|
|
|
|
"Setting Standards In The Comparison and
Identification" We have standards for everything. Standards for our own personal behavior, standards for our children's behavior, standards for the cleanliness of the food that we eat, standards for everything. This certainly is not novel. The IAI has a standardization committee. The Chinese philosopher Mo Tzu lived about 450 B.C., and said, "For any doctrine a standard must be established. To expound a doctrine without a standard is like determining the directions of Sunrise and Sunset on a revolving Potter's wheel." I kind of like that. In this way the distinction of right and wrong, and benefit and harm cannot be clearly known. Mo Tzu, who went on to say, "Therefore for any doctrine there must be the three standards. There must be the basis, a foundation; there must be an examination; and there must be a practical application." That's very interesting for the 5th century B.C.. It is surprisingly scientific: basis; examination; and application. The theory is to be tested and verified in actual experience. But what's the relevance of this: To of why I am here; why are you listening to me; what does any of this have to do with fingerprints. Because right now, certain of the doctrines of fingerprint comparison, the emphasis here is really on the identification of latent fingerprints, has been stirred up, and certain of the doctrines have yet to establish equilibrium. For many years there was something of a standard for the comparison of fingerprints that centered around the counting of points of identification. The Galton characteristics now generally referred to as level two detail. This has become to be known as the numerical standard. In the United States there was never really a standard. In other parts of the world it was, but in the United States it wasn't. But the standard now has been challenged and the challenger is the antithesis of the numerical standard. The Challenger proposes a non-numerical standard and is typified by the term ridgeology. Dealing with ridge path deviation, width, shape, pores, edges, and other details. The level 3 details that John (Vanderkolk) was speaking of. Standards applicable to rideology are more diffuse, are more broadly strewn around the landscape. It's probably helpful to take just a moment to trace the history of what constitutes an unequivocal identification by means of fingerprints, and then the history of a certain tension that is developed between the two divergent schools of thought. Those that adhere to a numerical standard, and those who have embraced a non-numerical standard. I guess the most fundamental question is how we know that fingerprints are unique? Speaking to this group, I have trepidation in just raising the question. I run the risk of losing you. The question sounds almost patronizing but I assure you that that's not the case and if you will indulge me for a moment, because I'm trying to build to something. I think the most honest answer to that question is that no two people have been found that have the same fingerprints. No two fingers have ever been found that are identical. No two fingers have been found that are anywhere near identical. So the best evidence is empirical. But there has always been an intrinsic appeal to develop a mathematical basis, a statistical basis for fingerprint uniqueness. And this interest commenced almost at the dawn of fingerprints as a means of personal identification with Sir Francis Galton. Now all of us here identify Galton with fingerprints but before he became interested in fingerprints, Galton was already established as one of the foremost anthropologists and statisticians in the world. He had enormous stature in the scientific world, which was one of the reasons that fingerprints were so readily accepted worldwide. Once Galton gave fingerprints his stamp of approval, all systematic skepticism ceased. One kind of interesting anecdote about Galton. He was interested in physical anthropology, and interested in differences in similarity among stature of peoples and races. And at one point he went to Africa and measured African natives. His Victorian upbringing prevented him from measuring bare breasted African women. So he set up a surveyor's transit some distance from the women he wanted to measure, and he measured the angle and then by means of trigonometry, determined the measurements of his subjects. it was an innovative thing that tells us something about Galton and how he thought. He came to fingerprints later in life and incidental to his interest in fingerprints. He proposed a statistical model to explain the uniqueness of fingerprints. His was the first, but certainly wasn't the last. Other statistical models proposed by Balthazard, by Cummings and Midlo, Wentworth and Wilder, by Sclove, by Osterberg, by Kingston, and by Stoney and Thornton. But all of the statistical models that may explain fingerprint uniqueness have something common. None of them has ever been tested, maybe the last, the last being the FBI. With the exception of the recent FBI study, none of them have been rigorously tested against an extended database of fingerprints. They may pass some test of reasonableness, plausibility, but not a direct test against fingerprints. The last attempt though was by the FBI in response to a Daubert challenge to fingerprint identification and the FBI produced this study in response to U.S. vs. Byron Mitchell in Philadelphia, and I'm sure that you know this case decision was passed down in September of 1999. Not too long ago. The FBI produced a mathematical model in which 50,000 fingerprints were compared with 50,000 other fingerprints. And their model told them that it was mathematically