Durable form is not the opposite of change but its organised management. Semper locates architectural form within material techniques and inherited practices of making; Wiener demonstrates that systems persist by receiving information about their own performance and modifying behaviour accordingly; Stewart Brand shows that buildings continue to transform through occupation, repair, replacement and layers moving at different speeds; Luhmann describes systemic identity as something reproduced through recurrent operations rather than stored in an immutable substance. These positions converge on a more powerful conception of form: persistence is continually produced. Buildings survive through cleaning, repair, adaptation, replacement and behavioural adjustment; institutions persist because teaching, administration, publication and evaluation recur; archives remain usable because files migrate, metadata is repaired and formats are updated; cities function through continuous maintenance of infrastructures whose deterioration and renewal occur at unequal temporal rhythms. Failure therefore becomes epistemically important. A crack, complaint, contradiction or outage can become feedback if information generated by failure returns strongly enough to modify subsequent organisation. This distinction also complicates resilience: survival alone demonstrates little if continuity is achieved by repeatedly transferring stress toward the same vulnerable bodies, districts or ecosystems. The relevant question is not simply whether a system persists, but through which recurrent labour, material expenditure and uneven distribution of costs it does so. Maintenance thus becomes an epistemological category rather than a secondary technical activity. To understand a form is to identify what reproduces it, what signals its failures, which components are permitted to change, which are protected, and who performs the work that makes apparent stability possible.