The Institution Model

Why Babylon models institutions as a third entity layer between substrate and agents, how the three-faction balance of forces works, why structural selectivity matters for gameplay, and how institutions relate to Althusser’s theory of state apparatuses.

Why a Third Layer?

The Babylon simulation originally had two entity layers: substrate (SocialClass, Territory, Community) and agents (Organizations). This two-layer model has a gap. Organizations come and go – a police department can be defunded, a union can be dissolved, a factory can close. But the function that these organizations serve persists. A new police department will be created to replace the old one. A new factory will open. The function of policing, of employment, of education – these persist because the social relations that produce them persist.

Institutions are the layer that captures this persistence. An institution is not an organization. It is the crystallized social relation that generates organizations. The Department of Justice is an institution; the FBI is an organization housed within it. The Catholic Church is an institution; a specific parish is an organization housed within it. Ford Motor Company is an institution; its Detroit assembly plant is an organization housed within it.

The distinction matters for gameplay. When the player attacks an organization, they face tactical questions: can we disrupt its leadership, cut its funding, reduce its legitimacy? When the player confronts an institution, they face strategic questions: what social function does this institution serve? Can we build alternative institutions that serve the same function? What happens when we destroy the organization but the institution persists and simply spawns a replacement?

This three-layer architecture – substrate, institution, agent – models the historical observation that revolutionary movements fail not because they cannot defeat individual state organizations, but because they cannot replace the institutional functions those organizations serve. The Bolsheviks dissolved the Tsarist police; they had to immediately build a new police apparatus, because the function of policing (however redefined) did not disappear with the organization that had performed it.

Althusser’s Apparatus Classification 

Every institution carries an apparatus_type drawn from Althusser’s distinction between Repressive State Apparatuses (RSA) and Ideological State Apparatuses (ISA), extended with a third Economic category.

RSA types – RSA_EXECUTIVE, RSA_MILITARY, RSA_POLICE, RSA_JUDICIAL, RSA_CARCERAL – operate primarily through repression. They are the organs of state violence: courts, prisons, police departments, the military. Their structural selectivity makes repression cheap and education expensive. An RSA institution does not need to convince anyone – it coerces.

ISA types – ISA_EDUCATIONAL, ISA_RELIGIOUS, ISA_FAMILY, ISA_LEGAL, ISA_POLITICAL, ISA_COMMUNICATIONS, ISA_CULTURAL – operate primarily through ideology. Schools, churches, media, the family – these are the apparatuses that reproduce class relations through consent rather than force. Their structural selectivity makes education and recruitment cheap, repression expensive. An ISA institution needs legitimacy to function.

Economic types – ECONOMIC_PRODUCTIVE, ECONOMIC_FINANCIAL, ECONOMIC_EXTRACTIVE – operate through surplus extraction. Factories, banks, mining operations. Their structural selectivity makes employment and fundraising cheap, reflecting the material basis of their influence.

The classification is not decorative. It determines the default action cost modifiers that shape how organizations housed within an institution can act. A revolutionary party housed within a university (ISA_EDUCATIONAL) can recruit cheaply but cannot repress. A security agency housed within the Department of Justice (RSA_JUDICIAL) can surveil cheaply but educating the public costs double. The institution’s structure selects for certain kinds of action and against others.

The Three-Faction Balance 

The most significant departure from treating institutions as static structures is the InternalBalanceOfForces model. Within every institution, three ruling-class fractions compete for hegemony:

Liberal-Technocratic faction seeks to maintain class rule through consent – assimilation, co-optation, procedural legitimacy. When liberal-technocratic hegemony prevails, the institution favors ASSIMILATE actions and has high escalation reluctance (0.7). This faction weakens when legitimacy erodes.

Revanchist-Fascist faction seeks class rule through naked repression. When revanchist-fascist hegemony prevails, the institution favors REPRESS actions and has low escalation reluctance (0.2). This faction strengthens during crisis.

Institutionalist-Bonapartist faction prioritizes the institution’s own survival and independence over the interests of any particular class fraction. When bonapartist hegemony prevails, the institution favors SURVEIL actions with moderate escalation reluctance (0.5). This faction strengthens under external threat.

