Now, isn't that just wonderful? ;)
Really, if that is going to be the case, it is a nice one.
Assume, for one moment, that You do not work professionally with search engine optimization. Assume further, that you are an "ordinary internet user" searching for ...well, widgets.
Conducting four searches within half an hour for the term "widgets", clearly, this is important to you. In your first search, the "superior-home-of-all-things-widget.com" even show up as number 1.
This is very nice, if that site is what you need. If not, well, I'd say that You would go about and conduct another search then. At least I would. And You did, too. Four times.
Now, if SE's were only a tiny little bit intelligent - just a little, that is - they would probably assume, that since you have been searching for that phrase in the immediate past, and then return, it couldn't have been the right #1 they picked for you. Obviously (if they were also a bit service-minded) they would present something else on the #1 place.
Of course, SE's are not very intelligent, they're just databases, and of course databases can't be service-minded per se, but some things can be done with probability and usage tracking, and the thought is nice.
..for an end user, that is. (the "Assume.." part)
Unfortunately i must admit that, even though the thought is nice, and even though this SE apparently employ a few fine thinkers, it's not very credible, as until now, focus has been on database manipulation (the so-called PR) and not on user behavior.
So, basically, this post is just yet another wildcard and speculation, and it shouldn't carry much weight, as i really don't follow the google whereabouts much.
I just tend to think that the world is not always static, and thus serps should not always be static either, but that's probably just wishful thinking.