The topline numbers (the % supporting whatever) may not tell you much in an absolute sense, but you can still get a lot of useful information from such polls if you do a broader analysis.

First, where question wording has been similar, you can do trend analysis -- is support going up or down, even if you don't have a good handle on what exactly that support means. Seeing how the numbers move in response to events and PR campaigns is also useful, to get a sense of just how firm or not opinions are (in this case, not very).

Likewise, one can do comparative analysis -- 60% say they support the manned space program. Great, is that a lot or a little for a federal program? IF most other programs get 30%, space looks good; if most others get 90%, space looks bad.

Comparing across different question wordings also can tell you a lot about what moves public opinion, and again, how sensitive it is to the way the problem is framed.

Of course, when you can get individual-level data and do multivariate analysis there's lots of stuff that can be teased out. I keep meaning to crunch a dataset I have from a CBS/NYT poll in early 2003 that asked both about support for the space program, and use of internet dating services. We can determine whether space supporters do indeed need help finding dates.

So, throwing out raw numbers from a single question doesn't tell you much, but one can learn something from larger datasets. Doing some of the trend/comparative analysis I mentioned is on my "articles to write someday" list, and would be a good excuse for my employer to send me to a space science conference someday...