At my father's funeral, there was a stranger present at the back of the chapel. As soon as I walked in, he approached me and immediately identified me as a member of the family, saying I looked like someone with our surname, which is fairly uncommon, ranking near the 14,000 mark for total population with that name in the 1990 census. It turned out he was doing family-tree research on our shared surname, and in his research he had encountered the same experience in a cemetery in an area where our shared name is more common locally; being recognized as a member of the named family simply by his physical characteristics.

As of consequence of this experience, DNA surname prediction seems perfectly logical to me.

Well, for this to work, they would need chromosomal data from a great majority of OTHER people with that surname. Obviously, it wouldn't work very well for common names, or those that are spread out more.

For more on the Y chromosome and surnames, I highly suggest Adam's Curse by Adam Sykes.

I agree with Yao Jun.

I think this is an interesting finding, but it might not work as well for those with common names - especially if, as in my country, some of us did not inherit those names at all.

At one point in our history, our colonizers decreed that we were all to have Hispanic names, regardless of our bloodlines. So I don't think this method would apply to, say, an Alvarez from the Philippines and an Alvarez from Mexico or Spain. Within the Philippines itself, there are so many dela Rosas, dela Cruzes, and Santoses who are completely unrelated.

But then, I don't know when, if ever, this kind of technology will be put to good use in the Philippines.