Raise The Banner

[Charles Parker]

The following is from the Fingerprint and Identification Magazine, Vol. 53 No. 4; October 1971, p. 15. It is part of a IAI committee report of the Dr. Morfopoulos report regarding a latent print determined by Dr. Morforpoulos to be different from the subject being on trial. Two latent print examiners from New York had testified it was a match. This is in reference to the Kent Case in New York (more about this case later). Part of the committee report:

“This committee views the term “dissimilarities” as used by Dr. Morfopoulos as an open question of semantics, or rather, the technical definition with reference to its applied use in the field of dactyloscopy. It is common knowledge, at least within the profession, that in all cases of a comparison of a latent and known print that “dissimilarities” per se will always exist, due to the different means by which the impression is either applied to the surface, or by the method that it is reproduced or developed. These “dissimilarities” are not crucial or significant unless they occur in the ridge structure itself (discrepancy--cjp), which in the case at hand, did not occur.

From SWGFAST Glossary

DISCREPANCY
A difference in two friction ridge impressions due to different sources of the impressions
(exclusion).
DISSIMILARITY
See Discrepancy.

Again I take up my banner on these two words. With dissimilarity a conclusion of Individualization or Inconclusive is the outcome. With the word discrepancy there can only be one conclusion and that is Exclusion.

With a dissimilarity one can describe, surmise; extrapolate a reasonable cause for two features to be not similar (dissimilarity) while with discrepancy there is no reasonable explanation available.

[David Fairhurst]

I know I'm only seeing snippets of SWGFAST definitions here and there on this forum. Please correct me if i'm wrong on the following.

<Paraphrasing>
Prints can be shown to be from different sources by the presence of dissimilarities.

Dissimilarities are caused by prints coming from different sources.
</paraphrasing>

This is circular reasoning, is it not?

[Steve Everist]

You can find the definitions at http://www.swgfast.org/Glossary_Consolidated_ver_1.pdf

But it sounds like you've got the idea regarding those definitions.

It's actually discrepancies caused by being from different sources, but since the definition of dissimilarity is "See Discrepancy" they're essentially interchangeable as per the glossary.

This gets a little more confusing when you try to include the one discrepancy/dissimilarity rule in the mix. Distortion, however is not considered a discrepancy/dissimilarity.

[Les Bush]

Since SWGFAST has been given the mandate to propose fingerprint policy for North America and hopefully for international consensus why not bite the bullet and separate the two terms so there is no confusion. When we use common understanding of the term 'similar' we assume they have like appearance (pictorial qualities) and dissimilar would follow that the appearances have unlike qualities. With that understanding the two terms follow the same thought process of visual observation. The fingerprint detail under examination will be sequential, spatial and of the same type but due to variables of production/detection have altered in appearance.
Discrepancy should only mean an unexplainable difference supporting the hypothesis that there are two sources for the fingerprints under examination. The Rule about differences in two impressions would then become the One Discrepancy rule.

[Charles Parker]

I want to think you for your comments. So where do we go from here. As Les said the next step is SWGFAST, but they will not be meeting for a while on the glossary. I think the time for review on it is in 2008. I will by that time throw my two cents at them on these two words.

Where else can we go? Why not take the big leap and deal with the "Single Discrepancy Rule". Why not put something together here on this forum that will last for decades. Anybody up to the challenge.

First I think it should not be a "RULE" but a principle (Not principal). The "Principle of a Single Discrepancy" or "The Single Discrepancy Principle".

Principle=A basic truth or law. A rule or standard. A fixed or predetermined policy.

So what should it say? It needs to contain some language that if a latent is borderline that a single discrepancy is sufficient for exclusion. But it also needs to contain some language that if there is overwhelming agreement then the principle does not apply.

Come on I need some wordsmiths out there to help write this thing that a majority will agree with.

What say ye, are you game?

[Dogma]

Hi Charles,

How about this:

The single discrepancy principle:

If when comparing impressions consisting of friction ridge skin characteristics, a single discrepancy is found, that is, a single feature present in one impression and not the other, that cannot be attributed to issues relating to matrix, surface, pressure, multiple impressions, recording technique, live-scan aberration, contamination or other causation, no individualization may be effected.

Regarding the second part that you seek, what is the scientific basis for the assertion that a single discrepancy is nullified by overwhelming agreement elsewhere in the print, especially in a time of AFIS, when candidates with similar appearing prints are becoming commonplace?

The other problem is the ambiguity of terms. "Overwhelming" is like "sufficient." What does it really mean? How is it quantified? How do we define the threshold that tips us over into "overwhelming"?

[Boyd Baumgartner]

“But it also needs to contain some language that if there is overwhelming agreement then the principle does not apply.”

