The Mediterranean Model is the analytical template Braudel developed in his 1949 masterwork: a civilization understood as the product of geographical constraints operating over centuries, persisting through political revolutions and technological changes. The mountains channeled trade routes. The winds determined shipping seasons. The crop cycle governed labor rhythms. Empires rose and fell; the structural constraints remained. Applied to AI, the model asks: what are the mountains and winds of the current transition? The cognitive limits of human attention, the social need for belonging, the material infrastructure of compute, the institutional tempo of adaptation — these are the durable constraints within which events and conjunctures play out.
Braudel's insight was that civilizations are not primarily political or cultural constructs — they are responses to geographical constraints elaborated over long durations. The Mediterranean produced a particular kind of civilization because it had particular seas, particular mountains, particular seasonal rhythms. Change the physical base, and the civilization changes. Leave the physical base intact, and no political revolution will fundamentally alter daily life.
The model does not claim geographical determinism. It claims that geographical constraints set the space of possibility within which human choices unfold. The Spaniards could not sail against the prevailing winds any more than the Ottomans could. The Alpine passes were fixed; the timing of when grain arrived from North Africa was fixed; the range of a galley before it needed to reprovision was fixed. Civilization had to accommodate these facts.
For AI, the constraints are different but analogous. The cognitive bandwidth of a human being is fixed (working memory ~4 items). The developmental requirements of children are fixed (decades of embodied, relational learning). The institutional tempo of meaningful reform is fixed (years to decades). The physical requirements of large-scale compute are fixed (power, water, chip supply chains). These are the mountains and winds of the AI landscape. No amount of enthusiasm dissolves them.
The practical consequence is a discipline of expectation. Plans that assume human cognition will scale with compute, that institutional adaptation will match product release cycles, that children will develop on the timescale of iterations — these plans are as delusional as an admiral planning to sail against the Mistral in February. Durable success requires aligning with the constraints, not ignoring them.
The Mediterranean Model is articulated across Part I of The Mediterranean (1949), titled 'The Role of the Environment,' which opens with mountains and concludes with the human geography that the physical geography produced.
Geography as constraint, not determinant. The physical base sets the space of possibility within which human choices and institutions unfold.
Persistence through political change. Political regimes rise and fall; structural constraints remain, shaping what succeeding regimes can accomplish.
The analogical move. Cognitive, social, material, and institutional constraints function for AI the way mountains and winds functioned for the Mediterranean.
The discipline of expectation. Plans that ignore structural constraints fail at scale; plans that align with them can produce durable transformation.
Critics (including Geyl and later Trevor-Roper) argued that Braudel's emphasis on geographical constraint verged on determinism and underestimated the role of culture, ideology, and agency. The defensible position is that the Mediterranean Model specifies the space within which agency operates rather than denying agency — a discipline that the AI discourse, prone to magical thinking in both directions, would do well to absorb.