Ask HN: Is Google cataloging my girlfriends?

[wonderwheelhuh]

I recently did a search on Google using the "Wonder Wheel", which lists related queries as spokes off of your original query. When I searched with my girlfriend's name, one of the spokes was my own name. Not a problem here, as many search results point back and forth between the two of us (blog, twitter, etc).

However, another spoke off of my girlfriend's name was surprisingly my ex-girlfriend's name. Now, of course, this might make sense if they had any related search results, but they don't. There's absolutely nothing on the web that directly links one to the other or contains both of their names. Moreover, there's nothing on the web that links me to my ex, so it can't even be an indirect connection of that sort through me.

Point is, I'm concerned as to how Google came to link the two. Since my current girlfriend and ex have never, to the best of my knowledge communicated online in any way (publicly or privately), the only explanation I can come up with is that Google has taken note that my girlfriend and ex may have done Google searches on each other. Has an invasion of privacy gone on here? Can you think of any other possibilities?

UPDATE: If we don't come up with a reasonable explanation, how should I take this up with Google?

FURTHER UPDATE: Anyone interested in trying the search from themselves can leave contact info for me somewhere. As long as you seem trustworthy and will not share them with others, I'll send you an email with our names so you can confirm/investigate.

[marltod]

Clearly this is an automated process. No one at Google has done research to determine who you have dated. It is possible that your parent, or friend or neighbor has done searches about both of them.

For example, an old friend may be curious if you are married so they search for you and your old GF names to see if you have a wedding website. Then by talking to a mutual friend at a later date they get your new girlfriends name and do a search for your new girlfriend.

And that is all it takes to make the connection.

[wonderwheelhuh]

If I understand what you're saying then despite the decent amount of info about each of us online (I'm a technical author, my girlfriend is a well-known blogger, and my ex does freelance writing for numerous periodicals), the relatively low-volume and infrequent queries that some people are making and that produce no search results are allowing Google to establish the connection?

Yes, the title on this post was a bit facetious, I don't think Google is actually cataloging my girlfriends, but I still don't buy your explanation. There is so much more info out there on each of us, and so many easier and better connections to make, that I am left to suspect that Google is not using the method you described to connect us because it is weaker than other possible connections.

Worse, even if I did accept infrequent GET requests to a search engine as an explanation for Google's connecting the names on the Wonder Wheel, I don't take it as a justification of that act. If a search engine is okay with drawing and publishing connections from just a few searches made, than wow is that a loop hole for fairly private information (especially about lesser-known people) to get out.

[orblivion]

That would mean Google is publicly (albeit indirectly) exposing his family members' searches.

[brazzy]

Could it be as simple as Google knowing that someone (namely, you) has previously searched for both these names, thereby establishing a connection between them?

I'm not sure whether that should be considered a privacy violation,

[greyman]

Yes, in my opinion, that would be a privacy violation.

[brazzy]

Why exactly, though? And can you suggest a practical solution that does not involve banning query correlation entirely (I don't think anyone is opposed to the idea of showing "Fred Astaire" and "Ginger Rogers" as related because many people search for both of them)

[ErrantX]

wonderwheelhuh emailed me the names and, just to confirm, I get the same results (so it doesn't appear to be a hold over from his Google settings/history).

One of the names (the ex's) is the same as a reasonably high profile writer (2K google hits) on the internet so there is a possibility that it is something she visited (perhaps Google reader?) has triggered the link. Edit: it seems his ex is that writer. So the Google reader etc. idea is a red herring :)

The link is one way (it seems) from the current gf to the ex.

(I know that reiterate as lot of what he already said..)

But it does appear somehow Google has linked his ex's name to his gf by some algorithm/process.

[dkarl]

Perhaps they were linked by one of them obsessively googling the other. I've received bitter comments about a current girlfriend's appearance, occupation, and religion from an ex with whom I shared none of that information. I did a Google search and came up with exactly the information my ex had. The same ex had, when she was my girlfriend, researched my previous ex and obsessed about her appearance, occupation, etc.

[Tichy]

Maybe they met in secret to conspire against you.

[wonderwheelhuh]

Ha, would that also imply Google even-more-secretly spying on them?

[jrockway]

One of the spokes off a wonder wheel for my name is "jonathan rockway fired". Does Google know something I don't!?

[ErrantX]

I'm even more concerned. Mine has "tom morton beatcroft" as a spoke....

Don't even know what that means!

[ABrandt]

And mine has "austin brandt god". Slightly more flattering, yet equally as disturbing.

[nostrademons]

Ironically, the top result for that query is this comment. How wonderfully recursive.

[wonderwheelhuh]

I'd prefer not to have to share their names, but honestly, if it means bringing a privacy violation to our attention, I will.

But yeah, for now, please see if you can come up with any other explanations.

