Bassically the service mission that has been proposed is, excuse my French, so much crappola. The mission isn't coming until 2008. There is a better than 50% chance that the Hubble will undergo enough failures to require that NASA deorbit it before the service mission can even be mounted. In 2004 the STIS failed. It was the last spectrograph on the Hubble, and the only one that ever took advantage of its supperior spatial resolution (which is, well duh, the reason for the Space Telescope to start with). The current mission as I've seen it doesn't adress the STIS's failure or contemplate repairing it.
That left the ACS camera the main workhorse of the HST. Last month, the ACS had a glitch that gave every engineer at NASA a heart attack. If the ACS dies (a reality that NASA is making plans for finnally) then HST is essentially a giant paper weight in space. To be fair, the fine guidance sensors are an amazing instrument. The only problem with them is that the number of astronomers that know how to use them can be counted on both my hands without using my toes. The WFPC2 is a bust. It's not it's own fault. It is just the typical classic failure of an instrument designed by a committee of ten thousand. The filter sets are no where as good as for the ACS and its resolution and field of view are also worse than the ACS. People will only use it if it is the only game in town.
COS is the new spectrograph they're touting. But it doesn't have any of the spatial resolution to do the things that the Hubble does best. It's basically for wide aperture spectroscopy of galaxys.... hey wasn't that what GALEX and FUSE were for? If yall haven't heard of GALEX and FUSE it's because they've bassically been a huge scientific bust. There has been some good ISM work to come out of FUSE, but that work is almost accidental because it was done by people piggy-backing on the main mission.... which was a bust. They and COS are poorly designed spectrographs that don't address the real issues in stellar and galactic astronomy which have classically driven ultraviolet spectroscopy. COS actually takes the clear images of the HST and smears them out on the imaging plane on purpose. That's like putting a pair of smoked glass Harry Potter glasses on the space telescope. Not a good plan if you ask me, especially when a previous instrument, STIS, almost got it right.
So my reaction to the announcement was "well la-dee-da!" because it is a case of too little too late. Hubble has become part of the huge problem with Astronomy in the US: we've got all our eggs in just a few baskets. When those baskets go bust we all go down with the the ship. We shouldn't continue putting all our money in huge projects, particularly huge projects that don't benifit the entire community (see Next Generation Space Telescope). The extent to which this whole shebang has become a political patronage football also turns my stomach. We don't build scientific instruments for real scientific reasons anymore, they're just a sound bite for the latest glory hound to call himself NASA administrator and pork for the right congressional districts.
Solider on brave space-cadets.