peers and held out for review. We have also seen that vehicle computer simulation is the subject of current university studies with ongoing research and a quest for reliability and ongoing tests of accuracy. Coupled with manufacturer's use of vehicle simulations in developing new vehicles, which shows that there is a general acceptance in the related community, computerized vehicle simulations meet the requirements necessary to elevate them to scientific knowledge.

Courtroom Uses of Computerized Motor Vehicle Simulations

A computer simulation is very different from an animation. In a simulation the expert relies on the computer to provide answers as to what would happen given a specific set of circumstances. Simulations have found wide usage because once set up, it allows engineers to change a variable and run "what if" scenarios. They have come into the courts in automobile cases because an automobile manufacturer can program in the characteristics of the car and push it until it fails, instead of driving an actual car around a curve at ever increasing speeds until it finally crashes. An expert witness might be able to take a paper and pencil and come out with the same results as the simulation, but the use of the simulation is more efficient. Additionally some simulations can generate a visual display of what occurs, offering a powerful way to present the results.

In Bartley v. Isuzu Motors Ltd. the plaintiff offered an expert witness who used a computer simulation to determine how he predicted a Isuzu Trooper would react under a variety of situations. (86) In Deffinbaugh v. Ohio Turnpike Commission the court allowed the use of a simulation to predict the movements of a tractor trailer truck. (87) The Supreme Court of Nebraska allowed the use of a simulation of a Fiat X1/9 into evidence to support a claim that the car was unreasonably dangerous in its handling characteristics. (88) With the high number of motor vehicle accidents, and the falling cost of computing power, it is natural to assume the use of computer vehicle simulations will rise in the future.

Are Computer Reconstructions Scientific Evidence

To explore exactly what accident reconstruction is, how it is performed, and how computers are used would take several papers of this size. Basically an accident reconstructionist starts where the accident ends. He surveys the accident scene and examines the vehicles. He speaks to witnesses and drivers. From this point he attempts to use the principles physics, human behavior, and his own experience to determine what happened before, during and after the accident. (89)Although an expert may use science to provide some information, he blends art and science into forming an opinion. (90)

My focus is on the portion of a reconstruction which uses physics to determine vehicle movement and the use of a computer to do these calculations. One principle use by reconstructionists is the conservation of linear motion, expressed as M1V1 +M2V2 = M1V1' + M2V2' or Momentum at impact = Momentum after impact, basically meaning that although the speed and direction of two vehicle may change in an impact, and energy may me transferred from one vehicle to another, the total momentum will not change because linear momentum is preserves. (91) For instance, an equation for determining the momentum of a moving vehicle is (92): M1 * V1 + M2 * V2=V1f + M2 * V2f. although it is certainly solvable by hand, it is a natural for computerization. (93)

An accident reconstruction program uses factors like energy analysis to determine the energy used to deform a vehicle (94), and from that determine the vehicle's speed at the time of impact. From that point the analysis of skid mark length and drag coefficient can determine the speed of the vehicle when the driver started to apply the brakes. Although the expert witness that testifies to this evidence does not need to be a scientist, he relies on data derived from scientific methods when he makes his determination. Although some accident reconstructions are done without the help of a computer, more and more are done with specialized software.

Training for an accident reconstructionist generally is taken in one or two week intensive courses and run from basic on scene investigative techniques to issues involving heavy trucks or motorcycles. Full curriculums are offered by Northwestern University and the Institute of Police Training and Management. The Society of Automobile Engineers (SAE) has a Accident Investigation and Reconstruction Practices standards committee. This group publishes papers and manuals dealing with various topics in accident reconstruction, including validation of computer software models. (95) Additionally the National Highway and Traffic Safety Administration sponsors investigation of motor vehicle accidents and reconstruction. Professional Journals exist for reconstructionists and actual crash tests have been used to validate reconstruction algorithms. (96)

If we use the Daubert analysis of what constitutes scientific evidence, at least to the point it is given in the decision, we need to see if accident reconstruction (1) can be tested as to accuracy; (97)(2) If it has been subject to peer review and publication; (98) (3) The known or potential error rate; (99) and (4) The general acceptance of the technique among the proper community. (100)Specifically above we have seen where actual accidents with known speeds and vehicle movements were compared to the predicted results from accident reconstruction programs. This allows the technique and software to be tested for accuracy. Error rates would be determined from the same situation. The SAE has numerous papers dealing with computer assisted accident reconstruction. These factors alone seem to weigh in favor of accident reconstruction programs fitting inside the nonexclusive test proposed by the court. Of course Daubert is very clear that there are other, as of yet undiscovered tests that may be used to prove a technique is scientific evidence.

