Quantity and Quality

[David Fairhurst]

I'd like to pose an open-ended question to you all.

I know that the qualitative aspect of a Q-Q analysis is described in terms of clarity and the presence of level 1, 2 & 3 detail.

In what terms do you describe the quantitative aspect and, if it truely is quantitative, how do you quantify it?

[antonroland]

Possibly not a really helpful response but my opinion is that any form of qualitative analysis (for court purposes at least) will always come back to level 2 detail.

Having a look at the FIG images will tell you that pattern type will not always work but it is handy most of the time...level 3 detail is rarely abundant in latents but often handy where visible PROVIDED that the known exemplar is recorded well too.

So, in a very roundabout way, I have difficulty in separating the qualitative from the quantitative to begin with as we might just find ourselves back to square 1...

Pattern types have very little if any identifying value as does 3rd level detail. Nice to have when but not always available or clear...

Local agency policy, state or national standards and /or court circuits will dictate the historic number of Galton details acceptible and this is what juries, judges and other roleplayers are comfortable with.

This could always easily be quantified and has indeed been quantified for a century or so...for good or bad...this is what we know and where we come from.

I cannot see how we can (with sufficient authority, scientific foundation and whatever else including the snowflake theory) really quantify any aspect of fingerprint identification. I cannot see how we can ever really get past the leap of faith as even a century of practise will not stand up to a flawed decision by a poorly skilled or (heaven forbid, a dishonest) practitioner.

I do, however, look forward to seeing some other folks' views on this.

[Gerald Clough]

When you think about assigning a value to Quantity, you can't do it without considering what you're counting. By that, I mean that if we imagine that we actually produced a numerical value for Quantity, the contribution made by each observed feature toward an accumulated total would depend on the nature of that feature. I guess I could call it a rarity factor, although I'd have to use the term "rarity" in a very loose way, so that it could mean plain rarity of occurrence and/or how rarely such fine detail could be observed as impressed. So I think it can only be a rather complex process to develop a Quantity, since the things being "counted" can have different weights for so many different reasons.

Of course, it's both futile and perhaps even misleading to try to actually assign a numerical value. Quantity can only be a bit like using an analogue meter with a very broad and fuzzy needle. The value "feels" relatively large or relatively small, a big value being big only in relation to what I would consider small. This is not to say that it's impossible to assign a specific, reliable value for Quantity. But to do that, you'd have to be able to put every observable detail into its own class and assign each class a valid weight. If you try that, you'll see how terribly difficult it is.

To what I think was the point of the question, I'd say that Quantity and Quality are terms used to help describe a process that is far more complex than assigning values and performing some calculation with them. The examiner does not consciously consider these two ideas as measures and, in fact, does not really think of them specifically at all. In actual practice, they both simply put names to two interrelated (and arguably interdependent) aspects of the internal analogue* computing process that happens in examination. For any given examination, we can easily discuss the observations that determined our internal judgment of Quantity or Quality, but we can't really give either a specific value. That's all the terms are, ways of talking about aspects of what we observe. That's how I see it, at any rate.

* A simple analogue computer example might be an adder, the controls of which are three knobs and a red light. You set one addend on one knob and the other addend on the second knob. You then turn the third knob until the red light comes on. The scale on the third knob tells you the sum. The circuit is very basic and has no digital components. It's accurate to the degree of accuracy of the dial calibrations. In our examiner analogue process, we just don't have any numbers on the dials. We just set relative values. And the third knob, the threshold for identification knob that makes the light come on is locked in one position, and the circuit is designed so that any result from the two inputs that meets of exceeds the threshold makes the light come on.

[ER]

- number of unique features in the print
- physical area of the print

There are probably other ways to quantify a print. However, I'm not sure that separated quality and quantity is possible in analyzing a print. Any question of 'How clear is it?' is immediately followed by 'How much is there?' and asking 'How much is there?' must be followed by 'Well, how clear is it?'

[kevin]

I think quantity is more a matter of ridge counts. The minutiae are, of course, a matter of position and type but the relationship of those minutiae is what affects the identification. The interval and intervening ridge counts (and I would include 'the more the better' when it comes to this relationship i.e. quantity) between minutia are just as important as the minutiae themselves. More minutiae includes more ridge counts and better quantity. I've always viewed a latent of 'sufficient quality' as one that includes enough intervening ridge area to establish good relationships.

[Gerald Clough]

That's kind of what complicates the notion of a Quantity. It's not just the details and their characters going to Quantity, it's the number of relationships, the associated complexity of relationships that's also part of Quantity and arguably the whole of Quantity, since if you have the relationships, you have the details by default, and the details without the continuity to relate them are useless. It might be more meaningful to talk about Richness rather than Quantity. A richer density of related details. Or a richer available collection of related details. And we have to remember that the concept of Quantity arose largely as one argument in the Q-Q function that might be used to talk about threshold, as in f(Quan,Qual)>=Threshold as a condition for identification, where the required value of Threshold and the definition of f(x,y) vary with the examiner and are not well-defined anywhere but in the brain of an experienced and consistent examine but are assumed to fall within reasonable but undefined professional ranges.