So let me get this straight, a discrepancy is a discrepancy unless it’s a dissimilarity? I think there’s an inherent “paint yourself into a corner” logic that prevails with the whole one discrepancy topic.

I think the underlying premise of the one discrepancy topic is accuracy vs precision as it relates to the deposition of the latent, and should be evaluated in these terms.

Formal definitions are as follows (a la dictionary.com):

Accuracy: the condition or quality of being true, correct, or exact; freedom from error or defect

Precision: The ability of a measurement to be consistently reproduced

Working from these definitions, the two terms are evaluated in the following manner:

Accuracy is measured with respect to reality

Precision is measured with respect to detail

The misdiagnosis of a discrepancy can be attributed to one of two factors:

A) The latent may not be a precise reproduction of the exemplar

B) The exemplar may be inaccurate due to injury or other physiological factors incurred since it was taken (working from file prints)

Considering that we usually only have one latent from which to work, it’s speculative, given the quality of most latents and the lack of omniscience of a suspects dermo-trauma that you can definitively recognize a singular ‘discrepancy’.

[Charles Parker]

Dogma, nice to see you back.

The single discrepancy principle: If when comparing impressions consisting of friction ridge skin characteristics, a single discrepancy is found, that is, a single feature present in one impression and not the other, that cannot be attributed to issues relating to matrix, surface, pressure, multiple impressions, recording technique, live-scan aberration, contamination or other causation, no individualization may be effected.

Dogma I like that. I also like principals that are short sentences but I have not been able to shorten yours to a point that I think is better (yet).

Regarding the second part that you seek, what is the scientific basis for the assertion that a single discrepancy is nullified by overwhelming agreement elsewhere in the print, especially in a time of AFIS, when candidates with similar appearing prints are becoming commonplace?

There is no basis that I am aware of but then Professor Thornton says the same thing about the “single discrepancy rule”, no scientific basis. I have a hard time with the concept that each rule, principle or standard must have a scientific basis. Some should but this one may be like Michele said in a different post. The single discrepancy principle is simply a quality control. Or it could be a standard which is acceptable to the majority of those practicing in the discipline. Of course finding out who is practicing it and who is not would be a task in and of itself.

The other problem is the ambiguity of terms. "Overwhelming" is like "sufficient." What does it really mean? How is it quantified? How do we define the threshold that tips us over into "overwhelming"?

You are correct. It is a term I have been grappling with for awhile. Our discipline has several ambiguious terms. The other one I am struggling with here recently is “Tolerance”. Out of tolerance, within tolerance. The old saying “I know it when I see it” just does not have any weight with me. Still struggling. However in another post I quoted a professor/scientist that basically said that in science with its precision it has never been able to show exactly how much is enough.


Boyd, you bring an interesting mix in this for me.

“

But it also needs to contain some language that if there is overwhelming agreement then the principle does not apply.”

So let me get this straight, a discrepancy is a discrepancy unless it’s a dissimilarity? I think there’s an inherent “paint yourself into a corner” logic that prevails with the whole one discrepancy topic.

Boyd, I think that discrepancy or dissimilar are just labels or tags that are subject to change depending upon the information available. For example you walk into a room and at the far end you see a piece of fruit and think to your self. An apple! But as you get closer and obtain more information you finally see that it is a pomegranate. For a latent print you initially see something that you think is just a dissimilarity produced by distortion. But looking further and studying the latent print you determine that it is a discrepancy. We do this in other things such as first impression we have a loop pattern but closer inspection reveals it to be a tented arch. We change the label.

I think the underlying premise of the one discrepancy topic is accuracy vs precision as it relates to the deposition of the latent, and should be evaluated in these terms.

Formal definitions are as follows (a la dictionary.com):

Accuracy: the condition or quality of being true, correct, or exact; freedom from error or defect

Precision: The ability of a measurement to be consistently reproduced

Working from these definitions, the two terms are evaluated in the following manner:

Accuracy is measured with respect to reality

Precision is measured with respect to detail

The misdiagnosis of a discrepancy can be attributed to one of two factors:

A) The latent may not be a precise reproduction of the exemplar

B) The exemplar may be inaccurate due to injury or other physiological factors incurred since it was taken (working from file prints)

This is good. I have seen the Accuracy and Precision before and I have always thought it needed to become a part of this discussion and others. However with A) a latent print is never a precise reproduction of the exemplar.

Considering that we usually only have one latent from which to work, it’s speculative, given the quality of most latents and the lack of omniscience of a suspects dermo-trauma that you can definitively recognize a singular ‘discrepancy’.