Some courts have disagreed however, when faced with engineering or technical witnesses. In Compton v. Subaru of American, Inc., the Tenth Circuit rejected a lower courts reasoning that an engineer's testimony was based on scientific knowledge. (101) "The language in Daubert makes clear the factors outlined by the Court are applicable only when a proffered expert relies on some principle or methodology. In other words, application of Daubert factors in unwarranted in cases where expert testimony in based solely upon experience or training." (102) This expert relied on SAE papers dealing with rollover accidents and roof crush, proposed standards for maximum allowable roof crush, examination of the accident vehicle and eight years of personal experience as a "consulting engineer." (103)

Discovery Issues

Traditional discovery issues, dealing with both computer simulations and reconstructions, usually reflect the opposing party wanting extremely broad discovery issues. Many rely on FRE 705 which allows an expert to give an opinion even when the underlying data used to form it is inadmissible. (104) The comments reflect that the opposing party may choose "to bring out any facts or data... unfavorable to the opinion..... [This] assumes that the cross-examiner has advance knowledge which is essential for effective cross-examination." (105) Some courts are realizing that it is not practical to expect an attorney to extract the underlying principles of a simulation or reconstruction from a witness during cross examination, then proceed to effectively identify and point out flaws while thinking on his feet. (106)

In a 1993 discovery motion, the US District Court of Colorado dealt with a request for disclosure of data from computer simulations modeling motor vehicle behavior. (107) The defense argued that experts often run computer simulations many times in order to get the desired result and requested information of every computer simulation run the expert ran. (108) The court said, "When one party seeks to present a computer study, in order to defend against the conclusions that are said to flow from these efforts, the discovering party not only must be given access to the data that represents the computer's work product, but he also must see the data put into the computer, the programs used to manipulate the data and produce the conclusions, and the theory or logic employed by those who planned and executed the experiment. All the information used in generating the computer simulations is relevant to Defendants' challenge of this evidence." (109)The court went on to order the plaintiff to preserve records of all computer runs in such a way that the Defendant could identify all input and output data. (110)

Not all decisions are so enlightened, though, The Georgia Court of Appeals upheld a ruling that allowed admission of "two computer runs" that a plaintiff was unaware of until they were offered. (111) The court felt that the ability to examine the evidence and speak to the witness overnight was sufficient, even though pre-trial discovery requests had included information on all computer simulations the witness had conducted. (112) Basically one of the parties had to revamp their case overnight to accommodate surprise evidence. As we shall see below, when dealing with computer evidence it simply is not that easy.

The law in the area of discovery related to computer generated evidence is in flux at this point in time as the courts attempt to deal with technology that changes literally everyday. There is no clear consensus and in the coming years will see where today's changes lead us.

The Technology at Issue

Computer software is written in Source Code. (113) This is a Language (114) that allows a computer programmer to develop programs in a way that is easier for humans to understand. The source code is then compiled into a form that the computer can understand called Machine Language. (115) The computer cannot understand or run the source code directly, it is an in between step to allow computer programs to be written easier and quicker. If for instance, a judge ordered the proponent of a simulation to turn over the program used the printout could look like this:

21 f4 65 h1 aa 43 aa 23 76 38 5c 4e 22 d2 11 03 58 f4 31 8f ff cf

and go one for hundreds of pages. This would be a machine code dump of the program and technically would be the computer program. It is almost worthless and although a computer programmer could eventually determine what the computer was doing, it wouldn't be easy.

There are several commercial programs in the market available to a reconstructionist, including ED-SMAC, (116) PC-CRASH, (117) WinSMAC, (118) and m-smac. (119) In addition there are other programs that are in the public domain to make similar calculations. Many of these are based on the original Simulation Model of Automobile Collisions (SMAC) program designed in the early 1970's. (120) Although there have been validation attempts (121) there are still questions about the accuracy of any given program. (122) In addition, some of the computer programs that are used to determine speed based on crush damage may be off as much as 58% in accidents that have a crush depth of 10 inches. (123) These show there is continuing controversy over particular programs and the way they reconstruct the accident event.