impossible, one chance in ten to the 16th, that's one in 10 million billion, for two fingerprints to agree in more than four Galton characteristics. Well, maybe. But I suspect that a goodly number of us here have seen two fingerprints from two individuals that have agreed in just four characteristics. At least for the level two detail and with respect to a localized area of fingerprints. The late Bob Olsen, the editor of Scott's fingerprint mechanics, was fond of showing a spurious fingerprint, palm print identification with nine points of agreement, and it was cropped. His display was cropped so as to show only the two very restrictive areas in questioned, but it was a convincing display. So when I think of, or when I consider this recent study by the FBI, I'm prepared to withhold a measure of respect for the conclusions because I don't think that it's in continence with reality. So much for mathematical models. In the meantime some standard was needed, and some standard was applied. In the United States there was a very loose convention that was adopted, basically 12 points of similarity , that is 12 Galton characteristics would constitute a bomb proof identification. The British were more conservative and insisted on 16 points. Now I don't have a dog in that fight. I want to speak here to the setting of standards and not get embroiled in whether we should expect 12, or 16, or 8, or whatever. Now there may have been some fingerprint examiners that required 12 points for an identification, and counted up to 12, and insisted on 12, and wouldn't be moved off a requirement of 12. There might have been some examiners but certainly the majority did not. In my own mind what the 12 point standard, if we are going to use the term somewhat carelessly, what the 12 point standard said to me was that if I had 12 points I may as well stop counting because I certainly wasn't going to alter my thinking if I had 20 or 18 or some other number. The FBI laboratory took the position that eight characteristics was enough, particularly if they were clear. I never found that to be very helpful because whenever I have a print with only eight characteristics it's generally because the prints is a miserable print that was lacking quality. But there was no scientific basis for requiring 12 points, sixteen or any other number. The IAI recognized that and took a position in August 1973 with a declaration that there is, "No valid basis exists at this time for requiring that a predetermined minimum number of friction ridge characteristics must be present in two impressions in order to establish positive identification." And certainly there isn't anyone else that can argue otherwise. This was reinforced in June of 1995 by the Ne'urim declaration. Ne'urim is a place on the Mediterranean in Israel and it's a retreat for police officials, high-ranking police officials and it's a conference center. In summertime in Israel it is fairly cool. So it was a place where a conference was convened of fingerprint experts from all over the world. And following the conference a declaration was declared, saying the same thing that the IAI had said in 1973. The Ne'urim declaration reads, " There is no scientific basis requiring a specific number of matching ridge characteristics for a fingerprint identification." So the Ne'urim declaration certainly wasn't novel, but it was at least partly spawned by political considerations. The British had been stuck with a 16 point standard and it was chafing them. Quite understandably and they didn't want to say yea to 16 points and say nay to 15. They wanted some ammunition by means of, to depart their 16 point standard. They already had the support of the IAI, but they were looking for more a broad-based worldwide backing, and the Ne'urim conference had that sort of clout. So now the worldwide fingerprint community was behind the notion that no specific number of characteristics was needed. The ground was now fertile for something else. That's something else was ridgeology. I will announce my own position here and then move on. I am not antagonistic towards ridgeology, but I would oppose any position that said that the counting of points is inappropriate and has no legitimate place in fingerprint identification. Ridgeology, the term, was coined by David Ashbaugh in an article published in 1983. He has been very influential in the area of fingerprint identification and his work has extended as you heard here in the previous hour now to other areas by means of the application of the ACE-V concept. Ashbaugh has been an evangelist of a new way of looking at fingerprint identification. He rejects the necessity of counting Galton characteristics to a particular number. So is just about everybody else. He emphasizes third level detail. He's the Johnny Apple seed of level three detail. Ridge width, shape, edge contour, pore distribution. He does not discount Galton characteristics, level two detail, but in the way he develops his approach to fingerprint comparisons he does not allow the Galton characteristics to dominate the identification process. He develops his thinking as to how the identification process should proceed, that is the ACE-V concept. I have no real quarrel with ACE-V, although in my experience I have not found level three detail to be that helpful. If I have a print that is clear enough to show reliable level three detail, it invariably has an abundance level two detail as well. And if I have a print, a marginal print where the level two detail is shaky, then in my experience generally the level three detail is hopeless. Then in 1995 the FBI recognizing the needs for standards in the procedures for examining fingerprint evidence, brought together