[ExI] google cousins. was: RE: online resources for identifying symbols

BillK pharos at gmail.com
Thu Dec 12 17:44:45 UTC 2013


On Thu, Dec 12, 2013 at 4:50 PM, spike wrote:
> I am looking at a couple things.  I noticed there isn’t much correlation I can tell
> between my Google cousins, although I think the first row first image is the
> same guy in the third row second image.  Can you tell?
>

If you click on the link for the image, you may find both photos come
from the same web site. Or there may be attached info on the photos to
identify them.

>
> I want to see if a woman gets mostly women Google cousins for instance.
> Also, if I notice there is only one image that is pi()/4 profile, the woman on the
> third row fourth image, and with my bad luck, she would be the stalker, oy, eeesh.
>
> I see only one guy who looks even a little like me, fourth row fourth image, but
> oddly enough, that boy to his right looks a lot like I did when I was about 8 yrs old.
>
>

The way the Google image analyser identifies features, weighs them,
and selects photos is probably undergoing constant change. As you
note, a white suit image produces mostly white suited photos.
Similarly I would expect a female photo search to match mostly female
photos.

As a matter of interest, when I do a search with your image, I get a
different set of matches, mostly male but including one hot blonde
female face. Even on image search Google changes the search results
depending on who submits the query and what it knows about them.

I think that Google doesn't really know what it is matching, unless it
finds an exact match.

BillK




More information about the extropy-chat mailing list