This is not to say that accident reconstruction is unreliable, however. Cornell University did a study where vehicles were crashed into each other under controlled and recorded conditions. Reconstructionists then studied the vehicles and were able to establish impact speeds within three miles per hour of the actual recorded speed. (124)


Daubert's Open Door

Daubert explicitly handed the trial judge "the task of ensuring that an expert's testimony ... rests on a reliable foundation..." (125) While the original case took the position that evidence not allowable under Frye might be allowable under Daubert, it is not inconceivable that some litigators will look to Daubert to exclude testimony that would have been allowable under Frye. Previously, as long as a methodology was generally accepted, the proponent could usually get it admitted, without the judge doing much more than holding an opinion poll. Now the court must take a good look at the methodology and pass judgement on it based on it's merits.

This leads to some interesting, and actually quite complex questions when expert witnesses rely on computer programs to assist them in forming their opinions. In this situation the computer is presenting testimony just as much as the witness. In some situations an expert might be the person who wrote the computer code and can testify as to how it arrives at the results that it does, but in most of the reported cases the expert did not write the computer program and is relying on someone else's computer expertise.

Imagine, if you will, an automobile crash case where the plaintiff seeks to offer the testimony of an accident reconstructionist. In investigating the accident he used a commercial reconstruction program that produces a visual representation of the accident as well as determining vehicle speeds based tire traction marks and crush damage. The reconstructionist has relied on the program to make the calculations, and it is a program that has a good reputation in the accident reconstruction field.

The normal vehicle for admitting this information would be FRE 703, allowing the opinion formed by an expert relying on out of court facts and data as long as those facts and data are of the type normally relied upon by experts in the field. This begs the question however, who formed the opinion, the expert or the computer. If the expert is merely a spokesman for the computer, then it is not addressed by rule 703. There is some disapproval of experts who cannot explain the interim steps in forming an opinion. One dissenting judge said "... an expert's testimony does not 'assist' the trier of fact if the expert does not explain the steps he took in reaching his conclusion. We should not require the trier of fact to accept blindly the expert's word to fill the analytical gap between proffered 'scientific knowledge' and the expert's conclusions." (126)

Under Frye, the court could avoid analyzing how the computer arrived at it's conclusions. There was a feeling that if a significant group of scientists in the appropriate group used the program, that spoke to it's reliability. (127) If it had major problems, they would have been discovered through use and unreliable programs either abandoned or at least recognized in their limitations.

Daubert poses a new puzzle, what to do when the defendant objects to the use of the computer program by the expert claiming it is not proper scientific evidence. In the hypothetical proposed the expert witness will not be able to explain the precise code used by the computer in arriving at it's conclusions. He will not have first hand knowledge as to how the computer arrived at the results, and the burden must shift to someone who can testify as to validity of the program. Because accident reconstructions utilize scientific principles the underlying methods must be scientifically valid and the evidence must authenticated as fair and accurate. (128) How should this be accomplished?

When there is a challenge as to an expert's proposed testimony regarding the cause of an accident, in order to evaluate his claims the opponent needs to know how the expert arrived at his opinion. The Federal Rules of Civil Procedure state, "A party may through interrogatories require any other party to identify each party whom the other party expects to call as an expert witness at trial, to state the subject matter on which the expert is expected to testify, and to state the substance of the facts and opinions to which the expert is expected to testify and a summary of the grounds for each opinion." (129)

When discovery reveals that a computer program has been used, some opponent litigators have requested copies of the computer program itself to attempt to identify flaws in the programming. The Manual for Complex and Multidistrict Litigation states its essential for to disclose underlying data, programs and programming techniques, and all relevant input and output far in advance of trial. (130) In Jochims v. Isuzu Motors, Inc. the U.S. District court ordered the plaintiff to turn over approximately 1,500 pages of material a number of computer disks which included the confidential source code for a vehicle simulator program. (131)

Other courts have held that producing the underlying data is sufficient, as long as the methodology is available from other sources, and source code is not required. (132) This is short sighted though. The purpose of discovery is not merely to allow a plaintiff or defendant to develop their own theory of how an accident occurred. If that were true then there would be no need for discovery of an opponent's expert witnesses at all. Everyone would simply produce their own evidence and present their own experts. Discovery of an opponent's expert witnesses is to show why he is wrong in what he says. As such, a witness who is allowed to get on the stand, say here is the underlying data, and here is what it means, is unimpeachable if he uses a computer program and the source code is not available. The computer is being as an out of court expert who is beyond the reach of cross examination. An attempt to make a Daubert challenge to his testimony is impossible because the methodology, the computer program, is unobtainable, and the court cannot fulfil the gatekeeper role that Daubert requires of it unless the underlying source code is available.

