COLLEGIUMInstitutional Wisdom for the Space Age

Hall of Shoulders

Enterprise Architecture

Roger Sessions

Roger Sessions is known for Treating complexity as a *measurable, mathematical* property of an architecture; reducing it through *partitioning*; the *Simple Iterative Partitions* (SIP) process; the claim that complexity grows faster than linearly with size and is therefore controlled only by decomposing systems into mathematically valid partitions. Adversarial reviewer-brain for space-policy, space-systems-architecture, SSA/SDA, and governance dissertations

Built

Sources

48

Primary + secondary

Citations

0

ARGOS-tracked

FTS5 Chunks

48

Retrieval index

Councils

0

Memberships

Core Concepts & Space Translation

Complexity is a measurable property of an architecture, not a feeling

Sessions's foundational claim is that architectural complexity is a function of the number of elements and the number of connections (states and dependencies) among them, and is therefore computable *before* the system is built. This reframes "this system is too complex" from a subjective complaint into a falsifiable measurement. The academic literature treats his complexity metric as a formal object to be refined and corrected (Šolić, Šebo & Jović 2011, MIPRO - "adjustment of the Sessions' methodology"), and the broader science of complex networks confirms the premise that complexity is a structural, analyzable quantity rather than an inherent mystery (Liu & Barabási 2016, 10.1103/revmodphys.88.035006).

Space translation

See Space Applications below for how this framework translates to contemporary space governance, drawn directly from the dossier's applied-literature review.

Complexity grows faster than linearly with size (the super-linear law)

Sessions's most consequential and most falsifiable claim is that the complexity of a system increases *non-linearly* with the number of its elements and interconnections: doubling the functions in a single undivided system more than doubles its complexity, because the number of possible interactions grows combinatorially. This is why "just add it to the existing system" is, for Sessions, almost always the wrong move. The control-of-complex-systems literature formalizes the same intuition - that interaction density, not element count alone, drives the difficulty of controlling a system (Liu & Barabási 2016, 10.1103/revmodphys.88.035006).

Space translation

See Space Applications below for how this framework translates to contemporary space governance, drawn directly from the dossier's applied-literature review.

Partitioning is the only reliable complexity-control lever

Because complexity is super-linear in a single undivided system (F2), the dominant strategy is to *partition* the system into smaller, autonomous subsets whose total complexity is far less than that of the undivided whole. Two small partitions of size n/2 are dramatically less complex than one system of size n. This is Sessions's translation of decomposition into a quantitative argument: partitioning is not stylistic tidiness, it is the mathematically forced response to F2. The microservices-decomposition literature is the contemporary engineering instantiation of exactly this lever - breaking a monolith into autonomous services to manage complexity and improve resilience (Abgaz, McCarren & Elger 2023, 10.1109/tse.2023.3287297).

Space translation

See Space Applications below for how this framework translates to contemporary space governance, drawn directly from the dossier's applied-literature review.

A valid partition obeys mathematical laws (synergy and dependency)

Not every decomposition is a valid partition. Sessions imports the mathematical definition of a partition (every element in exactly one subset, no element in two, no element in none) and adds design laws: elements that are *synergistic* (must change together) belong in the same partition, while *dependencies across* partitions must be minimized and made explicit. A decomposition that scatters synergistic elements across partitions, or that leaves dense hidden cross-partition dependencies, is not a real partition and does not reduce complexity - it relocates it. This is the test that separates an architecture from an org-chart.

Space translation

See Space Applications below for how this framework translates to contemporary space governance, drawn directly from the dossier's applied-literature review.

Simple Iterative Partitions (SIP) - the method

SIP is Sessions's repeatable process for producing valid partitions: identify the elements, group the synergistic ones, iterate the partitioning while measuring the complexity at each step, and stop when further partitioning no longer reduces total complexity (over-partitioning re-introduces cross-partition coordination cost). The method is iterative because the right partition is discovered, not assumed; it is measured because each candidate partition is scored by F1. The EA-complexity research that builds on and corrects Sessions treats SIP as a measurable, repeatable procedure rather than an art (Šolić, Šebo & Jović 2011, MIPRO).

