But I think your post is in error because it implies that the raw data is not eventually made public. To my knowledge, this is not true -- especially for Earth missions, I think all that data is public. Perhaps the level 0 data [essentially, raw telemetry] is not, but everything from level 1 up will be public.
The data-release culture is very discipline specific. In planetary (e.g. Mars), there are embargo periods -- partly to ensure the systematics are understood. In cosmology, a four-year embargo period for Planck was ended just this summer. This kind of makes sense -- they didn't want half the astrophysicists in the world off chasing a phantom caused by their very sensitive noise-suppression tools. But I understand the community was getting pretty impatient. So, there is push-back.
For Earth missions, after an initial calibration period, I think all the data is made public as fast as it can be processed. Indeed, MODIS even broadcasts the raw data to whomever wants to pick it up (http://modis.gsfc.nasa.gov/data/directbrod/), like disaster responders.
Like I said, it's discipline-specific. For disciplines where time matters (Earth, space weather, seismicity/geodesy, time-domain astronomy) there is a real advantage in releasing data fast, and that's what has happened. I could give websites in each case where data is released in near-real time, say, less than 1 hour after taken.