1σ confidence contours for cosmological and nuisance parameters, as in Fig. 13, now for forecasts employing the shift model for internal redshift recalibration (Sect. 2.1) – we thus include contours for the first, third, and fifth scalar shifts, δz_i, and exclude the power spectrum normalization, σ₈, in the figure for increased visibility. We show here the difference δp = p − p_ideal for parameter p relative to the idealized ($n_z^{\mathrm{true}}:\mathrm{do}\mathrm{nothing}$) case for each probe set (or the fiducial truth, i.e. the actual difference in the tomographic means, in the case of δz_i parameters), since the correct shifts δz_i depend upon the (in)coherent form of the redshift bias. Most δz_i forecasts see a reduction in bias and a loss of FoM on the S₈ − Ω_m plane (given by the legend) relative to the ‘do nothing’ model (Fig. 13).