There are other benefits to requiring production of source code. When a computer program is used in a courtroom, it becomes a tool of law, not of science. The harsh examination of this tool, by a motivated adversary intently looking for short comings, can only serve to eliminate questionable implementation and assumptions that the computer program might make. The more times a program is tested and found sufficient, the closer it comes to a situation where it could possible reach judicial notice of it's accuracy (although the expert's use of a program would always be subject to review). The tool will become stronger for having gone through the fire.

A programmer who knows that his work will be displayed for litigators to poke at will have to ensure a high degree of accuracy from the start. An expert will be motivated to use only the highest quality programs, and to check his work carefully. It will cause the level of acceptable standards to be higher from the beginning. Traditionally tort law has always been looked to as a source of external quality control because it looks critically at products and engineering decisions. It has driven product safety to higher levels. In the same way production of source code would drive the quality of the programmer of these specific types of programs to higher levels.

If a car manufacturer uses a simulation program to ensure that their vehicles are safe, there are some additional incentives if they plan on using the simulation in defending tort actions. Since they already rely on these programs for the safety of the buying public, there is a general benefit to society in making sure that these programs are of the highest quality. Production of source code serves this end.

Of course production of computer source code entails it's dangers, also. Many programs are copyright material that the author makes a living by producing and selling. Most programs are probably trade secrets and entitled to special protections. But a program used in determining facts and expert opinions in litigation are far different that the latest version of Donkey Kong. Courts deal in issues that are not releasable to the general public all the time. Any court ordering production of source code would need to address this issue through appropriate nondisclosure orders. Production can also be limited to relevant portions of the code. There is no need to disclose any portion that doesn't relate directly to the issues in question.

Additionally, a software author who writes in a field like accident reconstruction should realize from the start that his work is going to be subject to strict review in an adversarial setting. He must realize that accidents often end up with drivers in either civil or criminal court. If he holds out his work a suitable for use in such a litigious field, there exists a strong possibility that the results of work will be the center of contention. The very marketing of these programs envisions their use in the courtroom. (133) What other use is a moving graphic of an accident except to prove to someone else how the accident occurred? As such it is no hardship to subject these programs to the most detailed review.

1. 542 F. 2d 111, 125 (1976)(Van Graafeiland, J., dissenting).

2. In Daubert v. Merrell Dow Pharmaceuticals, 509 U.S. 579 (1993), the supreme court held that the test taken from Frye v. US, 293 F. 1013 (D.C. Cir 1923) (The Frye test) was inapplicable to the Federal Rules of Evidence. The focus in regard to scientific evidence was whether an expert witness proposing to offering expert scientific evidence was offering scientific evidence that would assist the trier of fact in determining an issue of fact. It set a non exclusive list of factors to be considered and set the judge as the gate keeper in keeping out unreliable scientific evidence. Since then there have been numerous law review articles dealing with Daubert's effects on the Federal Rules of Evidence.

3. In Frye, The court held that novel scientific evidence could only be admitted if a majority of the scientists in the appropriate field felt that the methods and research was reliable.

4. Jay P. Kesan. "An Autopsy of Scientific Evidence in a Post-Daubert World", 84 Geo. L. J. 1985, 2020 (1996).

5. Daubert v. Merrell Dow Pharmaceuticals, Inc., 43 F.3d 1311, 1315 (1995). It is interesting to note that in remand, the excluded testimony which was the basis of the Supreme Court decision was again excluded under the new rules and the original District Court decision was affirmed again.

