CLPEX Chat Board

ER wrote:Careful around the delta. It is a pattern forced area that may contain similarities to other people.

ER wrote:This is not just a game of semantics. This distinction is vitally important for our field to move forward in becoming more accurate and more transparent. The above listed definitions of identification and exclusion come from the most current and widely accepted latent print body in the country. It is supported by the IAI, SWGFAST, and a majority of practitioners in the field.

Aside from the IEEGFII can anyone point to a validation study, a SWGFAST definition, a Problem Ident in which this concept(pattern force) is echoed or it's prevalence in erroneous ID's demonstrated?

It would seem that published research shows something different. Namely, that you are more likely to incorrectly associate minutiae from Plain Arches, which in theory and by definition are the least likely to suffer from pattern force. If pattern force was a valid phenomenon, wouldn't the most prevalent instance of matching impostor minutiae (term from the paper) come from the whorl, which has at least 3 pattern forced areas?



So, for consistency's sake shouldn't we also refrain from referring to and teaching such concepts since they are equally inaccurate, suffer from lack of transparency and are not recognized in the largest latent print body in the country nor supported by the IAI, SWGFAST or a majority of practitioners in the field?

Does your exclusionology class teach that this is a valid anchor by which to devalue minutiae when excluding?

Sure thing.

Cedric Neumann, Christophe Champod, Mina Yoo, Thibault Genessay, and Glenn Langenburg. Quantifying the weight of fingerprint evidence through the spatial relationship, directions and types of minutiae observed on fingermarks. Forensic Science International 248 (2015) 154–171.

Figure 7 shows how the delta region of a fingerprint receives a relatively high match score against 5-10% of a large database even with 12 minutiae, while 12 minutiae when including all regions of a fingerprint only receives a high match score against 0.1-1% of a large database. (see graph below)

I've had an example in my Exclusion workshop and also now in the Exclusionology class where the clearest area is near the delta. Typically, more than 20% of attendees erroneously ID this print. My experience and the above research leads me to the conclusion that pattern forced areas have less discriminating power or are generally "less unique" (to use the term incorrectly) than non-pattern forced areas.

I would argue against the idea that arches are the least pattern forced. They could be viewed as opposing outflows of loops, therefore, having pattern forced areas on both sides. The above research suggests that outflows are not as dangerous as deltas, but I also have class examples that result in bad ID's from the outflow of loops. I'd even say that (to some extent) the Madrid error was in a pattern forced area. That is, under the center of the "core" of the arch, which can tend to resemble the area under a delta.

Does your exclusionology class teach that this is a valid anchor by which to devalue minutiae when excluding?

Absolutely not! The delta or core and the minutiae around it are the best kind of features to use for exclusions. They are only devalued as features for identifications. This is because the two decisions are based on very different and somewhat opposite concepts. (I hope I'm understanding your point correctly.)

ID's require quantity and specificity or discriminating power. The White Box made it perfectly clear that we mostly base our ID decisions on quantity. It's also to some degree modified by the overall quality of the compared prints.

Exclusions are based on a different type of information in the prints. Think of how much easier it is to exclude a latent with a delta or a core. These features give confidence of where to look. There is a single point of correspondence (delta or core) and clear differences are found around it. It doesn't matter if it's a super-rare quadfircation (which would be AWESOME for an ID), as long as the features are different in the two prints... Exclusion.

The main concept is: What is sufficiency for exclusion?

josher89 wrote:This might be digressing somewhat, but when does, or can, a judge make a decision that will affect case law? Does it have to go to the USSC for case law to be set? Or, can a judge make a ruling and have that ruling be the new precedent? Something I've always wondered and maybe Ms. Steele would be able to give some insight into this. If this is better for another thread or chat board, I'm open to asking it there.

This can get complicated, but, first divide courts between federal and state courts.

State trial courts have to follow the U.S. Supreme Court (on federal const. issues), their state intermediate appellate court (if any), and their state supreme court. Decisions of other trial judges may be persuasive, but are not binding on them. Cases from federal appellate courts, other state courts, federal district courts, likewise may be persuasive, particularly the federal appellate court for that circuit and the federal district courts for that state, but are not binding.

State intermediate appellate courts have to follow the U.S. Supreme Court (on federal const. issues), their state supreme court, and try to be consistent with decisions they have made in the past (precedent/rule of stare decisis). Trial court decisions are not binding on them. They will look at federal and out of state stuff, and tend to favor their federal circuit court and sometimes the local federal trial court.

State supreme courts have to follow the U.S. Supreme Court (on federal const. issues). They try to be consistent with their earlier decisions. They are not bound by the intermediate appellate court, but often will try to be consistent with it unless they think it is wrong. They will look at federal and out of state stuff, and tend to favor their federal circuit court and sometimes the local federal trial court. The State supreme court has the final word on interpretation of state statutes, evidence rules, procedural rules, and the state constitution -- so long as the interpretation doesn't raise a federal constitutional issue.

It is possible to appeal a federal constitutional issue to the U.S. Supreme Court, but odds of review are very low.

In the federal system, same concepts -- the federal district courts are bound by their circuit's appellate court and the U.S. Supreme Court and are often persuaded by each other. They will often look to the state courts when they have to interpret state statutes. The Circuit Courts of Appeals are bound by the U.S. Supreme Court and try to stay consistent with their own decisions. They are often persuaded by other federal appellate courts and sometimes state appellate courts.

