WHICH NOIR ELEMENTS HELP SEPARATE FILMS INTO SUB-TYPES? A LOOK AT TRIANGULATION...

We've seen that it's very possible to graph noirs against the data in the Noir-o-Meter™ (most recent post). Here is a more global look at the distribution of the "noir universe" based on the parameters used previously--a two-dimensional scatterplot mapping the level of "toughness" (hard-boiled) in the character elements vs. the overall relationship between hard-boiled elements and melodrama elements (what we call the MELO rate).

As you can see, this chart--which uses about 40% of the current Noir-o-Meter database--gives us a very good sense of how the noirs range from "tough" (to the left and up) to "tender" (to the bottom and the right).

From here, there are dozens of ways to slice the Noir-o-Meter data to look for patterns and trends. Since the different types of elements can co-exist, it's likely that many of the "noirest" films based on their raw scores (the 200-point scale that's assigned via scoring each of the 27 noir elements) will be closer to the mid-point (and you can see on the above chart where that is, as we've drawn in the averages for these two measures--it's where the red and the green lines cross). That's something we'll examine in a subsequent post.

Right now, we're interested in seeing if we can find any of the noir elements that isolate or cluster films in either the "tough" or "tender" region of noir. The elements that are likeliest to help us do this are, in fact, the ones derived from melodrama. And among the likeliest in that category is the element entitled "Degree of character triangulation."

Why this one? Because romantic triangles involve feelings and emotions. They evoke vulnerability in characters. It stands to reason that films which have a marked emphasis on character triangles are going to skew toward the "tender" side of the graph.

So we separated a sizeable subset of the Noir-o-Meter database into two groups: those which have a score of 9 or 10 in "degree of character triangulation" (an element assigned a maximum of 10 points), and those which have a score of 5 or less. We are surmising that this should produce two graphs that will give us some significant spatial separation.

Let's begin with the "tough" side...which we should find with films whose "degree of triangulation" is below average (5 or less is definitely below the overall average of 6.7/10). Here's what that chart looks like:

(Note we didn't include the axis labels that you saw on the first graph--but they are the same as above. And as you can see, we added some shading effects in the graph...because we're suckers for greyscale.)

Looks like that gives us mostly films in the upper left quadrant, which is where the "tougher" noirs are located.

Now--will we get a similar clustering when we look at the films whose "degree of triangulation" element score is 9+? Let's take a look:

And the answer is a resounding "yes." Films with a high degree of triangulation between characters have a strong tendency toward high MELO RATEs and low hard-boiled-to-melodrama character element ratios, pushing a larger preponderance of them toward the bottom right-hand region on the graph.

Of course, this too is not monolithic in nature, just as was the case for the low triangulation subgroup. But as you can see, at this level of triangulation there really are only a few films that cross over into the "tough" side of noir.

We can probably expect a graph of those films whose "degree of triangulation" scores are in the mid-range (6-8) to fall roughly halfway between these two extremes.

Which means that it looks like the "triangulation" element is a kind of bellwether for separating film noir sub-types.

We'll want to look at the films in the lower left and upper right portions of the graph, as these groups tend to resist the separation effect. Their placement may tell us something that those films have in common--higher or lower raw scores, for example.

At any rate, these uses of the data in the Noir-o-Meter provide strong indications that the system is fundamentally sound in its assumption that noir's dualism is ultimately its most defining characteristic, the quality that gives it its essential tension and allows it to be a protean form that operates across a wide range of story types, even invading established genres. Stay tuned for more...

A HARD BOILED FESTIVAL USING THE NOIR-O-METER...

While the Noir-o-Meter™ is an item that often provokes eye-rolls among aesthetes due to its quantitative approach to film criticism, its advances over the past ten years are quite promising. Film noir, that ever-elusive concept, is slippery because of its multiplicity of dualisms--and one of the most significant is the tension between the "regular rules" of (melo)drama and noir's effort to "crash in" on those rules by inverting them.

As we defined more clearly the elements that exist in varying gradations within film noir, we were able to determine which of these stemmed from melodrama--elements that are more totally assimilated into the "rules of drama"--and those which arose from noir's "hard-boiled" incursion into character and story. Noir tries to explode beyond the normal rules of drama, which is why the techniques of expressionism and the desire to "novelize" narrative, often to the point of radical disjunctures in time (flashbacks and other non-linear narrative devices) became associated with noir.

The Noir-o-Meter can tell us "how much" noir there is in a film (it can do this for "regular melodramas" such as NOW, VOYAGER, or even for comedies, in order to show us how far the "regular rules of drama" take us toward noir--about one-third to two-thirds of the way, as a general rule). More interestingly, it can show us the "shape" of that noirness by looking at the ratios between elements derived from melodrama and elements which stem from the "hard-boiled." (That latter is often mistakenly considered to be the true and only defining set of characteristics for noir, but this is a grossly inaccurate oversimplification that persists to this day due to the circumstances under which noir was first popularized by film critics.)