The statement that we cannot “definitively recognize a singular ‘discrepancy’”, blew me away. If that is true then the single discrepancy principle is false. I went back to my books and images trying to find one example of a singular discrepancy with minimum or less of agreement and a single discrepancy that moved it one way or the other. I do have shared prints with a single dissimilariy but the agreement is such that there is no doubt in mind that they are the same. Then I have some that have 2-3 discrepancies with minimum agreement, but I have yet to find one with a just a single discrepancy that I would say if I removed it that yes those two images are from the same source, but if I added it I would say no, they are from a different source. I was looking for two images where the weight of the conclusion rested only upon one feature. Perhaps you are right and the following should be the principle.

The principle of single discrepancy states: No conclusion for exclusion or individualization can be based upon a single anomalous event.

That only leaves you with inconclusive.

[Michele]

Charles,

I have 3 topics I’d like to address.

First, I don’t think we (as a profession) have adequately defined what a discrepancy is. The SWGFAST definition is:

Discrepancy
A difference in two friction ridge impressions due to different sources of the impressions (exclusion).
SWGFAST, Glossary - Consolidated 09-09-03 ver. 1.0
http://www.swgfast.org/Glossary_Consolidated_ver_1.pdf

I feel like this definition tells us the result of a discrepancy but it doesn’t tell us how to identify a discrepancy. Once we’ve determined the result (an exclusion or an ID), then it seems like we’re using the conclusion to support if something is a discrepancy or not. I think of this as circular reasoning. Even though I don’t like this definition, I don’t have a suggestion other than to stop defining things as a discrepancy.

Second, about the scientific basis for nullifying a discrepancy if there’s overwhelming evidence towards an ID. I don’t think the discrepancy is nullified, what we see still exists and we need to document it (otherwise it looks like we are only looking for information that supports our conclusion instead of looking for contradictory information). We acknowledge its existence but we shouldn’t feel obligated to explain it. It’s scientifically acceptable to say that we don’t know how it happened but there’s overwhelming evidence towards the conclusion we arrived at. Everything in science doesn’t have to be explained. Here’s an example from another science: Suppose we were trying to arrive at a conclusion that gravity exists. We keep dropping items and they keep falling towards the earth. What if we dropped an item and it didn’t fall towards the earth? Would that single event nullify our hypothesis? Of course not. We could try to figure out why this happened. Sometimes we might be able to figure out a reason and other times we might not be able to. Regardless of whether I know the reason or not, overwhelming evidence exists to support that gravity exists. If we can’t figure out the reason, we shouldn’t feel obligated to guess at a reason and insist its fact when we have no evidence to support that’s the reason.

I’d also like to add that I’m not trying to say that people shouldn’t look for visual signs of different distortional aspects. I think this is very important, I just don’t think that we can always figure out why or how something happened.

My third topic about your post is about how much is enough? What constitutes a sufficient amount? What is an overwhelming amount? What is considered to be within tolerance? I agree with you that “I’ll know it when I see it” isn’t a good enough answer. I don’t think this is unique to our profession, it’s a common question in most disciplines. Scientifically speaking, the answer is enough to justify the conclusion, enough to gain acceptance, enough to diminish doubt, enough to stand up to scrutiny, and enough so that your conclusion stands the test of time. I guess this could be referred to as a scientific standard for sufficiency, couldn’t it?

There are mathematical ways to determine how much is statistically enough data for certain studies, but I don’t think our profession can use these quite yet.

[mdavis]

I'm afraid, until we can come up with some reliable, accurate mathematical, analytical description of the complexity of the everyday latent print image, there will never be a consensus as to what constitutes "adequate", "sufficient", or any other such label to nail down a comparison's potential for accurate and absolute individualization. We just abandoned the concept of "point counting." It would seem that this is a step back in that direction, albeit with greater potential for misunderstanding and complexity. If you simply look at the current state of the art in AFIS search algorithms and their "hit list" and error rates in scoring, it becomes painfully obvious that we are nowhere near achieving such mathematical sophistication, semantics not withstanding. "Education, training and experience" may not be a very precise answer, but it is still the only answer to date if you add the traits of honesty, integrity and conservative caution.

[Charles Parker]

Michele

I feel like this definition tells us the result of a discrepancy but it doesn’t tell us how to identify a discrepancy.

I feel the word difference tells us how as the only way to determine a difference between two things is from one or more of the five senses. In this case visual/observation. Perhaps adding what those differences are in the definition would explain more of the how.

Once we’ve determined the result (an exclusion or an ID), then it seems like we’re using the conclusion to support if something is a discrepancy or not. I think of this as circular reasoning.

I did not get that from the conversation or posts. I cannot speak for everyone but a conclusion/decision/finding is the last step in my examination. Labeling what I observe through the steps of the examination and using what I observed to reach a conclusion is what I was talking about. I would not recommend reaching a conclusion and then labeling something to justify it.

