Must've never heard of caching.. http://www.latimes.com/news/nationworld/nation/la-100301safe.story
Inevitable next step: Enterprising cypherpunk registers censoredfedinfo.org, hunts through google's cache, posts everything there, etc.
Note that there are a relatively small number of Googles on the Net.
The trouble with Google and most other spiders is that they cannot access the DBs behind the sites. Various industry estimates place the amount of data not accessible to crawlers at up to 500x the html content. What's needed are open access data mining sites using more sophisticated crawlers like http://telegraph.cs.berkeley.edu/
More to the point, most spiders don't cache images, only text, so much of the interesting content isn't cached. I'm not sure how many of them cache PDFs; some of those are searchable and indexable, while some are just bitmaps. On the other hand, the Feds generally don't have as much fancy-graphics-design-for-inaccessibility, so more of their text may be cacheable than typical business sites. Shutting down web sites with data that terrorists could use has been going on for a few years - apparently many of the haz-mat sites are no longer accessible to the public, including one of the Bay Area sites that shut down a few weeks before we had a major refinery fire. Yes, there are potential threats to public safety if terrorists can use this data, but there are more serious threats if the public can't use it to determine what's near them, and far more serious threats if fire departments can't access the data conveniently.