This morning, my wife and I were sitting up in bed discussing the best route for her to take to park at St Joseph’s Hospital in Hamilton, the next city over from ours.
We had never communicated before about this location in any way.
Clare tells me the parking lot is at 50 Charlton Ave. I open Google Maps on my Android phone to look it up. I type 5, then 0 then SPACE, and voila, I’m presented with suggestions, the first one being 50 Charlton Ave, Hamilton.
This struck me as way more than a coincidence. I know that Google will correlate a whole bunch of things it knows about you to enhance search, so I searched my email to see if Clare had sent me that address before, which she had not. I had never searched for it before. We have driven past there recently, but we have driven past #50 of any number of streets.
The only explanation I could think of was that Google Voice (OK Google!) had picked the address or other hints such as cross streets out of our recent conversation and had used this information to narrow the search.
From the papers in front of us I had Clare clearly state the address of a medical building at 1960 Appleby Line, very close to us. I went to Maps, entered 1 then 9 then 6 and it immediately completed 1960 Appleby Line. Now it’s possible that there are no other addresses within a 5 km radius of our home that begin with 196 but I’m not entirely convinced.
We tried a couple more with less success, and even the two examples cited don’t work any more, but these two instances have me very close to convinced that recent ambient conversation is being used to augment search results.
As a programmer, I think of the data exchange that would need to take place for this to happen, and I wonder how much of a private conversation could be picked out of the data.
I would be interested in hearing from anyone who can replicate this behaviour.