The Shirky Principle, as Kevin Kelly named it in a 2010 essay, is the observation that institutions — whether corporations, government agencies, professional associations, or educational institutions — will systematically act to preserve the problems they were created to solve, because the problem's existence is the condition of the institution's continued relevance. The principle is not a claim about individual malice or conscious self-interest. It is a structural observation about how institutional incentives align: the people whose careers, identities, and livelihoods depend on the institution's continued function will, collectively and without explicit coordination, resist any change that would eliminate the problem the institution addresses. The principle's power lies in its predictive specificity. It predicts not what institutions will say about disruptive change — they will typically welcome it rhetorically — but what they will actually do, which is to absorb, redirect, or neutralize the change in ways that preserve their core function.
The principle applies with particular force to the AI transition because AI threatens to dissolve or radically alter the problems that many established institutions were created to address. Professional software firms exist to solve the problem that software is hard to produce; AI tools make software production accessible to non-specialists. Universities exist in part to solve the problem that knowledge is scarce and hard to acquire; AI tools make knowledge instantly accessible at a level of sophistication that would have required graduate training a decade ago. Publishers exist to solve the problem that producing and distributing written material requires significant capital and expertise; AI tools produce written material at marginal cost.
The predictable institutional response to these threats follows the pattern the principle describes. Professional software firms have not responded to AI by acknowledging that their core value proposition has been commoditized; they have responded by rebranding their offerings around AI integration while maintaining the pricing structures, staffing levels, and organizational forms appropriate to the previous era. Universities have responded to AI not by restructuring curricula around the capacities AI does not replicate but by developing policies to detect and punish AI use — preserving the assessment mechanisms that the technology has rendered unreliable rather than replacing them.
Shirky's own confrontation with the principle in his work as NYU's Vice Provost for AI and Technology in Education produced some of the most candid institutional analysis in the current debate. His observation that universities know what they need to do but 'don't know how to do it' acknowledges the structural bind: the institutional actors with the authority to implement necessary changes are the same actors whose roles would be most diminished by those changes. The principle does not predict that the changes will not occur; it predicts that they will be slower, more partial, and more contested than a straightforward analysis of the technology's capabilities would suggest.
The principle has a corollary that this book calls the ascending friction thesis: institutions that successfully adapt do so not by lowering their standards to match the new reality but by relocating their function to a level of difficulty the technology cannot reach. Shirky's proposed medieval turn in education — toward in-class assessment, oral examination, real-time demonstration of knowledge — is an ascending response in exactly this sense. The lower-level assessment (the take-home essay) has been rendered unreliable; the higher-level assessment (the live demonstration of understanding) ascends to a level where AI cannot substitute for human performance.
Kelly formulated the principle in a 2010 blog post responding to Shirky's observation, in a panel discussion, that 'institutions will try to preserve the problem to which they are the solution.' The observation itself was drawn from Shirky's earlier work on institutional adaptation, but Kelly's naming crystallized it as a portable analytical tool that has since been applied across domains from healthcare to telecommunications to higher education.
Structural incentive alignment. The principle does not require individual malice; the institutional incentive structure produces problem-preservation through the aggregate of rational individual choices.
Predictive specificity. The principle predicts not rhetoric but behavior; institutions disrupted by change typically welcome it rhetorically while acting to preserve their core function.
The ascending adaptation response. Institutions that adapt successfully relocate their function to a higher level of difficulty rather than abandoning it; the function survives, the specific mechanisms change.
Scope of application. The principle applies to any institution whose existence depends on a specific problem's persistence — which is, on examination, most institutions.
The governance implication. Because incumbent institutions cannot be expected to adapt at the pace that disruptive change demands, new institutions must often be built alongside the old ones; reform is insufficient.
The principle has been criticized as too cynical — reducing institutional behavior to self-interest in ways that obscure the genuine commitments that motivate institutional actors. The counter-response is that the principle is not a claim about motivation but about aggregate behavior; individual actors may be genuinely committed to the institution's stated mission while the institution as a collective produces problem-preserving behavior. The sharper criticism is that the principle understates the capacity of institutions to genuinely transform themselves when confronted with sufficiently severe challenges, pointing to historical cases of institutional reinvention. The response developed in this book is that such reinventions, while possible, are vastly less common than institutional adaptation that preserves core function by relocating it — and that the difference matters for how reformers should allocate effort.