In a recent preprint, Pepper et al. (2002) evaluate the sensitivity of photon-noise limited transit searches for habitable planets. They conclude that these searches have the largest chance of success if they are optimized toward the surveying of M stars. Such an optimisation implies several constraints and trade-offs. The principal one is that M star surveys need to observe at fainter magnitudes than surveys of solar-like stars. The required shift to fainter magnitudes is in principle possible, as M stars require less photometric precision in order to detect transits of their habitable planets. However, relatively higher noise from sky-background and increased crowding at fainter magnitudes requires the employment of optics with adequately small point-spread functions, with consequences on the observable dynamic range. A review of the merit of this strategy is given, first in terms of habitability of M star planets. Though being tidally locked to the central star, these planets are valuable and interesting targets, and can still potentially host life. Second, potential implications for the Eddington mission are discussed, and estimates for the noise sources and stellar crowding are given. A preliminary recommendation is to optimise Eddington for mid-K stars, employing moderately smaller apertures. Such an optimisation should allow Eddington to access both neighbouring specral classes (G and M) as well, and habitable planets with and without tidal locking may be detected.