It would then scan all sites 10'000'000 etc then return the results to dipsie, without us knowing about it.
Thats impossible. Any robot which obey's robots.txt and has a bot [dipsie.com] page, is of course, a real spider - which has a user-agent, IP etc. and when a bot visits your website, it leaves behind its IP, reffered URL, User-Agent etc. Its impossible for a robot which has a UA to visit your website and leave behind nothing. Thats one of the reasons I think Dipsie is a scam.
It promises to provide a HUGE index of 11 billion pages when it launches - and guess when its launching date is set to? 2004, which if, I'm not wrong, is this year. So for sure, if a search engine is promising 11 billion webpages its gotta be seen somewhere!