Second, about the scientific basis for nullifying a discrepancy if there’s overwhelming evidence towards an ID. I don’t think the discrepancy is nullified, what we see still exists and we need to document it (otherwise it looks like we are only looking for information that supports our conclusion instead of looking for contradictory information).

Again I must have missed something as I did not see where someone recommends nullification if there is overwhelming evidence. I think the discussion went to what is the definition of overwhelming. Yes it still exists and yes we need to document it, before the conclusion/decision/finding is reached. But it is just a label to describe an event in the documentation. Did you ever look at something and in the beginning felt it was a bifurcation but after further examination determined that it was just fill and it was an ending ridge? Just a label arrived at during the examination.

We acknowledge its existence but we shouldn’t feel obligated to explain it. It’s scientifically acceptable to say that we don’t know how it happened but there’s overwhelming evidence towards the conclusion we arrived at. Everything in science doesn’t have to be explained.

Amen to that. Yes we may not know exactly what is going on with the distortion. We should not guess at what is going on. Giving different possibilities as to how it occurred to me is the best route. However the position that we should not feel obligated to explain is a double edge sword. Have you ever seen an examiner just describe away something by saying---“I did not consider that as it is just distortion”. Being too descriptive is bad because sometimes we do not know. Being too cavalier in using distortion as a means to explain away something that should be considered is not good either.

My third topic about your post is about how much is enough? …. I agree with you that “I’ll know it when I see it” isn’t a good enough answer. I don’t think this is unique to our profession, it’s a common question in most disciplines. Scientifically speaking, the answer is enough to justify the conclusion, enough to gain acceptance, enough to diminish doubt, enough to stand up to scrutiny, and enough so that your conclusion stands the test of time. I guess this could be referred to as a scientific standard for sufficiency, couldn’t it?

I agree that how much is enough is something common in science, is subjective and sometimes difficult to get a good idea of. It is kind of like trying to quantify “Love”. Just not going to happen. But we can always try to refine, redefine to get a little closer to the concept. After making a difficult examination and before I put it to paper I run through these questions. 1) What conclusion did I reach in this examination? 2) What specific details support my conclusion? 3) What additional evidence supports my conclusion? 4) What underlying principals support my conclusion? 5) What disagreements might be raised about my conclusion? 6) In what ways should the conclusion be qualified? I cannot speak of your “scientific standards” as I am not qualified in that area, but those I just listed are my “analytical questions for my conclusions”.

There are mathematical ways to determine how much is statistically enough data for certain studies, but I don’t think our profession can use these quite yet.

I agree, but I believe it is coming.

[mdavis]

Most of what we do still takes place between our ears. Can we imagine the bottleneck that would occur if we were to have to document each and every detail, each L1D, L2D, L3D on each and every latent print that we analyzed, evaluated and/or compared on each and every case? We are, in fact, beginning to see the potential damage being done here in the world of drug chemistry in which only selected samples are being considered for analysis and many are simply ignored to speed workflow. Can we afford to ignore even one latent print of value? Conversely, can we afford (if indeed it were possible) to exhaustively document our entire thought process during ACE?

If so, then this should rightly spill over to every witness. Consider an eye-witness to a scene and the documentation requirements forced upon a poorly educated bystander who is marginally literate. Consider a simple police report of an alleged incident. No documentation, no testimony, no evidence, no conviction.

[Charles Parker]

Hello M. Davis, I hope you are doing fine.

Documentation is a worthy subject for its own special thread. But I am going to try and take a stab at it here. I do not document every detail of each latent (full level documentation Vs modified documentation). Especially if it is a street easy latent print. I do document (full level) those that I consider challenging or as the current terminology states “Complex”. My documentation is a combination of visual and written. In some weeks I will not document to the full level at all and in others I may do full level documentation on 3-4 latent prints.

ASCLD-LAB does not call for documentation to entail the “entire thought process”. They even go as far to say that they do not. All they do say is documentation must be present that any other qualified examiner can pick that up and see how you arrived at your conclusion. Of course that requirement is open to a lot of different interpretations. Having collected a number of analysis worksheets over the years, some require an extraordinary amount of documentation and some are very simple. All come from accredited labs. The variety is amazing.

The following is my POV and I will probably get slammed for it, but on street easy latent prints I see no reason to fully document the substrate (surface), matrix and development medium of the latent print. Someone turns in a lift card that is a powdered latent from the window of a card and that is already documented on the lift card, I see no reason to put it on a work sheet. However ETTR (exception to the rule), if the latent is complex and that information was used to help formulate my conclusion at the end of my examination then I will document it.

Bottom line, I use my best judgment as to when and how far to document my examination. If it does not meet the technical reviewers level, then I will document some more.