(I like "Richness." It has a nice hand-rubbing feel of satisfaction, the warm feeling of having sufficient information.)

[David Fairhurst]

Okay it looks like this has gone as far as it is going to go without a little more from me so here I go...

this statement form Anton sums up what got me started on this in the first place

Local agency policy, state or national standards and /or court circuits will dictate the historic number of Galton details acceptible and this is what juries, judges and other roleplayers are comfortable with.

Although we now work to a non-numeric, Q-Q, standard for identification in the UK our lawyers are still requesting information regarding the number of points in agreement. So I thought "Why?" are they taking the number of points as being the quantitative aspect of the examination. If they are, is this a valid way to interpret the number?

Then we get this from Gerald:

the contribution made by each observed feature toward an accumulated total would depend on the nature of that feature.

And this from kevin:

I think quantity is more a matter of ridge counts. The interval and intervening ridge counts (and I would include 'the more the better' when it comes to this relationship i.e. quantity) between minutia are just as important as the minutiae themselves.

I agree with both of you on these points. We all know that minutia densities are greater in a core or delta area than in the periphery of a print. Also the occurrence of, say, ridge endings that point away from a delta is to be expected and they can be said to have a lower rarity than one that you suddenly find goes the other way.

My own thought was more like ER's statement:

- number of unique features in the print
- phyical(sic) area of the print

I'd say that the number of points visible in the print is dependant on quality as well. Clarity can fall and even though a characteristic is present on the finger it is not observable in the print and therefore cannot be counted.
All we need to quantify then is the size of the print were dealing with. That is, Quantity = the size(area) of the print. and it can be quantified in units such as square inches, square millimetres.
What do you need to have more minutiae? A bigger print.
What do you need to have larger ridge counts between any two minutia? a bigger print
Minutiae do appear more densely in certain areas but these are of a type and orientation that are dictated by the first level detail and therefore have less rarity and value. Does the fact that you have the core or delta therefore change the total value of the minutiae that you have observed? I think not.
What do you all think?

[antonroland]

All we need to quantify then is the size of the print were dealing with. That is, Quantity = the size(area) of the print. and it can be quantified in units such as square inches, square millimetres.
What do you need to have more minutiae? A bigger print.
What do you need to have larger ridge counts between any two minutia? a bigger print
Minutiae do appear more densely in certain areas but these are of a type and orientation that are dictated by the first level detail and therefore have less rarity and value. Does the fact that you have the core or delta therefore change the total value of the minutiae that you have observed? I think not.
What do you all think?

This is all good and well but to what end will we now do this quantification and against what will we compare it?

To find a latent of, say, 60 square mm (10 mm x 6 mm) and to observe that it contains 10 nicely defined and clear minutiae and 5 of lesser value is the equivalent of 1 minutia per 4 square mm...I am not quite sure where you would like to go with this.

[Gerald Clough]

All we need to quantify then is the size of the print were dealing with. That is, Quantity = the size(area) of the print. and it can be quantified in units such as square inches, square millimetres.
What do you need to have more minutiae? A bigger print.
What do you need to have larger ridge counts between any two minutia? a bigger print
Minutiae do appear more densely in certain areas but these are of a type and orientation that are dictated by the first level detail and therefore have less rarity and value. Does the fact that you have the core or delta therefore change the total value of the minutiae that you have observed? I think not.
What do you all think?

This is all good and well but to what end will we now do this quantification and against what will we compare it?

To find a latent of, say, 60 square mm (10 mm x 6 mm) and to observe that it contains 10 nicely defined and clear minutiae and 5 of lesser value is the equivalent of 1 minutia per 4 square mm...I am not quite sure where you would like to go with this.

I wouldn't say that the core or delta (taken as a "countable" feature) necessarily has a greater identification value than any other feature. It likely makes exclusion considerably easier, in the practical sense of limiting the scope of what area has to be searched. To state the obvious, core and delta have the additional character of presumptive (not an entirely reliable presumption) location. That presumptive location facilitates the examination, but the identification still depends on all the observable features appearing in the same relationships. In other words, we presume that any given print might present with a core and delta, and that puts the mere fact of their appearance more into the realm of the limited information that can be gained from observing L1 patterns.

The notion of Quantity being purely a matter of area is interesting. I don't think we really think enough about what can be derived from lack of features, from areas displaying nothing but continuous and non-diverting ridges. In a real way, every aspect of the friction ridges is "detail." The nature of the forces producing any of the commonly named details is no different from the nature of the forces that produce a continuous ridge. Continuity itself can be seen as no less significant than the named details. An event occurs in the formation of pores. At any given point at which two adjacent pores emerge, only a few different sorts of events can occur. The adjacent pore may physically abut the other. It may emerge as to form a gap. It may emerge to abut with a pore in an adjacent continuous ridge. It may not emerge at all. My list may not be exhaustive of satisfactory to all, but the point is that An emergence that continues the running ridge is no different in nature from emerging with a gap of not emerging at all (producing ending ridge) or emerging to complete a connection with an adjacent ridge (producing a bifurcation). Why is a breaking or diverging event more significant than a continuation event? As pointed out, we don't consider than without value, but do we consider it more significant that two discontinuous features are separated by more continuous ridges? Is a large count of continuous ridges a more weighty factor than a lesser count, according to our judgment that it's more probable that minutia will appear as the ridge count grows? This starts to get a little hairy, since the expectation that something will break the series of continuous ridges depends on what portion of skin we're dealing with, which tends to make it circular, because we can't know the area unless we've already decided which skin we're looking at.

