Why it is a question
THEORY-383 reports one testbed (LIT-1118, read in NOTE-908): on Apache and Informix with cheap indexed selections, a query result stored in the DBMS and refreshed on every update lost to recomputation at five or more updates per second, which the authors explain by reading the stored result costing about as much as recomputing it and its refresh loading the same DBMS. Whether that explanation picks out the condition under which a stored result loses, on other systems and other query costs, is open.
What would answer it
A measurement on another database and web stack, with native materialized views and queries of graded cost, showing that a stored query result kept current on every update loses to recomputation exactly when reading the stored result costs about as much as recomputing it and the refresh runs on the resource that serves requests, and wins otherwise. More runs on one testbed with cheap indexed selections cannot settle it, since they only repeat the case held; nor can the paper’s Eq. 9, whose values were never computed against measurement.