A transition-edge-sensor (TES) microcalorimeter's shunt resistor (R sh ) and thermal conductance to the cryogenic bath (G) are often considered to be interchangeable knobs with which to control detector speed. Indeed, for otherwiseidentical TES-parameter models, there are many combinations of R sh and G that give the same decay time-constant (τ crit ). However, our previous work showed that with time-or code-division-multiplexed readout, the distribution of signal-to-noise ratio with frequency, which depends strongly on R sh and G, is just as important as τ crit . Here, we present a set of calculations to select the optimal values of R sh and G, given a linear TES model and count-rate and energy-resolution requirements. Lower G and lower R sh make multiplexing easier. Our calculations also determine the allowed combination of SQUID-readout noise (S Φ ) and multiplexer row-period (t row ) and row-count (N rows ). Recent improvements to S Φ and t row in the NIST time-divisionmultiplexing architecture have allowed a NIST eight-pixel TES array to be read out with 2.70 eV (full-width at half-maximum) average energy resolution at 6 keV. The improvements make the X-ray Microcalorimeter Spectrometer co-proposed by NASA and NIST for ESA's Athena X-ray observatory straightforwardly achievable, including engineering margin, with N rows = 16.