The balance shifts through alpha-smoothed dynamics:

Rising crisis intensity drives the REVANCHIST weight up.
Falling legitimacy weakens the LIBERAL weight.
External threat drives the BONAPARTIST weight up.

Weights are renormalized after each update to maintain the sum-to-1.0 invariant. When the hegemonic fraction changes, a FactionShiftEvent is generated. When the BONAPARTIST faction exceeds its threshold while both other factions are below the exclusion threshold, a BonapartistModeEvent fires.

The three-faction model captures the historical observation that state institutions are not monolithic instruments of a single class. The American judiciary, for example, contains liberal proceduralists, law-and-order revanchists, and institutionalist judges who prioritize the court’s independence above political alignment. The balance between these tendencies shifts in response to material conditions, and that shift changes how the institution acts in the world.

Structural Selectivity 

Poulantzas argued that the capitalist state is not a neutral instrument wielded by the ruling class, but a structure that selects for certain kinds of action and against others. A police department does not need to be told to repress – its structure makes repression the default, cheapest action available.

Feature 040 implements this through a two-level modifier system:

Apparatus-type defaults: Each ApparatusType has a default set of action cost modifiers. RSA_POLICE makes repress cost 0.6x (40% cheaper) and educate cost 2.0x (100% more expensive).
Institution-level overrides: A specific institution can override any apparatus-type default through action_modifiers. A reformed police department might have {"repress": 1.5, "educate": 0.8} – repression is now more expensive, education cheaper.

The lookup order is: institution override > apparatus-type default > 1.0 (no modifier). This means structural selectivity can be changed through institutional reform, but the defaults encode the structural bias.

For the OODA Loop System (Feature 032), structural selectivity modifies the cost of actions available to organizations housed within an institution. An organization that wants to educate within an RSA_CARCERAL institution pays 2.5x the normal action cost – the structure of a prison makes education enormously difficult, regardless of the intentions of the people within it.

Class Inscription 

Every institution carries a class_inscription: BOURGEOIS, PROLETARIAN, or CONTESTED. This is distinct from Organization class_character in two ways.

First, class inscription is more resistant to change. An organization’s class character can shift quickly through leadership changes or political realignment. An institution’s class inscription changes only through sustained class struggle, operating on the same alpha-smoothed timescale as the factional balance.

Second, class inscription reflects the institution’s structural relationship to class, not the intentions of its current occupants. The Detroit Police Department has class_inscription=BOURGEOIS not because every police officer supports the bourgeoisie, but because the structural function of policing – the protection of property relations – serves bourgeois class interests regardless of individual officers’ politics.

CONTESTED inscription means the institution is a site of active class struggle. A public university might be CONTESTED: it reproduces class relations through credentialing (bourgeois function) but also provides tools for critical analysis (proletarian potential). The contest is structural, not merely political.

Community Embeddedness 

The community_embeddedness graph query measures how deeply an institution is embedded in community networks. For each territory the institution occupies, the query finds community nodes with matching territory and computes an overlap ratio per community type.

An institution with high embeddedness in a community type has material presence where that community lives. A school (ISA_EDUCATIONAL) embedded in working-class neighborhoods has different political dynamics than one embedded only in affluent areas. The embeddedness score feeds into the institution’s effectiveness at ideological reproduction and its vulnerability to community-based organizing.

Relationship to Feature 031 and Feature 039 

Feature 040 sits between two existing systems:

Feature 031 (Organization Base Model) defines the organizations that institutions house. The housed_org_ids field on Institution and the HOUSES edge type create the structural link. The is_institution and institutional_persistence fields on Organization are now deprecated – their function is replaced by the Institution entity.

Feature 039 (State Apparatus AI) defines the AI decision-making for state organizations. The hegemonic_fraction_effect function bridges Features 039 and 040: it takes the hegemonic fraction from an institution’s internal balance and returns OODA modifier hints that influence how housed state organizations choose their actions.

The data flow is: Institution (Feature 040) determines structural context -> Organization OODA profile (Feature 032) is modulated by institutional selectivity and factional effects -> State AI (Feature 039) makes decisions within those constraints.