Composing the Whole

Sústēma, Systems Engineering, and the Sociological Cost of the Engineering Box

James Wolstencroft · July 2026

Archē · Paper I of III

Abstract

Call something a system today and nobody blinks. Markets are systems. Rainforests are systems. Governments, hospitals, software, families: all systems, all read through the same structural lens. The habit feels timeless, and it isn’t. The word it borrows, sústēma, began as the name for a handful of arrangements (musical, poetic, astronomical) and only later swelled to cover the whole of reality. This paper traces that swelling: from Aristotle’s organising form, through the Cold War’s forced adoption of a mechanistic systems engineering, to its culmination in Model-Based Systems Engineering. It then asks what got left outside the box when the system became an engineering problem rather than a philosophical one. Two threads run underneath the whole argument without being followed to their ends here: the history of the emergent whole, and the history of the trace that lets us track it. Each is picked up, in full, by one of the two papers that follow.

1. A borrowed word outgrows its origin

The system was not always everything. It began as the name for a few particular things.

The English “system” descends from the Greek sústēma: an organised body, a whole made of several parts, a musical scale. The word is a compound of syn (together) and histanai (to cause to stand), and behind that stands the reconstructed Proto-Indo-European root steh₂, to stand. To the Greeks, a system was literally a standing-together. And they were careful about what they allowed to stand together. Borrowed into Late Latin as systema, the word named bounded arrangements: harmonies, poetic compositions, sets of celestial bodies, troops in formation. It was reserved for cases where an organising unity was already visible. The pan-systemic habit (markets, biomes and software architectures all treated as the same kind of thing) is a modern acquisition, and a surprisingly recent one. By the 1610s “system” could describe the whole universe. By the 1630s, a set of correlated principles. By the 1680s, the vital anatomy of the animal body. Each expansion moved the word further from its modest referents and closer to the totalising instrument it has since become. The word grew. Whether our understanding grew with it is this paper’s question.

2. The whole before it had a name

Long before “system” meant everything, Greek philosophy had noticed something odd: a whole can hold a property that none of its parts possess.

Aristotle asked the question directly in the Metaphysics: what turns a heap of matter into a unified thing with properties of its own? His answer was form, an organising principle the elements do not, by themselves, possess. His example has outlived every gloss written on it. Take a syllable and dissolve it. The letters survive; the syllable does not. The syllable’s unity was therefore something more than its letters set side by side. He drew the same conclusion about flesh, and Galen later gave the mechanism a name: krasis di’ holōn, a total mixture, from which a new substance arises (in principle even a perceiving one) out of parts that perceive nothing at all. This is the oldest evidence on record that a system, properly understood, is more than additive. It deserves the full weight of its own history rather than a passing mention here, and it gets it. The tiger, the syllable, the total mixture and the nineteenth century thinkers who finally named the phenomenon are the subject of the paper that follows this one, Génesis.

3. The whole traced backward

A system that can produce an unanticipated whole raises a second question immediately: how do you recover where the whole came from?

The ancients had words for this too. The Greeks called the residue a higher reality leaves upon a lower one an ichnos: a trace. The Romans called it a vestigium: a footprint. Plato used ichnos to describe the traces that persisted in his cosmos before the Demiurge imposed order upon it. His Meno went further, and produced what may be the oldest surviving requirement for a traceable system. The dialogue distinguishes a fleeting “true belief” from stable “knowledge”, and insists that only aitias logismos, a reasoned tracing of cause, converts one into the other. An isolated fact is not yet knowledge. It becomes knowledge when it is tethered to its origin. Plotinus, Cicero and Augustine each carried the trace into new territory (mystical, political, theological), and by the nineteenth century the same concept had been formalised into the instruments that anchor modern measurement and modern engineering alike. That lineage runs all the way from ichnos to the Requirements Traceability Matrix, and it is exhaustive enough to warrant a paper of its own: Ichnos, the third in this sequence. What matters here is only the shape of the debt. Systems engineering did not invent traceability. It inherited a 2,400 year old epistemology and mechanised it.

4. The Cold War forces the system into a box

Sústēma stopped being a philosophical category and became a mechanistic one when twentieth century geopolitics demanded it.

The nineteenth century thinkers who inherited Aristotle’s insight still treated emergence as a genuine feature of reality: something a system could do that its parts, examined separately, could not predict. That appreciation was abruptly sidelined after the Second World War. Confronted with intercontinental ballistic missiles, the Apollo programme and early networked defence computing, the United States military and aerospace sectors needed a paradigm that could scale, and irreducible complexity was not something a procurement schedule could accommodate. So the new discipline borrowed its rigour from the one place that already had it: metrology. Following the 1875 Treaty of the Metre, physical measurement was already required to trace to a common reference through an unbroken, documented chain. In the 1960s a “software crisis” of chronic, runaway project failure pushed the same logic onto system design. At the 1968 NATO Software Engineering Conference the decision was taken to import hardware manufacturing rigour into software; in 1969 the US Air Force issued MIL-STD-499, the first formal systems engineering management standard, paired with a Waterfall model mandating strict, linear progression from requirement to deployment. The ancient vestigium had become the Requirements Traceability Matrix. The ancient sústēma had become an engineering deliverable.