6. Indeed, in footnote 11 of Daubert the Court says "Although the Frye decision itself focused exclusively on 'novel' scientific techniques, we do not read the requirements of Rule 702 to apply specially or exclusively to unconventional evidence." But the Court for the northern district of Iowa collected several cases applying Daubert solely to cases involving novel scientific evidence before holding that it was inapplicable to the case at bar. Waitek v. Dalkon Shield Claimants Trust, 934 F. Supp. 1068 (1996).

7. United States v. Markum, 4 F.3d 891, 895-6 (10th Cir 1993).

8. Compton v. Subaru of America, Inc., 82 F.3d 1513, 1518 (10th Cir. 1996). For a short summary of some cases since Daubert and how different courts have applied it see Bert Black, et. al., "Winning the Expert Wars in the Age of Daubert.", SB16 ALI-ABA 13 (July, 1996) and Thomas J. Mack, "Scientific Testimony After Daubert: Some Early Returns from Lower Courts.", Trial, August 1994 at 23.

9. Joiner v. General Electric Co., 78 F.3d 524, 534 (1996)(Birch,J, specially concurring).

10. Daubert at 590.

11. Daubert at 590, footnote 8.

12. Id.

13. Id. at 593.

14. Id.

15. Id.

16. Robert F. Seltzer,"Evidence and Exhibits at Trial", 387 PLI/Lit 371, 379 (1990).

17. Michael E. Cobo, "A strategic Approach to demonstrative Exhibits and Effective Jury Presentations", 3 PLI/Lit. 359 (1990).

18. Roy Krieger, "Now Showing a Courtroom Near You...", A.B.A. J. Dec. 1992, at 92.

19. Jennifer Robinson Boyle, " State v. Pierce: Will Florida Courts Ride the Wave of the Future and Allow Computer Animations in Criminal Trials?" 19 Nova L. Rev. 371, 380 (1994).

20. Wesley C. King, M. Marie Dent & Edward W. Miles, "The Persuasive Effect of Graphics in Computer-mediated Communication", 7 Comp. in Hum. Behav. 269 (1991).

21. Adam T. Berkoff, "Computer Simulations in Litigation: Are Television Generation Jurors Being Misled?", 77 Marq. L. Rev. 829, 846 (1994).

22. Carole E. Powell, "Computer Generated Visual Evidence: Does Daubert Make a Difference?" 12 Ga. St. U. L. Rev. 577, 579 (1996).

23. "The 'Rodney King' Case: What the Jury Saw in California v. Powell ." Courtroom Television Network Videotape (1992).

24. Berkoff at 582.

25. Currently you can buy lightwave 3d for $799.00 and lesser programs for as little as $195. Windows Magazine, Ad for The Software Source, December 1996 at 341.

26. Racz v. R.T. Merryman Trucking, Inc., Civ.A. No. 92-3404, 1994 WL 124857 (E.D.Pa. 1994) at *5.

27. Fed. R. Evid. 403.

28. Daubert at 2798 (quoting Jack B. Weinstien, "Rule 702 of the Federal Rules of Evidence is Sound; It should Not Be Amended", 138 F.R.D. 631, 632(1991)) .

29. Reference Manual on Sci. Evid. 37 (1994).

30. I. Neel Chatterjee, "Admitting Computer Animations: More Caution and New Approach Are Needed", 62 Def. Couns. J. 36, 44 (1995).

31. Berkoff, Supra note 21.

32. Robert Garcia, "Garbage In, Gospel Out"; Criminal Discovery, Computer Reliability, and the Constitution", 38 UCLA L. Rev. 1043, 1090 ( 1991).

33. Id.

34. Mario Borelli, "The Computer as Advocate: An Approach to Computer-Generated Displays in the Courtroom", 71 Ind.. L.J. 439 (Spring, 1996).

35. Garcia at 1082.

36. Id. at 1078.

37. Seltzer, supra note 16 at 382.

38. Marc A. Ellenbrogen, "Lights, Camera, Action: Computer-Animated Evidence Gets Its day in Court", 34 B.C. L. Rev. 1087,1101 (1993).

39. Mark Barrish, "Disclosure of Computer Reenactments During Pretrial Discovery", 16 Hastings Comm/Ent L.J. 691, 696 (1994).

40. Paul Marcotte, The Verdict Is...Jurors Confused, Bored and Critical of Trial Attorneys, Judges, A.B.A.J., June 1990 at 32.

