There are so many tunnel excavations, and so many who use either 'observational' (maybe double-shell) NATM or single-shell B + S(fr) selection from Q, or RMR, or less frequently RMi. Many then ask their younger (?) numerical modellers 'in the office' to check for 'plastic zone' size (often using the RocScience Hoek-Brown, GSI, Phase 2 / RS 2 software). We can alternatively use UDEC-MC or UDEC-BB to check for bolt and shotcrete loading. There are no 'plastic zones', just modelled joint adjustments: some joint shearing, joint aperture changes, and deformations. An altogether clearer idea of rock mechanics mechanisms. We can make improvements in bolt spacing or length or shotcrete thickness/strength if needed. Unfortunately it has been known for a long time that one should not add 'c' and sigma n tan(phi), either in linear Mohr-Coulomb or non-linear Hoek-Brown. This applies to the FEM continuum models. The nested equations involving a crudely estimated GSI (from a diagram, not from characterization) and the following opaque equations of c and phi need software as they are so complex. There is also a disturbance number D giving great freedom for the result, and a deformation modulus equation without depth or stress adjustment. Unfortunately the 'plastic zones' that the young modellers produce, though very colourful, may be grossly exaggerated, as found in a formal court case concerning a stable tunnel excavated without the need for shotcrete support. Is it not time to re-think the numerical modelling we are often doing in rock engineering? It is often based on so many a priori assumptions with even the input needing software. Mostly, the empirical methods, though with some inevitable weakness like subjectivity, are at least based on a posteriori principles.