Current science policy regarding the amount of information that should be archived regarding remotely sensed data is not adequate for producing long term climate data sets. It is essential to be able to discriminate differences in data caused by artifacts of the measurement process from differences caused by changes in the underling geophysical parameter of interest. Deriving geophysical parameters from remotely sensed data requires instrument measurements, calibration, and conversion algorithms. Information required to understand and replicate the entire measurement process is not currently archived along with the resulting data sets. Developing current data sets which will be the basis for ongoing climate studies often requires over 100 man years of science effort when done with full access to necessary instrument and processing system details. Papers describing these efforts currently published in the scientific literature do not provide sufficient information to enable anyone working independently to exactly replicate the measurement or analysis systems. This has not been an overwhelming problem to date. In most cases when there is a question it is still possible to go back to the people who made the measurement or did the analysis and ask “What did you do and how did you do it?” However some of the current data set efforts started as much as thirty years ago and none of the participants will be available to answer questions fifty years from now. There are two major impacts of this lack of information. First, it will not be possible to determine whether changes in the derived geophysical measurement between existing and future systems result from changes in the analysis method rather than changes in the underlying parameter. Second, time and money will be lost since it will not be practical to benefit from the preceding work by modifying or reusing the previous analysis code and techniques or to compare both new and old approaches for the same data. The concern is not that there will be blunders in the current or next data sets (though that may be the case) but rather that the two sets are different because of approach rather than underling change in the parameter of interest. There are things that could be done now to reduce this problem but they are not free and in a few years it will be impossible to do them for existing data if the effort is not started soon. This issue will impact both people producing climate data sets and those in the future who will use them. My hope is that this presentation will help to start consideration of this problem.