The Creepy Details of Facebook's New Graph Search
eff.org
eff.org
This assumes that the advertisers didn't have access to searches like this already. That seems very unlikely.
The only difference is that ordinary people can do it with their friends and no obscure query syntax.
I don't see why Facebook would allow advertisers to build databases of people when they can just force them to go through their own tool and keep people anonymous, like Google does.
Also, this is free, as opposed to advertising tools, which broadens the audience quite a lot.
By the way, you can click through the "create an ad" process quite far on facebook before you have to actually pay for it, so for those curious, why not give it a go and see for yourself what the process looks like? :)
For example: https://dl.dropbox.com/u/46245195/jealousy-machine.jpg
Graph Search is really interesting from a data standpoint, but I don't at all see how it's going to mesh with the realities of how people like their data to be managed.
The more specific the criteria, the more likely the "advertiser" is a spear fisher.
I can think of a couple exploits (Err- links) one might pay to send to "groups" like: All Oracle employees who like sailboats and tennis. All Universal Health Care employees who like Oracle. All apple employees who hate google maps.
The total cost of these 3 "campaigns" might be $30. You may hit these targets with a wider google campaign, but that would cost more and entail greater risk (i.e. people at mcafee who like nmap and Oracle.)
So for a query like "All Oracle employees who like sailboats and tennis", they check to see that the person has shared each of these facts with you, and only shows up if everything checks out. It's possible for you to be friends with someone who is an Oracle employee but they have only shared their employer information with a select group of people which doesn't include you. So they would not show up in that query you make.
The advertiser will presumably never see the actual profiles of the fb users, but they will know too much about the visitor(s) arriving from that Ad.
The post I replied to implied allowing "creepy" levels of targeting for Ads was not such a risk.
"Creepy" levels of targeting is pretty subjective. I don't think its creepy because the other side is essentially oblivious to who they're advertising to. I'd rather see no ads at all, but if I had to put up with ads, I'd rather see ads that advertise topics that I'm interested in rather than something that's completely irrelevant to me.
Of course, that's just my opinion.
This can't be said often enough.
I personally hate Facebook because of Zuckerberg's disregard for Steve Jobs's definition of privacy: to know what you've signed up for.
But there's definitely a lot of utility for those who willingly hand over their personal information. It's a trade-off, and I'm fine with it, as long as people knowingly, willingly accept that.
I understand why it took them this long to come out with something like this. For this to work, you need massive amounts of data, something Facebook has plenty of after receiving a firehose of data for so many years.
Google has failed in social search, because (ironically) they had the search first, but not the data. IMHO this is exactly what they were afraid of and the main reason why they keep pushing on G+ hard.
Most importantly, Graph Search finally "captures the intent" - which made Adwords so good - that FB Ads was badly missing. Expect to see their revenues soaring in the coming years.
You say that as if the competition was a specific event with a discrete ending time, rather than continuously ongoing process that ever ends.
I do understand. People don't evaluate risk correctly. They are fundamentally irrational in all sorts of strange ways. Humans have a strong optimism bias[1] so even if they know something bad might happen to some people they know that will happen to other people, not to them. The benefit, even when it's tiny in comparison, they get right now.
Thanks to the fascinating new neuroscience around cognitive biases you too can go from thinking you'll never understand to despairing for the future of humanity!
At least now someone will probably do a Firesheep-esque [2] "look how creepy you can be" app that will get some attention. And that kind of attention is exactly the antidote to the optimism bias as well as many of the others at play here, it's actually very effective because it plays into other well known biases [3][4]
[1] http://en.wikipedia.org/wiki/Optimism_bias
[2] http://codebutler.com/firesheep/
Maybe I'm going to far applying computer security concepts to real life but this was possible before, so making it easier to access only really changes how educated the users are about over-sharing. I mean, if you wanted to educate people in a "wow that's creepy" way like firesheep did you would build something like this.
If you shared it online and it's linked to your real name then you have to assume anyone can get a hold of it.
In the practical world you make an excellent point about discriminatory hiring, but I feel like this was going to happen anyway so better to open things up now and hopefully scare people into realizing what they are doing when they post private information publicly. The longer the time until this happens, the worse it will be when it does right?
I think my Firesheep comparison was better. Firesheep made something possible easy and the result was not an outbreak of hijacked email and social media accounts although that certainly happened in cases where it wouldn't have otherwise. The result was education through surprise and an actual improvement in online security practices (defaulting to https).
If someone is going to have this level of information more transparency about what there is and what can be done with it is the lesser of two evils.