I think if we had to answer, we would say it's a matter of frequency. Less likely events are more useful discriminators. We judge that for every discontinuous pore emergence, there are many continuous events. But we also have a rather high expectation that large areas in which there are no discontinuous events are less likely. When we venture into an area of plain continuous ridges, we expect to encounter something else before we go very far. (Core and delta might come into play in how strongly we expect to come across a detail according to our orientation to the core and delta.) Does the lack of any L2 detail over a significant area amount to, in itself, another contribution to Quantity? Is it itself another "detail?"

I suspect few examiners would use a broad, featureless area as an active factor in an identification, aside from it's value in describing the relationship between two details by the count of intervening ridges. But maybe it's useful to think about why that is. If it is merely a matter of the context of our accumulated experience, experience that has been related in terms of discontinuous features, that's legitimate. Our judgment of what is sufficient is almost entirely grounded in what others have judged to be sufficient and in experience that their applied sufficiency appears to exceed a threshold high enough to avoid error. But if we imagine that there is some countable Quantity of the commonly named discontinuous features that must be satisfied, whether we assign it a number of not, then I think we haven't thought enough about how we use every aspect of the observable arrangements of pore units.

So, maybe you're entirely correct that Quantity = Area. Maybe that's all it can mean. Maybe all it really means is that Quantity and Quality are just useful terms in talking about examinations and in no way should be taken to mean there would be a mechanical Q-Q analysis, if only we would quantify them.

[Boyd Baumgartner]

Apologies for going on a tanget here, but what about Orientation as a Qualitative element. Proper orientation confers qualitative aspects, which in turn can influence quantity as well.

As a practical example, the image below if oriented correctly and furthermore sourced to the correct area of friction ridge skin, confers qualitative value to the ridge events shown and drives quantitative requirements for sufficiency.



These are ridge endings extending from within the type lines of a left slant loop. Those that I talk with speak of them with a similar regard to ridge endings found in the hypothenar, relatively unremarkable. If you accept that notion, then you would need more of these events because their qualitative aspects (aside from quality as a function of ridge width/shape) are relatively unremarkable.

If there were intrinsic ridge anomalies that upped the qualitative aspects of the ridge events themselves, that would lower the threshold. This shows that there is interplay between the different aspects of Quality that is driving the quantitative requirements of the sufficiency threshold.

Thoughts, reactions?

[kevin]

I would throw the type of points you describe above in with the notion that 'not all points are created equal' and the location of those points may well be much more generic than details in another part of the finger (or palm). But ultimately I think this is because of interval relationships between the points and not the area. The fact that some parts of the finger (and palm) tend to have only one 'plane' or flow to them therefore makes the ridges in that area more generic. This may not preclude the area needing more FR skin, but in general, yeah; those areas tend to be more generic. Most areas that fall within the outlying areas of the ridge fields are probably going to be less probative in terms of ridge structures, minutiae, etc but it may not preclude sufficient detail being present in those areas. The inverse could probably be said of the 'richer' areas (to borrow from Gerald) of FR skin....

[David Fairhurst]

Thank you all for your contributions.

My original aim was to find support for the rebuttment of the notion that a minutia count can be used as an indicator of the quantity of detail. I think Gerald summed that up very well:

In a real way, every aspect of the friction ridges is "detail." The nature of the forces producing any of the commonly named details is no different from the nature of the forces that produce a continuous ridge. Continuity itself can be seen as no less significant than the named details.... Why is a breaking or diverging event more significant than a continuation event? As pointed out, we don't consider than without value, but do we consider it more significant that two discontinuous features are separated by more continuous ridges? Is a large count of continuous ridges a more weighty factor than a lesser count, according to our judgment that it's more probable that minutia will appear as the ridge count grows?

I was never going to suggest that we numerically quantify the detail even if it's as simple as measuring the area. The complimentary qualitative aspect is infinitely variable and unquantifyable. it would be like asking what's 6 x purple? It's nonsense. So we don't need to use numbers. I do feel that there may be some merit in using comparative measurements though. Like saying "The latent print has an area equal to the size of a nickel", or even "I would expect to be able to identify a latent the size of a dime, given clarity permits level 2 detail to be reliably observed."

I think the concept of "richness" is one that isn't addressed well by the Quantity=Size and Quality=Clarity model and maybe needs to be talked about as well.

I now feel validated in using this model to explain quantitative-qualitative analysis in shorter words for juries; and also in being able to argue that a minutia count has no validity as a measure of quantity.

Thanks again.

[ER]

Edit