Antifragility as a Measure
Traditionally, space habitat proposals focused on technical specifications: structural integrity, weight, etc. O’Neill and Sherwood brought architectural considerations of ‘earthness’ and ‘spaceness’, recreating Earth-like environment megastructure or modular scale. However, this thesis argues that IPA ought to be guided by metrics of antifragility.
In Nassim Nicholas Taleb’s Antifragile (2012), the fragile breaks under stress, the robust withstands stress, and the antifragile thrives on it. These stress events belong to two domains: Mediocristan (predictable, small deviations) and Extremistan (rare, high-impact). While current space architecture strives to be robust, it only mitigates events in the Mediocristan domain such as micrometeorite impacts. To move beyond robustness, antifragile design can be measured by optionality: the number of viable responses available when conditions deviate from predictions. High optionality indicates antifragility; low optionality, fragility.
By designing for ‘earthness’ and a ‘spaceness’, and thus optimising for 21st-century humans, O’Neill and Sherwood’s orbital cities are products of predictive design methodologies. They react to known stressors by correcting uncertainties rather than accommodating them. This makes them fragile per Taleb: fragile systems “depend on things following the exact planned course, with as little deviation as possible — for deviations are more harmful than helpful. [...] predictive systems cause fragility”. To recreate Earth-like conditions, they engineer systems characterised by interactive complexity and tight coupling. While each technological solution is intended to increase safety, it leads to more coupling and increased interactive complexity, inevitably making the habitat more prone to systemic, ‘normal accidents’.
This fragility extends beyond the technical domain into the social one. By treating habitat and inhabitants as separate entities, they abstracted social structures to focus on the habitat’s legible technical issues. Yet in a closed orbital environment, technical and social systems are inextricably linked: where structural failures impact the social orders while social disruptions place further strain on technical systems, creating an overlooked feedback loop.
Their attempt to recreate terrestrial conditions leaves no room to accommodate Extremistan events: Turkey Problems (blind spots about normal accidents) and Black Swans (unforeseen, rare, high-impact disruptions). By optimising their orbital cities for stability, they remove the optionality required to survive such events.
The Fragile Megastructure
To realise his vision of an anthropocentric, biospheric terrestrial simulacrum, O’Neill’s Island Three is highly coupled, and interactively complex. Earth-like conditions such as gravity, climate, and the biosphere are tightly linked to the technical performance of the orbital city, with heavy reliance on the cylinder’s constant rotation and the structural integrity of the pressure hull. As Charles Perrow stated, in coupled systems, failure in one subsystem propagates instantaneously. Within Island Three, stress events such as a mass imbalance or a structural breach cannot be isolated due to the common point of failure and is thus terminal. Furthermore, it suffers from three obvious Turkey Problems, that is the results of recreating an idealised Earth.
Firstly, O’Neill excluded ‘inefficient’ or ‘unpleasant’ species, and further zoned food production and psychological comforts separately. This also inadvertently stripped them of redundancies, creating low-diversity monocultures. Such systems are historically fragile, the 1840 Irish Potato Famine demonstrates how monocultures collapse under blight. O’Neill’s curated biosphere thus lacks pathogen and climate-failure resilience; with minimal optionality.
Secondly, its monolithic internal volume lacks compartmentalisation. This neglects airborne or waterborne hazards. Prioritising psychological comfort via an artificial sky and continuous horizon, O’Neill thus produced a twenty-mile-long atmospheric pressure vessel with no physical crisis containment. While on Earth, localised hazards can dissipate; yet in a closed recirculating environment, chemical leaks or other environmental hazards diffuse swiftly. without the structural ability to isolate them for remediation, they quickly integrate into the closed-loop climate, turning life-support infrastructure into a vector for the disaster’s propagation. Consequently, without the ability to decouple or isolate sections of the atmosphere, a minor technical error or human oversight could become an Extremistan event leading to ruin, bypassing protective measures targeted towards more legible risks.
Finally, O’Neill’s social model assumes stability, imagining a population “under the jurisdiction of the Energy Satellites Corporation (ENSAT) [...] a multinational profit-making consortium under U.N. treaties” that keeps them on “fairly loose rein as long as productivity and profits remain high” and organising themselves within communities. He thus presumes that material needs are all that is required. O’Neill further suggested that with everyone having spent a year working in life-support maintenance crew, ‘volunteers’ can replace regular government and professional maintenance crew should they get balky, viewing inhabitants not as political agents, but simply mechanical parts of the system. This simplification with cohesion within communities is reinforced by the functionalist logic he applied to zoning. This optimistic view is unwise. As there are many critical points of failure, for instance within the life-support infrastructure, civil unrest can easily lead to ruin. Due to the lack of alternatives to governance structure, resource distribution, and conflict resolution, O’Neill had engineered a highly fragile society.