Space translation

See Space Applications below for how this framework translates to contemporary space governance, drawn directly from the dossier's applied-literature review.

Simplicity is the prerequisite, not the byproduct, of governance

Sessions inverts the usual order: you cannot govern, secure, or evolve a system you cannot understand, and you cannot understand a system whose complexity exceeds the human cognitive bound. Therefore reducing complexity to within that bound is the *precondition* for every later property - agility, security, cost control, evolvability. An architecture that is "complete but incomprehensible" has failed at the one thing architecture is for. This is why fragmented, overlapping governance regimes - what the political-science literature calls *regime complexity* - are, in Sessions's terms, ungovernable by construction (Alter & Raustiala 2018, 10.1146/annurev-lawsocsci-101317-030830).

Space translation

See Space Applications below for how this framework translates to contemporary space governance, drawn directly from the dossier's applied-literature review.

Space Applications

Challenge 1: Fragmented space governance as an invalid partition (regime complexity): The single most Sessions-legible pathology in space policy is the structure of its governance. Oltrogge & Christensen (2020, 10.1016/j.jsse.2020.06.003) inventory the actual governance "system": operator self-regulation, four UN treaties, IADC and UN guidelines, ISO and CCSDS standards, and national regulators - an array of partially overlapping, non-hierarchical authorities. In Sessions's vocabulary this is not a partition at all (F4): the same element (a debris-mitigation duty, an orbital-slot claim, a frequency right) lives in two or three authorities at once, violating the rule that every element belongs to exactly one subset, while the dependencies *across* authorities are dense and largely implicit. The political-science literature names exactly this structure: Alter & Raustiala (2018, 10.1146/annurev-lawsocsci-101317-030830) define a *regime complex* as "an array of partially overlapping and nonhierarchical institutions," and argue that twenty-first-century cooperation is increasingly organized this way. Sessions's contribution is to say *why* this is dangerous and not merely untidy: by F2, complexity in an undivided, overlapping system grows super-linearly, and by F6 such a system passes the comprehension bound and becomes ungovernable by construction. The remedy is not another overlapping forum; it is a re-partition in which each governance element is assigned to exactly one authority and the remaining cross-authority dependencies are made explicit and minimized.
Challenge 2: SSA/SDA as a monolith that should be decomposed by partition: Space situational awareness is built as a sprawling system-of-systems whose complexity nobody measures (F1). The aspiration toward an all-encompassing integrated information system - visible in the AMOS work on an International Information System for orbital space data (AMOS 2017; AMOS brain, context) - is, read through Sessions, a drive toward a *monolith*: one undivided system absorbing sensors, catalogs, screening, and data-sharing functions, whose complexity by F2 grows faster than its function count. The contemporary engineering answer to exactly this problem is partitioning: the microservices-decomposition literature shows that breaking a large, complex monolith into autonomous services is what delivers scalability, resilience, and failure isolation (Abgaz, McCarren & Elger 2023, 10.1109/tse.2023.3287297). Sessions's SIP (F5) would have the SSA enterprise identify its synergistic clusters (everything that must change together to track a single object class) and assign each to its own partition with minimized, explicit cross-partition interfaces - rather than a single integrated catalog whose every new sensor type multiplies the internal interactions. A Sessions reviewer would not ask "is the catalog complete?"; he would ask for the *complexity number* of the proposed architecture and the partition that brings it under the comprehension bound.
Challenge 3: Mega-constellations as super-linear complexity injected into a shared commons: The constellation-debris problem is what F2 predicts when thousands of new elements are added to a single shared system with no partitioning of responsibility. Venkatesan et al. (2020, 10.1038/s41550-020-01238-3) frame orbit as an "ancestral global commons" - in Sessions's terms a single undivided system to which every operator adds elements (satellites) and therefore interactions (conjunction pairs), whose count grows combinatorially, not linearly, with the number of objects. The economics confirms the super-linear penalty in welfare terms: Guyot & Rao et al. (2023, 10.1073/pnas.2221343120) show that *uncoordinated* oligopoly competition between constellations reduces the economic welfare derivable from orbit use - i.e., the absence of a valid partition of orbital responsibility is directly costly. Sessions's diagnosis is sharper than "we need coordination": the orbital commons is a single partition into which all operators dump synergistic-but-unseparated elements, so its complexity is by construction super-linear and its governance is by F6 beyond comprehension. The architectural prescription is to partition the commons - by shell, by responsibility, by operator-bounded subset with explicit, minimized cross-partition conjunction interfaces - so that the total managed complexity falls below the sum of the parts.