5. The culmination: Model-Based Systems Engineering

When the discipline’s own artefacts grew too large to manage as documents, it replaced the document with the model.

By the time cyber-physical systems combined millions of interconnected software nodes with physical hardware, the document-centric approach MIL-STD-499 had formalised was collapsing under its own weight. Specifications became unwieldy. Traceability matrices became unmanageable spreadsheets. The text drifted quietly away from the systems it was meant to describe. The industry’s answer was Model-Based Systems Engineering, defined by INCOSE in 2007 as the formalised application of modelling to system requirements, design, analysis, verification and validation, from the conceptual phase onward. Its working language, SysML, gave engineers a shared graphical vocabulary for architecture, behaviour and constraint. And its practical results have been substantial. The Silesian Phoenix student rover project used structured MBSE to catch interface incompatibilities during design review rather than during physical integration, and the coupling of SysML models to virtual and augmented reality now lets stakeholders walk through a system before a single part is machined. None of this is in dispute. What is in dispute is what the model was quietly asked to replace.

6. Modelling for the sake of modelling

A discipline built to eliminate ambiguity is exposed to a pathology of its own: the model becomes the object of the effort, rather than the system it represents.

Industry retrospectives are candid about this failure mode, and explicit in warning against it: don’t start modelling for the sake of modelling. Three symptoms recur often enough to have earned names. The first is shelfware: SysML models that took thousands of hours to build and are then never touched by a downstream developer, because they are legible only to the specialists who built them. The second is “death by SysML”: a language meant to be a lingua franca across disciplines instead breeds, through its steep learning curve, a priesthood of modellers isolated from the mechanical and software engineers they were meant to serve. The third is the most dangerous, because it is invisible from inside the model: the illusion of completeness. A tool that enforces structural rigour and visual tidiness manufactures a psychological confidence that the system is now fully understood. No model computes shifting political requirements. No model computes human error, or the kind of emergence Génesis describes. When the model demands more energy than the system it describes, the discipline has defeated its own purpose. The map has displaced the territory.

7. What the engineering box left outside

The deepest cost of forcing every system into a hard engineering frame is sociological, and it was diagnosed decades before MBSE existed.

Working at Lancaster in the 1970s and 1980s, Peter Checkland applied hard systems engineering (the direct precursor to MBSE) to management and organisational problems, and watched it fail. It failed consistently, and for a structural reason: hard methods assume the system’s objective is given, agreed and quantifiable. That is true of a missile intercept. It is false of a hospital, a university, or a government department. Checkland’s Soft Systems Methodology named three casualties of the hard systems assumption. The first is Weltanschauungen: the worldviews and cultural contexts of the people inside the system, for which SysML has no native syntax; the method forces a single source of truth onto a situation actually governed by several contested ones. The second is aporia itself: the productive perplexity Plato’s Meno treats as a precondition of real learning, stripped out by a discipline built to eliminate ambiguity, and taking with it the elasticity that ambiguity gives a human organisation. The third is emergence, flattened: MBSE’s parametric simulations remain reductionist at heart, and organisational morale, institutional decay and systemic corruption trace back to no block in any Internal Block Diagram. Checkland’s conclusion still holds. Forcing a sociological reality into an engineering box does not tame its complexity. It only ignores the parts that refuse to fit.

8. Limitations and open problems

This paper leaves three of its own arguments open on purpose: two are owned by the papers that follow, and the third belongs to a different series.

The claim that Aristotelian hylomorphism anticipates modern emergence theory is asserted here and not defended; its full evidentiary chain (from Pliny’s tiger through British Emergentism) is the task of Génesis. The claim that aitias logismos is a philosophical ancestor of the Requirements Traceability Matrix is likewise stated and not proven; the complete philological and metaphysical lineage is the task of Ichnos. A third limitation is more fundamental. This paper diagnoses the sociological cost of the engineering box without proposing a remedy. That restraint is deliberate rather than accidental: the remedy belongs to a different series entirely.

Boundary notes

This paper supplies the widest lens only: the word’s own history, the shape of the engineering box it was forced into, and the sociological remainder the box could not hold. The history of the emergent whole, previewed in Section 2, belongs to Génesis. The history of the trace, previewed in Section 3, belongs to Ichnos, which carries the lineage forward to the Requirements Traceability Matrix and, from there, into the reconciliation problem the main series takes up in Breaking the Reconciliation Trap.

References

• Aristotle. Metaphysics, Book VII (Zeta).
• Plato. Meno; Timaeus.
• Galen. On the Elements According to Hippocrates.
• Metre Convention (Treaty of the Metre), Paris, 1875.
• MIL-STD-499, “Systems Engineering Management” (1969).
• INCOSE Systems Engineering Vision, MBSE definition (2007).
• Checkland, P. Soft Systems Methodology: A Thirty-Year Retrospective.
• Wolstencroft, J. Génesis; Ichnos (this series).
• Wolstencroft, J. Reasoning in the Fog (Building the Tekmerium, unpublished at the time of writing).