Polythetics and the Boeing 737 MAX [1]


In 2018, assisted by a Pentagon missing in action (MIA) team from Kaiserslauten, I excavated the remains of a WWII US bomber shot down near the village of Abtsdorf on the edge of the Austrian Alps in February 1944. Although the eight dead had been recovered in June 1945 and conveyed to a war grave in France, the location of the crash of Flight 42-31870 had dropped off the official radar and my project had rediscovered the site. We had used LiDAR: flown topographic laser data which essentially strips away the vegetation cover to provide a high-resolution digital contour map of the underlying terrain, including the hill of Kronberg, overlooking Abtsdorf village and the picturesque Lake Attersee in the beautiful Sound of Music landscape of the Salzkammergut.

In the images we observed some mysterious ramparts in dense woodland. These would turn out to be Bronze Age in date, but we could also see that the prehistoric site was peppered with craters. I initially assumed these were from jettisoned WWII bombs. On closer examination, however, the big depressions were found to contain a lot of plexiglass, including one large piece that—having been a model-plane nerd as a child (OK, I still am)—I was fairly sure came from the forward bombardier’s cupola of a B-17 Flying Fortress. We are still doing the post-excavation analysis of this aircraft, which will be published to the same analytical standards as all our discoveries.

The point here is a simple one, however: I guessed that the aircraft was a Flying Fortress using the standard technique, not unique to archaeology but very central to it, of typology: some of the fragments could be referred to a type. Nevertheless, no two instances of the type are exactly the same. That is to say, the boundaries of formal variance among the nearly 13,000 B-17s built by Boeing are complex. Our example was a B-17G-30-BO from the production block of two hundred aircraft, 42-31732 to 31931, and was given its unique designator and paint job, The Queen of the Skies.

2018 and 2019 saw two catastrophic air crashes involving another kind of Boeing aircraft: the domestic transport vehicle known as the Boeing 737 MAX. An archaeologically-informed analysis of typology and classification can tell us something about what went wrong, by explaining why we have a useful but also dangerous tendency to view our constructed artefacts as if they were natural kinds, akin to chemical elements or biological species, and therefore ascribe to them the idea of shared essential characteristics. There is a rather daunting specialist vocabulary that goes with this, but I believe the concepts covered by polythetic entitation, isochrestism, reverse isochrestism (which I am road-testing for the first time here) and emic versus etic taxonomy, could and should be of wider interest.

Polythetic entitation is a way of understanding fuzzy-edged groups of things, the products of human technology. It is easiest to understand by contrast with a biological entity, such the sub-phylum vertebrata. If you want to know whether the cat asleep on your chair is a vertebrate, you check whether it has a backbone. If it does, it is. This is an example of monothetic classification, where a particular attribute—the backbone—is sufficient grounds for making the call and (and this is important) at the same time, a necessary attribute.

Compare that situation with wanting to know whether the piece of furniture the cat is asleep on is actually a chair. That may sound crazy, as you know it is a chair. But you do not know it in the same way as you know that the cat is a vertebrate because—as it turns out—there is no single attribute that is at once sufficient and necessary to define your piece of furniture as a chair. It seems natural to dispute this. Does a chair have four legs, or three, or some other support? Refine the target to four-legged chairs and we find there are four-legged tables—which are not chairs—so possessing four legs could never be a sufficiently exclusive characteristic to define the object in front of you as a chair. There has to be a place to sit on a chair, but a sofa has to have that too, as does a stool. And the material variation is even greater. I could carve a chair out of a block of ice for example (and Inuit ice building does include carving sitting and sleeping platforms from ice). Chair expectations (what is normal vs. what is unusual) thus depend on the historical timeframe, the environment, and the culture in question.

Recent Western chairs display some standard conjunctions of attributes: wood and leather (or equivalent) dining chairs, with four legs, as found in homes and restaurants. But when chairs are bolted down in rows and given numbers (in a theatre for example), we refer to them as seats instead. Here the difference between the etic and the emic emerges, being the difference between the objectively measurable attributes (or, if you like, the scientific description) of the object, and the subjectively experienced cultural expectation—the feeling we know what a chair should be like, even if we cannot ultimately pin down an essence. What we think a chair should really be involves the emic, cultural insider, and often tacit and implicit definition.

We once had an oak corner chair with a triangular seat and three legs; it was a formal outlier of the fuzzy-edged (i.e. polythetic) category ‘chair’, and was equivalent in use with more familiar four-legged chairs. The term ‘isochrestic’ was coined by the anthropological archaeologist James Sackett to cover the idea of functional equivalence. His aim in formalizing such a concept was part of an attempt to better define another concept—style. Sackett’s idea was that if two things look different but function in the same way as one another, then any differences in their appearance can be seen as purely stylistic. Style thus becomes the residual difference among members of a class of artefact that all function equally well (or badly): one chair is made of dark wood, carved with hearts, and the other one, identical in shape and size, is made of light wood carved with roses. These chairs are isochrestic. The analysis is not perfect however: as Sackett swiftly realized himself, function is defined by context (difference in style can serve the function of identifying personal property for instance). So stylistic difference can always function to distinguish one example of something from another example.