The Illusion of Robustness
On the surface, Sherwood’s modular approach appears to resolve O’Neill’s fragilities. With his functionally differentiated modules(e.g. habitation, life-support), localised failures are allowed and containment can be achieved. However, this modularity masks the fact that the whole remains tightly coupled, with individual modules dependent on the collective infrastructure. For instance, a severed life-support module is rendered non-functional, unable to obtain energy or reject heat. As a potential Black Swan, the standardised modules could share design flaws. Thus while the whole system is robust to predictable stressors, it still lacks optionality as modules are specialized and fixed, thus unable to reconfigure its functions.
Sherwood’s approach introduces a new category of ‘Normal Accidents’ risk, due to the mechanical complexity of simulating varied microgravity. One of Sherwood’s key premises is that modules simulate different microgravity conditions by spinning at different rates. This requires complicated interconnections, such as radial elevator cabs crawling along tethers with “extensible berthing mechanisms to make pressurized connections at both ends” which have to grapple onto de-spun hubs whilst in motion, or the “large-diameter, pressure-containing, human-rated de-spin joints [connecting] modules rotating at different rates” with “[i]nternal conveyors operate along spiral tracks inside the fenestrated radius tunnel”. Thus risk is concentrated in these intricate moving parts.
Sherwood’s social model is similarly fragile. Without explicit governing structures and provisions for scaling, an inherently anarchic and unstable social system is formed. He categorised his parti’s inhabitants into four, researchers and expedition members, travellers for business or leisure, operations crew-member maintaining the space systems, and security for defence of space-based assets . The majority are meant to be passengers, people who have no agency. This characterises a system whose users are separated from risks, which Taleb argues to be inherently fragile. Sherwood’s passengers cannot intervene because the system’s complexity exceeds their abilities, thus they have no agency. Their lack of optionality shows fragility.
Antifragile, Semi-Autonomous Cells
For O’Neill and Sherwood, architecture acts as a stabilising mechanism enforcing equilibrium, preventing adaptation. Their systems maintain artificial stasis within a narrow band, and when enforcement fails, collapse is total. Both lack optionality, with no capacity to generate new survival configurations under stress.
By adopting antifragility as the principle guiding design, this thesis proposes a distributed network of semi-autonomous habitation cells scaled to Dunbar’s Number(~150 individuals).
Stable social relationships at this scale are maintained through direct trust rather than abstracted bureaucracy, as Malcolm Gladwell uncovered in his analysis of W.L. Gore & Associates’ organisational structure, wherein factories are capped at 150 employees.
By seeding IPA at the scale of a Dunbar community, the society is thus composed of individuals who have skin in the game, but who know each other. Inhabitants are neither passengers nor users of the architecture, but instead co-participants within their space-ecology with both other organisms and their living habitation cells in the activity of maintaining their survival. Social and direct relationships replace the bureaucracies of Island Three, facilitating the necessary trust-formation and conflict-resolution.
While IPA utilises modularity, it rejects the functional interdependence found in Sherwood’s LEO parti of modules specialised - whether for storage, life support, or power - and connected infrastructurally into a physically, materially, and functionally connected network. It also rejects the complex connections Sherwood described, allowing its network to be loosely free-standing, accessible via space-transports. Its modules are not conceived like O’Neill’s cylinders either, isolated and megastructural with everything it needs for its inhabitants to be contently sheltered, such that they never leave beyond work-related activities.
Instead, it proposes a network of generalist cells. Recognising that specialisation is a ‘fragiliser’, IPA’s cells remain generalist to stay metabolically independent. While they share common parts and resources, they can survive alone, thus allowing the network IPA describes to be one of generalists who might develop their own niches, creating variations between cells that were not pre-allocated, but instead emergent from the specific space-ecologies nurturing them.
This decentralised model allows for the logic of small failures, where variabilities causing both mistakes and adaptations demonstrates failures and successes everyone can learn from. Failures of any single semi-autonomous cell remains a localised tragedy providing vital lessons for the rest of the network without threatening the broader population, facilitating the evolutionary system where successful adaptations, whether ecological or cultural, have the potential to spread, while failures are decoupled and contained. By replacing monumental stasis with a distributed, semi-autonomous ecology, IPA thus moves beyond the robust, laying the grounds for a truly antifragile space-native ecology with human presence in the cosmos.