Not that interesting to be honest
I mean, it's not really a hack or even pilfering, is it? It's just using the DOM for one of the things it was meant for - inspecting the styles of page elements.
Start Panic is reliant on the script knowing the specific URIs it wants to check and second, it can't determine *what* you did on each site it detects. For example, it can detect that you've been to http://www.google.com/ but it can't detect that you've been to http://www.google.com/search?q=something+naughty unless it specifically checks for that precise term. Given how many variations there can be on Google search URIs that give the same results, and that you'd then have to check the history for all those possible URI variations and then do the same for each TLD Google operates under, it'd take forever.
To be even relatively sure that you'd scanned the user's history accurately for even one search term, you'd probably have to probe the history hundreds if not thousands of times. It's hardly worth the effort, is it?
Come back when someone comes up with JS code to automatically extract all visits to http://www.google.com/* from the history. That will be news but of course, it will never happen (well, maybe in IE...)