In an extreme case, you could argue that the emic classification ‘chair’ in itself serves as the monothetic, at-once-sufficient-and-necessary, attribute of a chair, irrespective of what empirically measurable, etic, attributes it has: it must be a chair if we call it a chair. We might, however, be asserting a relationship between one thing and others that might be disputed. One of the spoof advertisements dreamed up by the great British humourist J. B. Morton involves his favourite fictional commercial company, Snibbo: ‘At the Breakfast Table’ reads: Snibbo is this week launching instant bacon. You drop a beautifully packaged powder into hot milk. The whole thing coagulates in the form of rashers. ‘It tastes like spinach,’ said an executive, ‘but it’s bacon all right. Look. You can see the word “Bacon” on the packet.’ Did a similar logic and motivation operate among Boeing executives in their claims about the 737 MAX?

A 737-badged Boeing aircraft was first certified for flight by the US Federal Aviation Authority in 1967. The aircraft was 28.6 m long and carried up to 103 passengers; in 2019, the distant descendant of that aircraft model, the 737 MAX-10, was 43.8 m long and carried 230 passengers. In between, there have been all sorts of civilian and military variants, and the plane (‘the plane’) was immensely successful (so that in 2005 one quarter of all large commercial airliners worldwide carried the 737 badge). However, certain decisions, made at the very outset, constrained how aircraft of this kind could evolve. Now, I realize by talking about descent (in a genealogical sense), and evolution (in the sense of gradual change over time), I am already potentially getting caught up in a biological metaphor—almost as if I thought 737s got together and had babies, each generation similar to but different from themselves. Manufacturing firms, who make cars, or aircraft, or computers, use the terms ‘generation’, ‘next generation’ and so on to describe salient step changes in parts of a design chain which has both continuities and discontinuities. But how do we measure these changes, and who decides (at Boeing or elsewhere) which changes are radically discontinuous? When does one artefact type become another?

As an archaeologist, I face this issue all the time: it is the sword and dagger problem. The Bronze Age inhabitants of Kronberg lived at a time when stabbing swords were giving way to longer, slashing swords, which led to the stabbing function moving more distinctly in the direction of the production of the shorter, later Bronze Age daggers. These were the very artefact types that previously, in the Earlier Bronze Age, had been progressively lengthened from their knife origins into stabbing swords. Presumably people in the different phases of the Bronze Age knew what they meant when they called for a knife or a dagger or a sword (stabbing or slashing), or whatever names they applied to whatever breakdown of ‘types’ they themselves recognised. Going by the formal attributes alone, the outside observer may have little chance of deciding where the boundaries between forms and functions once lay, and will not know if the archaeological museum categories correspond to the emic ones of the objects’ producers.

Let me set that thought to one side for a moment and consider the constraints on the form of the original 1967 737-100: to make this aircraft commercially competitive, the engines were mounted under wings that were themselves very close to the ground; the whole plane was built close to the ground, with low landing gear, reducing weight and, more importantly, making the engines really easily accessible for maintenance.

The initial downside of this was that, for short-haul flying in wilder areas where runways were rougher, a modification kit of shields, blowers, and deflectors was needed (available from 1969 and still present in the so-called ‘gravel kit’ used by companies like Air Inuit today). The low stance made for an aerodynamic challenge right from the off, especially as the aircraft was built without a fuel-eject system, meaning it might be forced to land shortly after take-off, still weighing close to its original weight. This meant that the 737 was the first aircraft where standard landing had to involve an active autopiloting system, and where the shape of the leading edge of the wing had to be regulated in two ways: with leading-edge slats (narrow retractable airfoils) and leading-edge (or Kruger) flaps. This is fine if everything works, but it is certainly a complex trimming system.

The eventually fatal downside revealed itself by degrees, as the aircraft type was modernised throughout the 1990s with new-generation engines which had ever-larger air intakes to boost fuel efficiency. The result was that the engines had to be moved progressively further forward, with the engine nacelles becoming flattened into a ‘hamster pouch’ look. In the 737 MAX, the top of the nacelle is situated significantly higher than, and forward of, the leading edge of the wing, creating conditions of powerful additional lift that, while allowing the aircraft to make spectacularly steep take-offs, can also lead it to stall. It was for this reason that it seems Boeing quietly installed extra sensors and software, to force the aircraft nose back down automatically without really letting on that this aircraft had become aerodynamically significantly unlike the 737-100 of 1967.