Two measures that the noir-o-meter calculates help us show us the total dimension of the noir universe as the films occupy a version of the "stellar sequence" (noirs are not literally "red giants" or "white dwarfs," of course: this is just a rough analogy for what is really a distribution of wide-ranging sub-types). When we look at each group of elements--those which deal with the characters, those which address the visual information that films provide over and above traditional verbal/written storytelling, and those which anatomize how storytelling is altered and shaped toward a subversive inversion of standard plot and screenwriting techniques--we can compare the intensity of the elements which stem from the "regular rules of drama" (what we call the "melodrama elements") and those elements which come from forces that subvert/invert those rules (what we call the "hard-boiled" elements).

These measures, when applied, can distribute films noirs across a universe-like grid which demonstrates how much range exists within these elements as they combine and collide in individual films. Here we are going to look at the subset of those films that fulfill the region of that grid which houses the toughest, most hard-hitting, least melodramatic, and unsentimental examples of film noir.

How do we do this? We calculate two measures from the Noir-o-Meter data and create a scatterplot for them. What are these measures? The first compares the totals for the two "element sources" (melodrama/hard-boiled) across all the three element types (character/visual/plot-screenwriting). The second focuses only on comparing the totals in the character elements. The idea is that the character elements, where the action in the film has its most direct and immediate contact with the viewer, are the ones that give us the visceral "feel" of the action. So, if the character elements show a tilt toward the hard-boiled, we will get a measure that captures just how "tough" the film seems to the viewer.

Our first chart (above) shows a group of 150 films that have a hard-boiled-to-melodrama character element rate of 100 or higher. (The average for the more than 1500 films scored in the Noir-o-Meter DB for this measure is 89). The data points to the left and toward the top of this chart show the "toughest" films.

Think of this chart as the top slice of the film noir universe, with the next slice that you'd see being the one where the bulk of the films reside--in the middle region of a noir "galaxy" where these measures are both close to the overall average. On the other side of that large region will be the films that are the opposite of our "hard-boiled" extreme--those melodramas whose intensity has been coupled with at least near-criminal behavior and darker psychological conflicts than what is customary in films that otherwise are derived from the "regular rules of drama." (We'll look at that region in a subsequent post.)

The second chart zeroes in on the top end of the region depicted in the first chart. (You should be able to tell what subset of the data is common to each). These are the 32 films where the ratio between the character element comparison (hard-boiled v. melodrama) is at least 1.5 times as strong as the overall hard-boiled/melodrama ratio (what we call the MELO RATE).

The film at the top left of the chart in each case is Anthony Mann's BORDER INCIDENT. Those of you who know the film already will probably not be surprised that it grades out in such a manner. It is a particularly tough, violent, even grisly police procedural that features nasty, brutish, even racist characters who are caught up in an issue that still plagues America today: "illegal" immigration across the US-Mexico border.

The other film that seems well-separated from the pack of "tough films" (the one at the intersection of 50 on the horizontal axis and 200 on the vertical) is another film in which Anthony Mann participated (he completed the film on behalf of the original director, Alfred Werker). Many of you will nod in recognition upon reading its name: HE WALKED BY NIGHT, a film that still astonishes many for Richard Basehart's incredibly cold-blooded portrayal of the central criminal character.

These thirty-two films would make for quite a film noir festival, one which would demonstrate the most extreme manifestation of "toughness" that could likely be displayed. It might not be the most bitter, however, since that quality often contains melodramatic overtones that would color some of the other noir elements and add character relationships that would move the MELO RATE upward (to the right, as is evident by comparing the second chart with the first).

Take a look at my Blackboard post for a list of the films that are shown on the second chart.

FILM NOIR ORIGINALS VS REMAKES PER IMDB RATINGS

An interesting little study from Solomon at the Blackboard about the IMDB ratings of classic noir originals (40s-50s-60s) and their various remakes. As the line chart (below) indicates, none of the 40+ films in the sample had the remake score a higher rating than the original, though 5 of them had scores that were at least 90% of the original:

The average IMDB score for remakes is just about 80% of the score achieved by the originals. It's also clear that the folks who opted for remakes picked some of the better films of the classic noir era, which probably made it more difficult to match the scores.

The data that would be really interesting here would be to see the average for those folk who saw each film and cast votes for both in the IMDB ratings. That's a level of detail we don't have readily available, however. Our guess is that the score gap is somewhat greater, with the remakes scoring in the low 70% range among such voters.

A-B QUALITY FOR NC 16

The quality rankings for the films screened at NC 16 in San Francisco (ongoing through Sunday, February 4) can, like so many things these days, be shown in graphic form. Are A's better than B's? Will Manny Farber and his followers be outraged?

It's WAY too small a sample size to be taken seriously, but here goes:

For these films (12 of each, a comparison afforded by the structure of NC 16), A's outpoint B's by 7 points (84.7 to 77.6). The graph confirms this via its distribution across quality levels, with each sub-group showing something akin to a bell-curve structure even though we have only 12 films in each sample.

A larger sample from the full NC sample (or from a larger portion of it that can be readily identified using the "A-B" distinction) is possible...we will try to get to that at some point in time.