The redshift distribution of gamma-ray bursts (GRBs) is strongly biased by selection effects. We investigate, via Monte Carlo simulations, one possible selection effect that may be modifying the Swift GRB redshift distribution. We show how telescope
response times to acquire a GRB redshift may, via the Malmquist effect and GRB optical afterglow brightness distribution, introduce a bias into the average of the observed redshift distribution. It is difficult to reconcile a recently reported correlated trend between telescope response time and average redshifts unless we employ a redshift-dependent optical afterglow distribution. Simulations of this selection effect suggest that GRB optical afterglows may have been either intrinsically brighter early in the Universe or suffered less local host galaxy extinction.
We compare the detection rates and redshift distributions of low-luminosity (LL) GRBs localized by Swift with those expected to be observed by the new generation satellite detectors on GLAST (now Fermi) and, in future, EXIST. Although the GLAST burst
telescope will be less sensitive than Swifts in the 15--150 keV band, its large field-of-view implies that it will double Swifts detection rate of LL bursts. We show that Swift, GLAST and EXIST should detect about 1, 2 & 30 LL GRBs, respectively, over a 5-year operational period. The burst telescope on EXIST should detect LL GRBs at a rate of more than an order of magnitude greater than that of Swifts BAT. We show that the detection horizon for LL GRBs will be extended from $z simeq 0.4$ for Swift to $z simeq 1.1$ in the EXIST era. Also, the contribution of LL bursts to the observed GRB redshift distribution will contribute to an identifiable feature in the distribution at $z simeq 1$.
