• Astrometric calibration of ALL the world's images and all the world's digital cameras
• Dependence of camera calibration on temperature, focus, gravitational loading of all the world's astronomical imagers
• "Your image is located here on this part of the sky. If you like this part of the sky, you may also like this other part of the sky."
• "Your image is located here on this part of the sky. Dan Stern also imaged this part of the sky. Do you want to see his images?"
• "Your image is located here on this part of the sky. Here are direct links to the calibrated HST/Galex/Spitzer images that overlap your image."
• Provide approximate photometric calibration based on standard calibration stars.
• Determine the proper motion of every source on the sky
• Create an extremely heterogeneous but extremely dense astrometric standards catalog (AN-1.0)
• Provide not just WCS but multiple realizations of a WCS that are all within some specified likelihood ratio of the optimal.
• Send email to users saying "on such and such a date, we gave you WCS on an image. Because we received new data on this part of the sky or we received new information on the instrument used to capture that image, we have computed a better WCS estimate based on this new data. You can retreve the improved images with this wget command"
• Group images by both spatial clustering and by bandpass, and automatically produce combined images
• Make amazing composite image of all relevant interesting information indicated (interesting objects, galaxies, redshifts, etc)
• Facilitate the followup of transient, real-time, events (e.g. gamma ray bursts, supernova, gravitational lens transits)
• Continuously assemble all images into a multi-wavelength time-dependent map of the sky
• Interface effectively with the low-bandwidth, low-power, low-disk  OLPC
• Determine the parallax of every source on the sky
• Reprocess the Hipparcos/USNOB data
• Confirm promising/interesting proper motion and parallax stars using HST FGS
• Ultimate list of open questions for ANY object, or maybe ALL objects. (AN --> knowledgebase)
• Locate the worst places in our astrometric catalog (i.e. lowest quality data) and apply for telescope time in these horrendously imaged parts of the sky (NOTE this is TRIVIAL if the rest of astrometry is done, i.e. feed data into AN). We could probably get money for this easily.
• Expose an interface/API to the astrometry data such that automated searches for statistical anomalies can be written by 3rd parties
• Write an amazing visualizer similar to google maps (perhaps using the kdtree) which is graphical and pretty, which lets the user zoom around and inspect the data on a whim, in a way that is FUN, and encourages further investigation.
• Write an interface to the browsing system (be it graphical or on the web) such that astronomers with a pedagogical bent can write meaningful, informative, wikipedia- or tutoria-style entires about these objects to help educate the interested public. (tagging)
• Write a scientific american article (or other popsci rag)
• Interoperability with the future
• A kdtree database of triangles where we represent every field ever taken such that given a ray from the earth to some RA/Dec, we know exactly what fields this intersects
• (re: above) use case: find all HST images intersecting with this set of 60 000 rays, and return the exact pixel in each image which intersects that ray.
• Become the data archive for all telescopes on the entire earth. Back of the envelope calculations show that at current rates, that is approximately 30TB/year, which is cheap enough that a single wealthy individual could afford it. Considering the amount of money spent on telescopes (100's of millions of dollars), it seems reasonable that the large telescopes would divert 0.1% of their budget to AN for managing their data forever.

YOUR IDEAS HERE