“So the image search system might come back and say, oh… Image results are, hmm… kind of important for this query and here’s a set of images that I have for that.
And if we get the web results back and the web results say, oh, web results are super relevant for this query, then we might show the web results and not the image results.
On the other hand, if we hear back from these different systems and they say, oh, image results are super important for this query and the web results are kind of okay as well, then we will almost certainly show kind of an image one-box somewhere in the search results.
It might be on top if we think that’s super critical, it might be somewhere in the middle, anything like that.
And the same thing happens for video and I imagine for the different other types of one-boxes and things that we have as well… I don’t know, maybe Top Stories, I don’t know how all of that flows in.
But essentially all of these different systems come back to that one central place and say, here are my results and here’s how relevant I think they are.
And based on that we try to figure out which of these elements we should show.
So that’s kind of how the images flow in there. It’s not that someone is manually saying, oh we should show images for this query and not for that one.
It’s really these systems are trying to understand how relevant the results are that they can bring back and based on that we try to make an automated decision.
And that’s also why this can change over time, where if you search for something now, then maybe we won’t show images. And if we recognize that everyone is actually looking for images for this query, then over time we will start showing images in the web search results as well.”
