I think Google has an intelligent but slightly twisted sense of humor. I would not be surprised in the least if they designed their search engines(web+image) to mirror a '1 keyword is worth 1000 images' in terms of how much actual traffic they send. That being said...
scottb, there's possibly a side issue here.
JS_Harris has noted in another current thread...
I put my entire image folder in a url directory removal request and blocked Google-image bot from indexing anything in that folder via robots.txt. Note: Googlebot and google-image bots behave differently when you block them from a folder. Googlebot finds urls and indexes them with no descriptions but image bot simply doesn't index them at all.
Result
- Total image serp impressions down 100%
- Overall CTR improved greatly because web searches convert
much better than image searches on my site
- Total clicks from search is also up though I'm dealing with a no new data problem in GWT since the 23rd of Feb.
That last line, and the fact I now see real traffic from web search to the affected pages, tells me images were indeed blocking the page from getting indexed in web searches as a non-image result. The way they have image reporting set up in GWT leads you to see 'related images' impressions as textual web impressions. I explained that in the other thread, if you're interested.
Unless Google fixes this oddity, and ideally re-designs image search again to send more traffic, I see no reason to unblock Google's image bot. I'm actually not vulnerable at all to hotlinking, backlinks from crappy image gathering sites or outright image theft from people scouring image search this way, works for me.
If the web/image ranking 'oddity' is intentional I don't see how it benefits users which leads me to believe it *should* get looked into and *should* get fixed soon enough. If/when it does it would take all of 10 seconds to remove the google image bot block so it's not a huge risk to try out on your site.