a number of latent print examiners to discuss the development of consensus guidelines which would preserve and improve the quality of service provided by the community nationwide. This group evolved into the Scientific Working Group of Friction Ridge Analysis Study and Technology, SWGFAST. First it was TWGFAST working group and all of that and then SWGFAST. Latent fingerprint examiners who are persuaded by the ridgeology approach and the ACE-V concept are well represented among the thirty-some members of SWGFAST. But there has developed a certain tension among those that are comfortable with a numerical standard, we may as well call them the ridge counters and most of them, I guess I would include myself among them, take no offense at that Appellation, and the group who favor a non-numerical standard. The two groups have become polarized and their issues between them have yet to be resolved. You can see some of that, the differences in the polarization in your own Web site, the Web Site for your organization. Because there are some articles put up on your Web Site from Fingerprint Whorld, one of them "Fingerprint Evidence Standards. Significant Milestone in the Change to a Non-Numerical STANDARD", and that last word standard is all in caps. It is the only word that is all in caps in the title of the process. Another article, " The Identification Process, SWGFAST and the Search For Science" as if somehow with SWGFAST, that this is the genesis of a search within the fingerprint examiner community for a search for science. And I think that's a crock. We speak of the science of fingerprints, the FBI has a publication of that title and for those of you who ought to be reminded the definition of a science is an orderly body of knowledge with principles that are clearly enunciated. And I think that the community of fingerprint examiners has met that criterion. In the Mitchell case, in Philadelphia, there was testimony from some credible people that fingerprint identification is not really a science. Then I do not think that that is a supportable position. I think that it does meet the criterion of a science as an orderly body of knowledge with principles that are clearly enunciated. But the term standard, to numerical standard and non numerical standard, is being held hostage in a dialogue between the two groups. And this is unfortunate. It is unfortunate because it deprives the profession of fingerprint examiners of a certain measure of grace to which it's legitimately entitled, and in my mind it really is absolutely unnecessary. So we have now kind of a goofy situation where just about everyone disavows a numerical standard, not counting to a particular number, but where many people disclaim counting points altogether, but continue to do it otherwise, continue to do it anyway . And this has led to the type of professional schizophrenia, and it has led to certain people being closet counters. And I think that maybe we made a wrong turn. When we decided that there was no validity to requiring a specific number of characteristics, somehow some people jumped to the conclusion that there was no utility to the counting of points. And if there was anything there's one thing that I would hope maybe you might remember that of anything I said here today, I think that would be it. That although there may be no validity to saying that there is a professional imperative to count to a particular number of points, I think that certainly there is some utility to doing that. I don't think that it necessarily followed that because there was no validity, that there was no utility. The logic does not follow. We may have thrown out the baby with the bath water. Clearly the counting of points has some functional utility. If someone tells me over the telephone that there is a 20 plus point identification, it tells me something. If someone tells me over the telephone that they can only squeeze out five points, then that tells me something as well. When I read the 1973 IAI position or the Ne'Urim declaration, I don't get the sense that either of those statements say that it is improper to count points. Only that is it unreasonable to expect an identification to be based upon a particular number. I recently reviewed a case where the latent print analyst, an adherent of ridgeology, testified that there was no particular number of characteristics needed for identification. Okay, so far so good. But since he was not required to demonstrate a particular number of characteristics, he chose not to demonstrate any at all. He went without a display of the latent and the inked print. And the identification rested then on his professional opinion, and his ability to express it, and defend it. Well, that's okay if you can get away with it. But all of us here know that going to court is like licking honey from a razor blade. If you can get away with it, its real sweet. If you can't, then it's going to hurt. It puts the premium on being able to express and defend a particular opinion of a particular person rather than having the force and logic and the power and glory of IAI and other professional organizations behind you. In this particular case in a strange reversal of reality, the defense then brought in a display of the latent and inked print and cross-examined the analyst from the defense display. So the analyst was then invited to comment on second and third level detail as evident on the defense exhibit. The danger in this is that you're now playing on the other guy's turf. You've lost control to how the evidence is to be presented. I don't think I'm over drawing the situation when I say that point counters and non-numerical standard people are somewhat polarized. Here's a