"Stability" refers to at least two distinct concepts in scientific computing:
- inexactness of floating point arithmetic, backward error analysis, etc
- discretization/operator stability, a necessary part of a well-posed discrete problem
The stability tag claims to be the former, but actually contains more of the latter. The tag was created after several of the defining questions in that category, and really needs to be revised. Should the floating-point tag be clarified and used to catch those currently using "stability" to mean floating point stability, with "stability" revised for discretization/operator stability? Or some other scheme?