41. Tape supra note 23 at 8:24 timer.

42. Id. at timer 01:26.

43. Id.

44. Id. at timer 00:36.

45. Id. at timer 00:50 and 00:56.

46. Id. at 01:06.

47. Vicki S. Menard, "Admission of Computer Generated Visual Evidence: Should There Be Clear Standards?", 6 Software L.J. 325,April 1993.

48. Chatterjee Supra note 30.

49. Elaine M. Chaney, "Computer Simulations: How They Can Be Used at Trial and the Arguments For Admissibility", 19 Ind. L. Rev. 735, 1986.

50. Without pointing an accusatory finger at anyone in particular, the literature is rife with different terms and attempts to make new phrases. Additionally, some authors do not seem to grasp what different types of computer software actually does, or how that relates to an end product.

I have attempted to use words in their common sense definitions and normal usage in the computer field, although the term animation is often used to describe any moving picture generated on a computer screen.

51. Borelli, supra note 34 at 439.

52. Kathlynn G. Fadely, "Use of Computer-Generated Visual Evidence in Aviation Litigation: Interactive Video Comes to Court", 55 J. Air L. & Com. 839, 842 (Summer, 1990).

53. Cornell v. State, 265 Ga. 904(1995) says at 904: "In his remaining enumeration, appellant asserts the trial court erred in refusing to admit demonstrative evidence in the form of a computer reenactment . Whether the computer simulation ..."(emphasis added).

54. Webster's College Dictionary (1996) defines animated in part as "containing objects that appear to move in a lifelike fashion; an animated display ." (emphasis in original).

55. Barrish, supra note 39 at 707.

56. Commercial Union Insurance Co. v. Boston Edison Co. 591 N.E. 2d 165,169 (1992).

57. Webster's College Dictionary (1994) defines Simulation in part as "the representation of the behavior or characteristics of one system through the use of another system, esp. using a computer."

58. reconstruct is defined in part as "to recreate in the mind from available information." Id.

59. Menard, supra note 47 at 348.

60. 61 F.3d 45 (1st Cir. 1994).

61. Id. at 57.

62. Federal Rule of Evidence 703 provides "The facts or data in the particular case upon which an expert bases an opinion or inference may be those perceived by or made known to the expert at or before the hearing. If of a type reasonably relied upon by experts in the particular field in forming opinions or inferences upon the subject, the facts or data need not be admissible in evidence."

63. 591 N.E. 2d 165 (1992).

64. Id. at 549.

65. "Demonstrative evidence, including such items as a model, map, chart, photograph, or demonstration is distinguished from real evidence in that it has no probative value itself, but rather serves merely as a visual aid to the jury in comprehending the verbal testimony of a witness or other evidence." Michael H. Graham, Handbook of Federal Evidence, 401.2, 4th Ed., West Publishing Co., 1996.

66. Pierce v. Florida, 671 So.2d 186, 190 (1996)(emphasis added).

67. Id. at 291(emphasis added).

68. Id.

69. Civ.A. No. 91-920, 1995 WL 20667 (D.N.J. 1995).

70. Id. at 23.

71. Id.

72. The first line of the body of the Daubert decision says "[i]n this case we are called upon to determine the standard for admitting expert scientific testimony in a federal trail." Daubert at 582.

73. 880 P.2d 689 (1994).

74. Borelli, supra note 34 at 439.

75. Bledsoe v. Salt River Valley Water Users' Assoc., 880 P. 2d 689 (1994).

76. Id. at 692.

77. Id..

78. Id.

79. Id.

80. CarSim Vehicle Dynamics Software, at

81. Abdullatif K. Zaouk, et. al., "Evaluation of a Multi-purpose Pick-up Truck Model Using Full Scale Crash Data with Application to Highway Barrier Impacts.", FHWA/NHTSA National Crash Analysis Center at the George Washington University via Internet

82. Id. at page 5.

83. Id.

84. Available via the internet at

85. Michael W. Sayers and Dongsuk Han, "A Generic Multibody Vehicle Model for Simulating Handling and Braking", A paper presented at the 1995 Symposium of the International Association of Vehicle System Dynamics, Ann Arbor, Michigan to appear in a special edition of the journal Vehicle System Dynamics , 1996.

86. 151 F.R.D. 659 (D. Colorado 1993).