[z92]

Even if you decide to post the actual names, do it slowly. Remember that once you post the names here, all search results that you are talking about will get destroyed by this information now being available on a single page.

[ErrantX]

> the only explanation I can come up with is that Google has taken note that my girlfriend and ex may have done Google searches on each other

That seems.... unlikely. Mostly for logistical/practical reasons than anything else.

Were you logged in at the time? That seems more likely.

Did the name pull up actual results on your ex? Is there a more reasonable explanation (i.e. could it be someone else your GF is interested in).

[wonderwheelhuh]

> Were you logged in at the time? That seems more likely.

I am not.

> Did the name pull up actual results on your ex?

It does not.

> Is there a more reasonable explanation (i.e. could it be someone else your GF is interested in).

Of course some other connection between my girlfriend and ex may be possible, but none exist or are record online to the best of my knowledge. The problem with this answer is that all of us have considerably more readily associated information, and even if there is a rare case of indirect connection, it seems unreasonable that Google would weight it so greatly so as to make it one of the 8 Wonder Wheel connections.

[jcdreads]

Is it possible that each of them has searched both for her own name and for yours?

[dutchbrit]

One question, do you use GMail?

[wonderwheelhuh]

Yes, my girlfriend and I use GMail.

I am unaware of whether my ex uses GMail or not. We were in school when we met and had uni email addresses.

[wonderwheelhuh]

I just did a search and found that my ex does now use GMail.

I also searched through the contacts and conversations for both my GMail account and my girlfriend's (she consented, of course) - result: in neither did my ex show up.

[samwillis]

Are you logged in to your google account while searching?

I have found that the results change when you are and it will promote things that you have search for and emailed about before.

[wonderwheelhuh]

Unfortunately not - I am not logged in when I make the query.

[terminus]

Could you try the same query after deleting all the google cookies?

[wizard_2]

The wonder wheel search is useless for my name... maybe it doesn't have a full index.

You might have overlooked that your ex could have posed information about you online. Google probably figured out that you were related to her in some way and it's why she shows up as a "personable" (and similar way you're linked to your current gf) link to you.

[nostrademons]

I worked on Wonder Wheel, though on the frontend and not the backend. Drop me an e-mail with the queries. I don't know enough about the backend algorithms to guess what may be causing this (nor do I know if I could reveal them if I did), but I may be able to route this to someone who has a better idea than I do.

[nemoniac]

Look up "Latent Semantic Indexing".

That should lead you to an explanation that'll ease your paranoia.

[DrSprout]

I'm not sure if it's more or less sinister than Google tracking your searches, but it's also possible it's noting your most frequent email correspondence.

Maybe it takes into account words that show up frequently in your email?

[wonderwheelhuh]

Some more info:

1) When I do the search described, I do it without being logged into any Google account.

2) To the absolute best of my knowledge, no email correspondences to my ex or about my ex have ever come out of my GMail account or my girlfriends.

I would consider email tracking more sinister for sure, so in some ways I'm glad it's not that.

That said, it's definitely not the volume of searches people are doing on anyone's names that have created the association (we all have a fair amount of individual web presence). So again, my only hunch is that Google has taken note that my girlfriend while logged in has searched for my ex (I have this confirmed) and that my ex likewise may have done the same.

If that is the case, I definitely 'feel' privacy violation going on since they're not using queries on aggregate to craft results, but instead are considering low-volume queries between users as a sign of association.

[ErrantX]

That seems a little unlikely - simply because, as you say, there is no link between the two girls. So why has Google made that link (when I am sure a large number of people with the same name exist).

What it could be is your girlfriend Googled her name while logged in (i.e. rather than matching the two it just recognizes it as something she searched for). Even that seems unlikely - how does Google know you mean that person when Googling her - does she even have her name registered with Google. and so forth.

There are so many variables I really think that, using Occams razor, it is a highly unlikely solution :)

Also; Wonder Wheel has always given very slap dash to me. I've not observed this level of sophistication in it before...

Would you consider giving the name(s) to someone you trust (but not nearby) for them to perform the same search. That might give us some better data to figure it out.

EDIT: email is on my profile if you want me to check.

[tokenadult]

Wonder wheel on my name shows my professional interests, and doesn't show any names of women at all. (My wife of twenty-six-plus years has little Internet presence.)

[jmakov]

Seems they finally copied http://searchpoint.ijs.si/Result.aspx

[yanw]

If a third-party (you) searched for your current girlfriend and then for your ex (or vice versa) that could make them related queries. Click on the link at the end of the spoke and follow the six degrees of separation.

[wonderwheelhuh]

Ah, interesting possibility, but I don't think it could have been I who catalyzed it.

My concern first arose when I searched for my girlfriend and surprisingly spotted my ex as a spoke. The only other search I had done was for myself.