It is not my purpose to detail the catastrophic failures in the MAX system on two occasions, in 2018 and 2019, which led to the deaths of 346 passengers and crew. They are the subject of ongoing inquiries and clearly new information will continue to become available (at the time of writing Boeing has issued a safety warning in relation to manufacturing faults on the aforementioned lift-controlling wing slats on three hundred 737-badged aircraft).

What I want to discuss is rather more philosophical and practically theoretical, as it may shine a different kind of light on how things could have gone so badly wrong. Part of the problem lies, for certain, with language, and how we extend the name of a single entity or object to others we take to be of the same class. ‘My cat is a vertebrate’, we might say, ‘and therefore has a spine’—and all cats are, and do. Can we say the same thing of a 737—that it is a 737, and has feature x, and therefore all 737s must also have feature x, as it is the sine qua non of a 737?

As an archaeologist, I am used to looking for diagnostic typological features: the plexiglass I found in the ‘craters’ on Kronberg immediately suggested to me an event after 1928, when methyl methacrylate synthetic polymer was first produced, and the shape of the fragments looked like part of an aircraft cupola. This suggested a B-17, but in fact not all B-17s had plexiglass; in addition, some other aircraft, such as the Lockheed P-38 Lightning, had similar cupolas of the same material; finally (and although in this context it is not practically relevant, it is analytically important) not all plexiglass cupolas made for B-17s were finally fitted—some remained spare parts. This may sound like an arcane argument, but the issue is one of where, or in what attributes, does ‘the aircraft’ essentially reside. Polythetics tells us that, whilst there is a sweet spot of features where most B-17s ever made will aggregate in terms of their attributes, no attribute is simultaneously sufficient and necessary to assign the category.

In thinking about how language could mislead us, it might help to switch back to the 737 case, where we imagine decisions were made in terms of putting bigger engines on a 737 (implying that a 737 could exist without engines); of building a longer fuselage for a 737; of adding winglets to a 737 (winglets are the bits that curve up at the very end of the wing: split in the 737 MAX in the form of split-tip winglets). In each case what we somehow accepted is a core identity, with just one additional aspect then being put on, built in, or added to. Yet it is the sum of all these attributes that is represented finally by whatever the aircraft itself is.

These modification processes, changing one thing and then another, have been going on, iteratively, for over fifty years at Boeing in the 737 program, with only two stable factors: the name of the aircraft type, and the unbroken development sequence (the history—which actually begins with a borrowing of 60% of a previous aircraft, the 727, made from 1963 onwards, including ‘its’ fuselage; and the 727 had inherited its nose and cockpit unit from the yet earlier 707). Yet, because of the name—the emic designator (remember Snibbo’s new ‘bacon’)—the FAA certification of the fourth-generation 737 in 2017 involved only light-touch approval. The only really rigorous test-result underpinnings were those of the 1967 machine. Whether it is an accumulation of small changes, or one critical change, that leads a performance threshold to be crossed, is as hard to pin down as the definition of particular performance thresholds themselves. What can be said, in this grey area, is that building redundancy into design plays to safety, while trusting single systems plays to economy.

It is time to return to Sackett’s provocative and complex concept of isochrestism: the idea that style resides in the areas that spill over from function; or, at least, that objects can look different but be ‘equivalent in use’. Despite a lot of literature on technology and design, the reverse of the isochrestism concept does not yet seem to have been characterised or defined. To my way of thinking, it seems that there may be a sort of reverse isochrestism where form stays the same while the demands of function change. The ‘same’ vehicle has to be incredibly more fuel efficient, carry twice as many people twice as far, and becomes modified in thousands of ways from what it was—the ‘it’ in fact being no more than the ghostly trail of an empty name, and the historical trail of the design sequence that was clustered around it. But the name and some aspects of the form carry on. Imagining myself as a future archaeologist, excavating a crash site that I did not know was a 737 MAX, and finding the nose and windscreen elements, I might well make the identification that this was a Boeing 707, as the elements are virtually identical. It may not simply be that the very old-fashioned-looking nose and windscreen on the 737 MAX is just an ugly leftover from 60s design; perhaps it is deliberately (albeit implicitly) retro, so as to act as a smokescreen for all the critical engine, wing and body changes.

In reverse isochrestism, a new function is desired, but a familiar form is thought to be needed, perhaps to support the impression that no change of name is needed, and no novel threat exists. This is rather like skeuomorphism, the phenomenon that describes what happens when a new material supplants an old one and both nostalgia and familiar expectation modify the new one in favour of the old (thus we like a smart office desk to have a surface that references or looks like wood, even though in most cases it is a laminate surface). But here I am trying to get at something a little different.

