Back when the 2003 rover mission was going to be part of a sample return program, (yes, I know -- I wrote a bit about that particular reality distortion field here) I remember being amazed when Steve Squires told me that they were seriously considering a return capsule with no parachute. The reason was that they would have to convince the planetary protection officer -- currently the most excellent John Rummel -- that the samples would still be biologically contained even if the parachute failed. Given that constraint, the parachute itself became a luxury. The mission had to be able to survive a hard landing so it might as well plan for it.
After the catastrophic failure of the Genesis re-entry/descent system, (6MB Quicktime video here) it’s a certainty that the Mars sample return capsule, when it gets designed, will definitely have a safe hard landing capacity. The risk isn’t so much the possible martian bugs getting out if containment is breached (though that’s what the public debate will focus on); it’s that earthly contamination might get in and ruin the science. However, I very much doubt MSR will be done without the parachutes, as they once imagined. Genesis is yet another reminder that if you can afford a belt-and-braces approach when trying new tricks in space, it's a worthwhile investment.
One of the original selling points of NASA's Discovery program was that with more, cheaper missions it would be possible to fail and learn from failures in a way that was not possible with infrequent and very expensive missions, and thus belts were not required if braces were available. That made a lot of sense, but the Discovery failure record is a bit more extensive than anyone would have wished.
So far 10 missions have been selected under the Discovery program, eight of which have been selected competitively. (The first two missions, NEAR Shoemaker and Mars Pathfinder, were included in the program ab initio, NEAR as a way of showing that someone other than JPL could play the planetary exploration game under the new rules, and Pathfinder so that JPL didn't get too upset by this competition. I caricature, but not that broadly.) The subsequent missions were selected competitively, and part of the program's brief has been that once a mission is selected, the people who proposed it take most of the responsibility for making sure it happens -- NASA headquarters is meant to keep its hands off, for the most part.
Of these eight missions, five have so far made it into space, including Genesis. Of the other four, one -- Lunar Prospector -- completed its mission pretty much flawlessly. Contour was lost soon after take off, in all likelihood because of a design flaw though possibly due to bad luck. Messenger, which like Contour and NEAR came from the Johns Hopkins Applied Physics Laboratory, made it on to its trajectory to Mercury, but has a long way to go and a hard mission to get through. Stardust has completed most of its mission, but to be a success it needs to get its samples back to the earth. Today, that doesn’t seem such a trivial step, especially since at least some features of the Stardust sample return design were the basis for the Genesis system (though I don’t know if they had any hardware in common). How much of a failure the Genesis mission has been remains to be seen, but I’m not that optimistic; complex bits of tech not designed to crash to earth at 310 kph normally don’t take such treatment in their stride. (Incidentally, though the probe's cargo came from the sun, the probe itself did not go "to the sun and back" as I have just heard them claim twice on the BBC; it stayed quite close to home and allowed parts of the sun to come to it, by means of the solar wind.)
This isn’t to say Discovery is an overall failure. But a decade and well over a billion dollars in, with one complete failure, one serious mishap and one completed mission, it’s hard to see it as a glowing success. The obvious response to the trouble would be to increase the oversight NASA exercises on the missions, which was one of the recommendations of the Contour mishap report. (Another recommendation was our old friend the need for constant telemetry during critical phases of the mission; I don't know if that was an issue this time.) Is it possible to increase oversight without adding to development time and costs? I don't know, but it seems intuitively unlikely. The lesson this year seems to be that the path to success -- as followed by the wonderful rovers -- is paved with budget overruns, best followed by labs with experience at what they're doing, and involves a very failure-sensitive headquarters breathing down your neck. Which is fine, as long as you can be satisfied with flying only a few missions.
That said, if the Kepler discovery mission manages to discover earth-sized planets (really earth-sized, not just closer-in-size-to-earth-than-to-jupiter), as I've long hoped it would, then Discovery will look pretty good, since no other NASA programme would have funded such a mission.