Why? (I feel the opposite, as what better system to get the data than the system it was designed for? I'd like to hear your opinion.) -- Almost all of the data that I recover is done at the physical media level. This is done either with analog to digital converters or modern digital data separators which have better recovery characteristics than the originals. The main reason for wanting to do this is to preserve all of the original bit streams and error checking information. In the case of magentic tape, it is possible to recover data from very old tapes using modern magnetorestrictive head technology that would be impossible to recover using the original heads, since MR heads require MUCH less contact pressure and are much more sensitive than the originals. While it may be practical to maintain microcomputers for this purpose for newer media, the types of data I have been trying to recover (mainframes and older minis) is impractical, either because the machines no longer exist, cannot be kept running given the tradeoff of machine ontime vs the time needed to keep it running, or that the reliablity of the data from older controllers is worse than can be obtained from direct low-level data aquisition. There is also the problem of file transation and transfer even if you can get the bits read on the original system. The problem with using these techniques is it requires detailed knowledge of how the data was written and information on things like file formats. This is why there is a strong bias towards this sort of information on bitsavers (in addition to the fact that I've discovered this information is REALLY hard to find for pre-minicomputer systems). This is also information that is VERY useful to people trying to write simulators.