direct quote from the October 1999 the issue of "The Print". "I would call for the point counters to relinquish their position and accept the standards which have now been established." Okay, what standards are those that we are speaking of? SWGFAST standards, ridgeology standards? I'm sure that in David Ashbaugh's book somewhere the word standard appears, I do know where, but I know that it does not appear in the index. SWGFAST is concerned with the big picture, global issues if you will. It isn't promulgating standards that you can find particularly helpful if you're faced with a miserable smeared partial distorted bloody print that is the only evidence in a homicide case where they're going to cut the guy loose at 2:00 unless you make the identification. So how is the standard to be set? Well, a good place to start is to recognize that in attempting to set standards we're not alone. The standards are being set by other groups, other communities, other disciplines. There are standards from everything from badger hairs to billiard balls. I'm not being smart ass when I say badger hairs, because badger hairs are used in shaving brushes and somebody showed me the standard on how much adulteration with squirrel hairs is permissible and still have the shaving brush labeled badger hair. There are organizations. For example, ASTM, the American Society for Testing and Materials and ISO, International Organization for Standardization. It's initials are backward because the formal title is in French. The organization is headquartered in Geneva. These organizations promulgate standards of all sorts. And we can learn, or we could learn much from these other efforts. For example, when a pathologist looks at tissue slice to determine whether cancer is present, he or she is making a subjective judgment call. Features are being abstracted, projected against a gestalt of past experience. Are there standards that have been developed for tasks such as this? Yes, certainly there are. Is there some common ground between other groups with their own parochial interests and our group of individuals interested in fingerprint identifications? Clearly there is. But someone, or some group of some ones, will need to identify those elements of other standardization elements, or standardization efforts that would be applicable to fingerprints. I applaud SWGAST's efforts in what they've done. And SWGFAST knows a lot about fingerprints, but they don't know much about standards. And then other groups, ISO for example, knows a lot about standards but they know diddly about fingerprints. Thus far there's not much melding of the two groups. If one of us was to go to a standardization organization, if we happened to find ourselves in Geneva for example, and were otherwise bored and stopped in and asked the ISO people for help in setting up standards for the fingerprint comparison, what would we be told? The first thing that would happen is that we would have a question thrown back at us. How we are using the term standards and standardization? Because the term standard means different things in different contexts. The dictionary would tell us that a standard is something that is established by authority, custom, or general consent as a model or example. This type of standard is the sort that may be in effect in your household for the flushing of the toilet after each use. It's a casual definition of standard. But ISO, and again ISO knows nothing about fingerprints, but they know a lot about standards, would tell us that there is a formal definition of a standard that they applied to everything. Everything. And their definition is, I'll read to you, "A document established by consensus and approved by a recognized body that provides for common and repeated use, rules, guidelines, or characteristics for activities, or their results in that the achievement of optimum degree of order in a given context." Yea, OK, that's a mouthful, but it's pregnant with implication. Its formal, and you might not like that. You might prefer something more casual than formal, but the problem of having your own casual, informal, laid-back untailored standards that are not the product of consensus, is that you will then be forced to defend them, as in court, and you won't have any assistance. It's just you. You'll notice also that the ISO definition said that a standard is established by consensus. That's really important. And the ISO definition states that a standard is a document. It's not an abstract ideal. It's not a warm and fuzzy feeling. It's not like the dream home, or the perfect orgasm, or some abstract ideal. In the view of ISO, a standard cannot exist unless it can be reduced to the written form. And there's a purpose here. It's to avoid to the extent possible any imprecision that would allow negotiation of the meaning of the standard. Back to SWGFAST. What are these standards that SWGFAST has presumably established? What are the numerical standards that ridgeology has developed? SWGFAST has really, they haven't developed standards, they have developed some guidelines thus far, and their work is not complete. Their work thus far has developed: minimum qualifications for latent print examiners; training guidelines for latent examiners, and that's a list of things you ought to know if you're going to look at latents; quality assurance guidelines. And one of the things, there is a manual for methods and procedures in detail to describe methods used for the development of latent fingerprints. But it doesn't speak to comparisons. SWGFAST has not developed standards for dealing with level three details. Now, it isn't enough just to enunciate standards simply to give voice to them. Their adequacy, their