87. 588 N.E.2d 189, 194 (1990).

88. Kudlacek v. Fiat, S.p.A.,509 N.W. 2d 603, 617 (1994).

89. Gary T. Walker, "Motor Vehicle Accident Reconstruction", 450 PLI/Lit 419, 421 (1992).

90. Id. at 431.

91. James G. Perry, iCAR documentation (a accident reconstruction computer calculator), 1993.

92. Lawrence F. Mazer, et. al., "Expert Testimony Regarding the Speed of a Vehicle : The Status of North Carolina Law and the State of the Art.", 16 Campbell L. Rev. 191, 199 (Spring, 1994).

93. In appendix A I have a short description of a windows based reconstruction calculator I have included with this paper. I displays the various formulas it uses for making calculations. The difference is that it does not attempt to produce visual evidence.

94. One description goes like this: "Perhaps the most often used and important use of the energy analysis method is the determination of the energy consumed by the physical crushing of the vehicular structure. The vehicle's horizontal boundaries are modeled as a collection of springs which possess mathematically linear deformation characteristics. That is, the force of the deformation is directly proportional to the depths of the deformation. Once depth of deformation is known, only spring constant remains to be determined for an accurate calculation of the energy crush." Lawrence F. Mazer, at 201citing Randall L. Hargens and Terry D. Day, Vehicle Crush Stiffness Coefficients for Model Years 1970-1984 I1-I15 (1987) in footnote 50.

95. A complete listing of available SAE papers in available via internet at

96. One example is the Accident Reconstruction Journal, P.O.Box 234, Waldorf, Md 20604. It is published 6 times yearly.

97. Daubert at 593.

98. Id.

99. Id. at 594.

100. Id.

101. 82 F.3d 1513, 1518-1519 (1996).

102. Id.

103. Id. at 1516-1517.

104. FRE 705 says: "Rule 705. Disclosure of Facts or Data Underlying Expert Opinion. The expert may testify in terms of opinion or inference and give reasons therefor without first testifying to the underlying facts or data, unless the court requires otherwise. The expert may in any event be required to disclose the underlying facts or data on cross-examination."

105. Fed. R. Evid. 705 advisory committee's note (1970 amendment).

106. Barrish, supra note 39 at 691.

107. Bartley v. Isuzu Motors, Ltd., 151 F.R.D. 659 (1993).

108. Id. at 660.

109. Id.(cites omitted).

110. Id. at 661.

111. Whitley v. Gwinnett County, 221 Ga.App. 18, 22-23 (1996).

112. Id.

113. Peter Norton and Richard Wilton, The New Peter Norton Programmer's Guide to the IBM PC and PS/2, (Microsoft Press, Redmond, Washington, 1988) at 432.

114. Different languages include Basic, C, Fortran, Assembler, Pascal, and many others. Some business develop their own proprietary languages for in house use. Although there is some similarities between some languages, for the most part they are entirely different. Id.

115. Machine code looks like a series on numbers when printed out. Id at 18.





120. R.R. McHenry, "A Computer Program for Reconstruction of Highway Accidents", Proceedings of the 17th Stapp Car Crash Conference, SAE Paper 73-0980, November 12-13, 1973.

121. T.D. Day and R.L. Hargens, "Further Validation of EDCRASH Using the RICSAC Staged Collisions", SAE Paper No. 890740.

122. For instance, McHenry Software points out that a competitor's product, PC-CRASH, allows for an arbitrary choice of initial contact, which can lead to inadvertent or unintentional bias in the results gained. If unintentional bias can result, certainly intentional bias is also possible.

123. Id.

124. Mazer, supra note 94 at 199.

125. Daubert at 597.

126. Joiner v. General Electric Co., 78 F.3d 524, 535 (11th Cir. 1996)(Smith, J. Dissenting).

127. Commercial Union Insurance Co., 591 N.E. 2d at 170.

128. Chatterjee, supra note 30 at 40.

129. Fed. R. Civ. P. 26(b)(4)(A)(i).

130. Manual for Complex Litigation (3rd Ed.) 21.446.

131. 145 F.D.R. 507, 509 (S.D. Iowa 1992)

132. Commercial Union Insurance Co. at 170 (footnote 5).

133. Many of the companies offering accident reconstruction programs also offer expert accident reconstruction services.