The U.S. Supreme Court, as a practical matter, does what a majority of justices agree to. It tries to stay consistent with its earlier cases and does look at what other courts and sometimes other countries do.

Add to this that judges don't like to be overruled, so they tend towards a conservative interpretation of earlier decisions. It is also somewhat easier to persuade them to do something that some other court has done first and nothing bad happened. So it is easier, if there is no Connecticut case directly on an issue, for me to say in that, say, Massachusetts has interpreted this statute in this way for 15 years and there doesn't seem to be any sign of a problem, then to say, here's this whole new way of interpreting this statute that looks like a good idea but no one has ever tried it before.

This is one of the problems with forensic challenges. There is a huge body of cases saying that X, Y, or Z technique is admissible -- some of which involved substantive challenges, many of which may not have. It is easy for a judge to follow the vast body of cases rather than venture into a new area and put limits on a discipline. You can hear the frustration in Judge Gertner's opinion in U.S. v. Green, 405 F. Supp. 2d 104, 109 (D. Mass. 2005) on restricting firearms ID testimony.

"I reluctantly come to the above conclusion because of my confidence that any other decision will be rejected by appellate courts, in light of precedents across the country, regardless of the findings I have made. While I recognize that the Daubert-Kumho standard does not require the illusory perfection of a television show (CSI, this wasn't),[5] when liberty hangs in the balance — and, in the case of the defendants facing the death penalty, life itself — the standards should be higher than were met in this case, and than have been imposed across the country. The more courts admit this type of toolmark evidence without requiring documentation, proficiency testing, or evidence of reliability, the more sloppy practices will endure; we should require more"

I have a couple of concerns based on your answer:

Pattern force isn’t even mentioned in the paper you cite so I don’t know how that qualifies as a validation study of pattern force. The term still doesn’t appear in any document held by any organization that I quoted you on above which you hold as authoritative.

A paper by the same authors referencing a tangential topic, says this:

There is an urgent need for development of standards and training to ensure consistency in the definition, selection, interpretation and use of observations made on friction ridge impressions. (emphasis mine)

You are using a non-standard definition and then using it in different ways to make your argument as it suits you which is in direct opposition to the recommendations of the paper by authors you find authoritative. And while it was convenient for you to argue a new application of pattern force once confronted with contradictory data, you have yet to overcome the data in the Chen paper that shows that a pattern with more areas of pattern force has a less likely impostor minutiae matching probability. Are you now saying that the magnitude of pattern force matters, and that less magnitude equals more likely to match? If so, this is an abstraction away from a concept that’s not even been validated . The abstract of the original paper you cite also states that the performance of the model that the spatial relationship carries more evidential weight than their type or direction, so there’s a problem with discrimination brewing on the horizon, but I’ll table that for now seeing how esoteric this discussion is becoming.

The paper I cite says:

On the other hand, the values for the delta region seems to indicate that configurations in this region are not very discriminative and that the matching algorithm is able to retrieve an abundant number of reference configurations for any given latent configuration. These results are confusing, especially in the light of the results presented in Figure 21a--‐d, which seems to indicate that the shape, minutiae direction and type of the configurations in the delta region are equally as discriminative as in the other regions.

It would appear that data is at odds with itself.

I have also yet to hear any rebuttal from the Hari Iyer presentation at the IAI this year where he specifically states that the distribution of the priors for minutiae cannot be known and therefore their equal weighting in the calculation of a likelihood ratio, makes the resulting calculations less than valid, so appealing to a box plot of data whose utility is in question is an overstatement of the data as evidence for your point, so there’s that. (which is probably the most distilled point of my post)

Lastly, you’re appealing to practices you recommended against.

ER wrote: RE: Exclusion to all others when an ID is made. That is, you have to actually compare something to exclude. You have to actually look at something to find sufficient disagreement. And no, I wouldn't bother comparing anyone else after I made an ID. But that's just the point. I'm "not comparing" anyone else. Very different than "EXCLUDING" everyone else.

Then you say:

ER wrote: ID's require quantity and specificity or discriminating power

What exactly is the common measure of specificity or discriminating power? Don’t you need to actually have to measure something outside the population of the print you are looking at in order to reference its distinctiveness? Otherwise you’re just referring to your experience, which also carries no common measure. This was the point I made in Sandy’s email thread of the same topic, when I quoted the Collins article that said

We contend that the use of verbal scales is inappropriate because it relies on the examiner’s impression and privileges untested ‘experience’. It cannot be readily assessed and it is not easy to explain the methodological frailties at trial

The second paper by the authors of the paper you cite above states this with regards to the quality of data

A large variability in the reported quality judgments and number of minutiae is observed between examiners for all trials. In addition, we can also observe differences in the decisions made by examiners perceiving similar quality and quantity of information. The variability can fairly be explained by the lack of consistency in the definition and understanding of these concepts within the fingerprint community.

I think this is exactly what you’re promoting by appealing to concepts that aren’t validated and speaking about specificity and discriminating power in a fungible way.

(For brevity's sake, the reply forced me to bounce around a little bit, so if you need me to clarify, just ask)

I do want to continue this discussion but have been unable to find the time with our AFIS upgrade going through. I'll start a new thread after I find some time.