appropriateness, their relevancy, their durability must also be tested. Mo Tzu said that 2500 years ago. The real test comes not from those that like the standards, but from those that are hostile to them, like defense attorneys. The SWGFAST people know a lot about fingerprints, but they are not the ultimate authority on standards in general. And what they've come up with thus far really won't pass muster when compared to the sort of standards that a hospital laboratory must meet in order to diagnose strep throat. The sort of standards that a factory must meet in the manufacture of ball bearings. For example, what about corrective action? In every other discipline, standards speak exhaustively as to how corrective action is to proceed whenever there is a departure from the standard in question. A failure to have a well-developed position with respect to corrective action is almost certainly to harvest the criticism that these aren't really standards. They're just the undigested rumination of a group of people playing at the development of standards. SWGFAST has five lines on corrective action. Points to consider in establishing non-numerical standards. Non-numerical standards are inherently and necessarily more subjective, and that carries with it certain complexities. Because anything that's subjective is harder to communicate to someone else. I'd be very careful about setting standards that may indeed be met by somebody that has practiced in the field for decades, but then again expecting those of standards to be met by somebody new on the job. You can set someone down on their first day on the job with two fingerprint cards, and two glasses, and a pair of ridge picks, and have them count second level detail features. What would you say concerning third level characteristics to that person on their first day on the job, and how would you evaluate their work? You might say, "Well we wouldn't expect anyone on their first aid to be able to appreciate third level characteristics, that would come later." All right. But your standards must be able to accommodate this situation. SWGFAST expects a measure of standardization in training as well as casework. But when I look at the SWGFAST document entitled training to competency for latent print examiners, I don't see that this issue is really adequately addressed. There is a phrase, section 2.6 that says that an examiner must demonstrate an ability to render a proper identification decision. Okay. That's all well and good, but that is not in my view sufficiently explicit to give me much of any guideline. I don't get a compass heading from that portion of the guidelines. Identifications based on level three detail have yet to be rigorously tested either in a scientific venue or in court. I suppose that David Ashbaugh and possibly a number of people here today would take issue with that. In his book, Ashbaugh takes pains to develop the basis of ridgeology. And I think he does a commendable job of doing so, but it's an argument. You may think it's a good argument. You may think that it's a not so good argument. But it's still an argument and argument is not proof. Dusty Clark said it is well as anyone I think could: "The repeatability of the finite detail that is utilized in the comparison process has never been subjected to a definitive study to demonstrate that what is visible is actually a true third level detail, or an anomaly." Personally I don't think I've ever seen a level 3 detail comparison of a latent fingerprint that didn't require some level of rationalization. In the Inked fingerprint the ridge goes this way, but in the latent it goes that way, and then the rationalizations begin. Well yes, but this can be explained on the basis of a difference in pressure, distortion, or under processing, or over processing, or some of foreign material on the finger or surface debris or surface texture. Now how you write a standard, and it needs to be written, it needs to be reduced to writing or the lords of standards will grumble about it. How you write a standard that specifies how rationalizations are to be dealt with? There's a movie, The Big Chill, and there's a great line where someone said that rationalizations are more important than sex. And someone says why? What do you mean? And the other person says well, had you ever gone all week without a rationalization. Most of us haven't. Once standards are established you must follow them implicitly. Not following your own standard procedures is truly the kiss of death. And ASCLD accreditation has taught us that. ASCLD accredited laboratories are required to have written protocols for each type of evidence that is examined and the laboratories accreditation status would be severely compromised if the ASCLD audit team compares what was done in a given case with what the lab said it ought to have done in that case and finds a discrepancy. And defense attorneys have picked up on that. Defense attorneys demand in a discovery request copies of lab protocols, quality assurance documents, standard procedures, and so on. And if your work in a given case does not comport to the standards you yourself have adopted, then you just don't have a leg to stand on. Is there a reconciliation? Yes I think there is. I believe that there's reconciliation that is possible between the point counters and the non-numerical standard people. But in this, an outstretched hand of tolerance certainly would be helpful. And then we can proceed with the important work of developing rational standards that we all desire. There is possibly some inroads in that regard. In May of 1999 the FBI hosted an international symposium on setting quality standards for their forensic community which