I do wonder if this wouldn't be preventable with a better-designed web. Why fetch static media from the source anyway? You could grab it from your peers.
There's plenty of interest in replacing transfer and naming, but I don't see a lot in developing a post-WWW hypertext system. Are there any projects or did the WWW and HTTP kill off research in that area?
I suspect in this case, the fact that the static page links to a dynamic one on the same server, right up top, and urges you to go there helps not at all.