Challenge 4: Space traffic management as a partitioning method, not another forum: STM discourse keeps proposing new integrated regimes and new coordinating bodies, which by F2 and F4 makes the problem worse: each new overlapping body adds elements and cross-dependencies to an already invalid partition. ACTA's space-capacity / STM-interaction analysis (Space capacity management and its interaction with space traffic management, 2025, 10.1016/j.actaastro.2025.01.069; ACTA brain) implicitly recognizes that STM and capacity are two partitions whose *interaction* (the cross-partition dependency, in Sessions's terms) is the hard part - exactly the quantity F4 says must be made explicit and minimized. The control-of-complex-systems science underwrites the move: Liu & Barabási (2016, 10.1103/revmodphys.88.035006) show that controllability of a networked system depends on its structure, and that decomposition into controllable subsystems is the route to tractable control. Sessions would have STM stop adding coordinating forums and instead *partition the traffic-management enterprise* - separating, say, conjunction assessment, maneuver authority, and licensing into valid partitions with synergistic elements kept together and the inter-partition handoffs (the genuinely hard interfaces) explicitly engineered and complexity-scored.
Challenge 5: Integrated everything as the anti-pattern (the comprehension bound): A recurring instinct in space architecture is to *integrate* - one integrated catalog, one integrated regulator, one integrated mission-management agent layer (the agent-based distributed-systems direction surveyed in Acta, Agent-based approaches for distributed space systems and mission management, 2025; ACTA brain, context). Sessions's whole point (F6) is that integration past the comprehension bound is not a virtue but the failure mode: a system that is complete, integrated, and incomprehensible cannot be secured, evolved, or governed, and integration is precisely the operation that drives complexity up super-linearly (F2). The corrective is counter-intuitive to the integration instinct: the more capable and consequential a space system becomes, the *more* it must be partitioned, not less, so that each partition stays inside the bound and the whole stays governable. Where the field reaches for "one system to rule them all," Sessions reaches for the smallest valid partition that keeps each piece comprehensible and the cross-piece dependencies explicit.
Synthesis: Across all five challenges the same Sessions diagnosis recurs: the space enterprise treats complexity as unmeasured and integration as virtuous, when complexity is in fact a computable, super-linear quantity (F1, F2) and integration is the operation that makes it explode. The governance literature reaches the symptoms empirically - Oltrogge & Christensen (2020) on the fragmented authority landscape, Alter & Raustiala (2018) on regime complexity, Venkatesan et al. (2020) on the un-partitioned commons, Guyot & Rao et al. (2023) on the welfare cost of uncoordinated competition - but lacks the *arithmetic* to say what to do. Sessions supplies it: measure the complexity, accept that it grows faster than size, and partition the system into mathematically valid, synergy-respecting subsets until every partition fits inside the human comprehension bound and the cross-partition dependencies are explicit and minimized. The engineering world already runs this play - decomposing monoliths into microservices (Abgaz et al. 2023) and decomposing complex networks into controllable subsystems (Liu & Barabási 2016). Space policy has not. Where Zachman audits which descriptive cells are empty and Spewak audits the order of operations, Sessions audits *the number*: how complex is this architecture, and what valid partition brings the number down? For space, the answer is almost never "integrate more" and almost always "partition correctly." ---

Council Memberships

No active memberships.

Brain Record

Brain ID: hos-sessions
FTS5 path: brains/sessions/sessions.db
Corpus: Full corpus + dossier
Last indexed: 2026-06-14

Query this brain

search_brain.py --brain sessions "<query>"

Open BrainTrust entry