Some people argue that technological artefacts have clearly defined, essentially monothetic, attribute features, that they are ‘memic’—memes being a sort of cultural counterpart to genes. I am not among them. I do admit that things can be slavishly reproduced (and it is no surprise that the word ‘meme’ has attached itself to the internet, where the same packets of binary information get reproduced as a meme spreads; other quasi-memic artefacts, like coins, may better be dealt with as type-instances of a single artefact—that is, be considered as ‘partitive multi-part artefacts’, similar in some ways to the ‘tokens of types’ in Wollheimian art-historical object analysis—but I digress). Generally speaking, as we move from the smallest attribute to the largest, most complex, artefact, every stage can only be defined in fuzzy edged, polythetic terms.

A recently unveiled prototype design for a new generation air passenger transport under development by the technical University in Delft, The Netherlands, and KLM Royal Dutch Airlines, the Flying-V (inspired by the form of a 1958 Gibson Flying-V guitar) has twin fuselages that are also the wings; these fuselages—or wings—come together at the pilot’s—or sharp—end; the twin engines are top mounted at the rear. Ergonomically this may prove very effective, making massive fuel savings, but whether it will be able to manoeuvre aerodynamically is questionable. The key point here with the Flying-V is that it reveals both ‘wing’ and ‘fuselage’ as polythetic entities too. The prototype design has both or neither. Perhaps it has dual fusewings? Eventually, novel attributes and artefacts will stimulate and prompt new word applications (computer, mobile, and the merger of the two in the smartphone). Conversely,  when old words remain associated with emergent forms there is a danger that an unconscious constraint is placed on the evolution of the new form. The Flying-V story also shows that the inspiration (the ancestry, if you will) for an innovation can just as well come (and may even more effectively come) from a musical instrument as from conventional design history (the conventional idea of the transmission of unitary and definable technical elements in a genealogically-related, quasi-biological sequence—something that the idea of memes tries to promote as reality).

I took great interest in Sackett’s ideas, not least because I am interested in style from the aesthetic perspective, and also in whether aesthetics themselves are pervaded by some appreciation of fitness for purpose. Put it this way: Boeing have built some great aircraft but I did not like the look of the 737 MAX even before it had its calamitous accidents: I didn’t like the huge pendulous air-intakes protruding in front of the stick-thin wings—it all looked out of kilter. As a model-aircraft-building enthusiast as well as an archaeologist, I often think that when form follows function, and function reaches an apogee, an object that is somehow balanced and beautiful is produced. But it could be a prejudice, and I cannot back it up. The B-17—the ‘flying hedgehog’ (‘Fliegender Stachelschwein’) as the Axis called it—was a truly great aircraft, and I sincerely hope Boeing will haul the MAX situation back, with new technology and software that provides the correct level of safeguard.

Evolution means nothing without environment, and environment, at least for humans, is reflexively defined, both by other artefacts and technologies, and by value systems. In 1967, jet fuel was cheap and concerns about anthropogenic climate effects were hardly discussed, so the environment which made a low-set fuselage and easy-to-reach engines the best choice was one where labour cost figured as the key competition element. To return to chairs, I often wondered why three-legged stools and tables were very much more common in the past than today, even though they are far less stable. As an archaeologist, I should have seen at once that, standing on older and less even flooring, a tripod design is far more stable, as all three supports touch the floor simultaneously. Only when floors became regularly level did the environment that chairs have to survive in change enough for four-legged furniture to start outcompeting its three-legged rivals. And by the time that happens, expertise in the old way starts to get lost.

We no longer have our three-legged corner chair. After providing our guests with some exciting moments, and after several attempts to repair it, I put it on the bonfire: a once fine concept, it was badly executed and became a frustrating and potentially dangerous curiosity object.

Being able to name things is a huge advantage, but it is very easy to mistake the name of a thing with the thing itself, and to assume the same essential stability of identity, and thus of function. Calling something a cat, a chair, a B-17, or a 737 MAX works very well in specific contexts, but philosophically speaking the danger is that an essentialist claim is advanced. A cat is only a cat in a specific historical setting (one of the most destabilizing aspects of Darwin’s idea of species was that, when set in a long-term perspective, the set of all cats loses boundaries and identity: synchronically monothetic it is diachronically polythetic). That is cognitively challenging enough (and I have some sympathy with evolution deniers who would like words to fit to things, each according to its type, and no blurred edges), but it is a level more unsettling with technology, which evolves even faster than biology. Perhaps, paradoxically, as we pick up our telephone, the rapid rate of change makes us cleave all the more to the familiar name, as the only sure thing about it. In reality, when William E. Boeing founded his company in 1916, he was living in a world where telephones were just becoming established, and were distinguishable from slightly older camera technology on the one hand, and cutting edge television technology on the other. A century later we still fondly maintain the three categories even while the objects to which they refer are dying as separable units. Whether anyone will want to keep the designator ‘Boeing 737’ rolling over, as newer technology develops, is a moot point.