The next generation of cosmic microwave background (CMB) experiments c
an measure cosmological parameters with unprecedented accuracy-in prin
ciple. To achieve this in practice when faced with such gigantic data
sets, elaborate data analysis methods are needed to make it computatio
nally feasible. An important step in the data pipeline is to make a ma
p, which typically reduces the size of the data set by orders of magni
tude. We compare 10 map-making methods and find that for the Gaussian
case, both the method used by the COBE Differential Microwave Radiomet
er (DMR) team and various forms of Wiener filtering are optimal in the
sense that the map retains all cosmological information that was pres
ent in the time-ordered data (TOD). Specifically, one obtains just as
small error bars on cosmological parameters when estimating them from
the map as one could have obtained by estimating them directly from th
e TOD. The method of simply averaging the observations of each pixel (
for total-power detectors), on the contrary, is found generally to des
troy information, as does the maximum entropy method and most other no
nlinear map-making techniques. Since it is also numerically feasible,
the COBE method is the natural choice for large data sets. Other lossl
ess (e.g., Wiener-filtered) maps can then be computed directly from th
e COBE method map.