was held in San Antonio. A portion of the program dealt with identification criteria in fingerprint evidence. Here's what the conclusions were with respect to identification criteria. "The basis for individualization is that the friction ridge arrangement for each finger, palm, and foot is unique and permanent. Identification is a composite of several factors, each having a different relevance for each comparison conducted. The comparison process therefore is both qualitative and quantitative." And that is precisely what David Ashbaugh has been saying. That is what John Vanderkolk was saying here earlier today. It goes on to say that there are four elements to identification criteria, and this first one I think is, well, I think I will let you make up your own mind, "There must be an agreement of fiction ridge formation. At a minimum, level one and level two detail must correlate. Level three detail can be used as necessary and is generally relied upon during the comparison process, but commonly not during the identification decision, unless the quantity of level two detail is minimal and the quality of level three detail provides sufficient clarity." That seems reasonable. The other components: "The sequential relationship of all elements must be the same; the Print must be devoid of any unexplained discrepancy; there must be sufficient uniqueness to individualize. This element requires an assessment of both quality and quantity of information contained in both the unknown print and the known exemplar." Now if there's going to be an assessment of the quantity of information contained in level two detail, how else would we do this, other than to note the number of points? We can scarcely say that there's oodles, or a scad of agreement between the evidence and the inked print. What would we say about the print if we only found half an oodle, or half a scad? The business about level three detail not having been rigorously tested in a scientific venue may change in the near future however, and with respect to second level detail as well. The National Institute of Justice has a solicitation out right now for friction ridge (fingerprint) examination validation study. It is not a particularly well crafted title, but it's for a goodly chunk of money, $500,000.00. Proposals are to be made by October of this year. This is in response to something that started before 1997, but in 1997, the American Society of Crime Lab Directors (ASCLAD) requested that the National Institute of Justice (NIJ) assess the current status and needs of the forensic sciences. A number of needs were identified including the validation of the basis of print, that his friction ridge individualization. And that's what they're funding in this cycle. And standardization of comparison criteria and that is still possibly in the future. In its solicitation for the validation of basis of friction ridge individualization, NIJ says "that basic research to determine the scientific validity of individuality and friction ridge examination must be based on measurement of features, quantitative, and statistical analysis." It goes on to say, "That the validation could be performed in the context of existing systems such as AFIS", well maybe, maybe not. "Solicitation is expected to be in compliance with SWGAST standard operating procedures", and I'm not sure what that means or the implications there, "and is expected to address first, second, and third level detail. Procedures must be tested statistically in order to demonstrate that following the stated procedures will allow an analyst to produce correct results with an acceptable error rate." This has not been done. And then they love it to death, "Methodologies which may be useful include quality-control; experimental design; protocol evaluation; and cognitive and perceptual psychology." Yea, well okay. But the money is there. That's probably the important point and the research is going to be done. But hearken back now to the 1973 IAI resolution that said "No valid basis exists at this time for the requirement of a determined number". There was no valid basis in 1973. There's no valid basis in the year 2000. But in 2005 or so there may be some recently developed, recently derived basis requiring a minimum number of characteristics. I would not discount that possibility. So for anyone that's inclined to view point counting with total disdain, it may be thrust upon you again sometime in the near future. In the ongoing quest for professional equilibrium between a numerical and a non numerical standard, I would urge that you not allow a lot of momentum to build up. Momentum that would drive a wedge between the two cultures. I believe that there's room in the ark for everyone. Thank you. We have four minutes for questions. Question: I would like a show of hands of those in attendance at this presentation that do not do any ridge counting at all. Just out of curiosity. Response: It's OK with me. How many people would not do ridge counting? (No hands were raised) So it's unfortunate there is some momentum that is built up that says, that ridge counting, you count ridges, you really are doing something that runs counter to current thought. I don't buy that. The validity of the opinion is coupled with an ability to defend that position, and both are founded in one's personal knowledge, ability, and experience. And that's all well and good, but it places such a premium on a knowledge, ability, and experience. And being able to count points, even though your not counting them to a particular number, I think that that leavens the playing field of knowledge, ability, and experience. I would not want to discard the point counting. Thank you very much.
|