Google+ Circles: And The Inverted Personal Privacy Dilemma

What circles do you think you have been added to?

tldr:
Google+ introduces a new type of privacy problems which I can only describe as the inverted personal privacy dilemma. It is no longer an issue of “What I choose to share with Google and then what Google shows to others”, the issue is now: “What friends/complete strangers choose to share about me to Google, and my inability to do anything about it”. The responsibility of maintaining my personal privacy has been partially removed from my hands, and placed in the hands of the people of internet.

Thought experiment:
Put 100 privacy aware/concerned people into a room. What would you suppose is the best way to create a global interest graph/taxonomy for this group of individuals? Algorithmically? Manually? First, ask them to fill out very basic personal information about themselves. We will probably have reasonable success. Next we ask them to fill in very specific personal information/categorizations about themselves. How do they react? Probably not as successful. Now ask the same group of people to write down and categorize the people around them, but remind them that their categorizations will remain secret. Each individual’s privacy has been maintained, and collectively we have created a very rich interest graph/taxonomy which is possibly far more accurate than any automated solution.

Doing this algorithmically is hard* [full stop]. The process of taking a very soft analog characteristic, and turning it into a digital signal can be very difficult to get perfectly. Humans are very good at doing this. This problem situation seems very familiar:

Before people-tagging came out, I think most people would have said that the best way to figure out who’s in photos was to have some face-recognition algorithm. But it actually turns out that the best way is to just have people tagged.
-Mark Zuckerberg

Background:
As Google throws it’s hat into the social ring once again, we have all been treated to a very interesting and exciting new alternative on the web to social networking. I’ve read many very interesting opinions on the new product, some heavily praising, some really bashing it. I’ve resisted my natural urge to do an overview of the UI/general UX. I will not be doing a product or feature review. I will however look at some interesting privacy implications some of these new features/concepts this product has introduced.

Google’s famous mission statement: “organize the world’s information and make it universally accessible and useful” has lead to some of the most brilliant and amazing products I have ever used. Using my own mission statements, I see Google Search (organize the web) of course topping the list, their amazing Maps product (organize the physical world), GMail (organize my digital mail), Google Images (organize the web’s images) etc all as fantastic products. It became clear in the late 2000’s and now into 2011, that the social web and social networking was the next necessary territory grab online. Staying inline with Google’s motto, I personally see Google+ as: organize my social network, or perhaps organize the world’s content with social context for me, or maybe even: organize the people in the world.

The Dilemma:
To me, Google may have accidentally introduced a new type of privacy issue, something I call the “inverted personal privacy dilemma”. I call it a dilemma because while I feel like my privacy is safe from others, I am essentially exposing information about other people without their knowledge. And I use the word inverted because to me, it seems opposite to all of the traditional privacy problems that exist on the web.

This is also a dilemma because I personally don’t like seeing myself comparable to an indexable book, nor do I feel like I am a place on a map. I’d hope that there is a human element that lets me be who I want to be, beyond any indexable attributes or properties. There is already a pretty large index of people on the web, it is called Facebook. I use it, and I provide as much information about myself that I feel comfortable with, and this is what it looks like for me:

The information here was provided by me, and is essentially the most amount of information I’ve ever really been comfortable disclosing with people online. It is the sufficient snapshot of my life which made it easiest for my old friends/coworkers to find me if they chose to look. I personally don’t add anything else, although Facebook gives you the option to do so, and many do willingly. I don’t add music, movies, religion, favourite book, etc, although again, many people choose to do. And that is fine.

Circles:
Interestingly enough, in addition to the basic personal information I’ve added, the Google+ circles metaphor allows a very similar view of my networks. My circles are explicitly created/labeled, private views of my social graph. This data certainly says something about me and my network from my perspective:

Right now, I am notified when someone has added me to a circle, and I can choose to ignore it, or maybe add them to one of my circles. Fairly harmless. From the perspective of my Google+ account, this is a visualization of most of what Google knows about me based on what I have explicitly done on the site (circles/bio etc).

I have added my location to my profile, my alma mater and that I have previously lived in Toronto. Additionally, I have created circles for those properties which further re-enforces the data graph that these are true characteristics of the person Ian Chan. I have chosen explicitly to share this subset of information with the web, not too different than the minimal amount of info I have shared on my Facebook account.

Inverting:
Now let’s invert this graph, or rather, lets visualize how other people on the internet have added me to their circles and what the implications are. It’s worth noting, a large portion of people who have added me to their circles are complete strangers:

Now due to the existing privacy constraints, I cannot actually see which circles I have been added to. This satisfies most people’s concerns about the circles metaphor. But it is easy to speculate how people may curate their graph. If one of the purposes of circles is to allow topic based feeds of data, and narrow cast content to these circles, the above graphic is fairly reasonable for me. At the bottom, I display the expected: Toronto, SF, UofT groups which compliment the same circles I have already created. Given the bi-directional (explicit outbound + implicit inbound) taxonomy group agreement, it is a fairly strong/confident signal about my personal characteristics. The more people who add me to their SF circle, likely the stronger that signal becomes. Javascript, UX, startups is another group of circles I can imagine myself being placed in. Since I often talk about these topics, and am active in each respective community, it would make sense for the industry connections I make to slot me into these bins.

The last property, Poker, is something that gets a bit more interesting. Yes, I know a lot of people through my hobby playing poker, and it is not unrealistic for one of my friends/acquaintances to add me to a poker circle. The action of them doing so creates a new edge on my implicit interest graph for which a data point now connects me with the  “poker” taxonomy entity. The more people who add me to a “Poker” (or similar) circle, the stronger that graph edge becomes in the global taxonomy. Now, I am pretty open about my poker hobby, but I wouldn’t explicitly add that as a hobby on Facebook;  Mostly because I don’t need advertisers knowing that “gambling” is an enjoyable pastime of mine. As I mentioned before, my Facebook interests are basically empty, and as a result, I rarely get ads targeted to me beyond the generic targeting for a “male in SF”.

Implications:
Advertising and control. The prices for targeted, contextual ads are significantly higher than non-targeted ads. Facebook learned this a long time ago, especially in the social context. From the Facebook advertisement documentation:

I like this (Facebook’s) model. It is within my control. Poker is obviously just an example here. I would not be devastated if I were targeted for a poker ad, but I might be a bit uncomfortable with it. But there are a lot of other characteristics about myself that I would not want to share with an ad targeting system. My religion, my political leanings, my ethnicity, etc are just a few characteristics that are all reasonable circles one may innocently add me to. And the current circles model allows essentially anyone to “tag” me with whatever they want, as a result, constructing an interest graph on my behalf.

Advertising:
Google+ does not currently have display ads, which is very nice! However it is clear that their large app suite is pushing towards some kind of convergence, so there will be ample advertisement opportunities down the road. Could information/context provided by another person about me really find its way onto my screen? Lets try:

Unfortunately, most recruiting emails I get are not too far off from this spoof

I sent an email to myself as Joe Co-Founder who was looking for a Rockstar Ruby Developer. I did not choose to get this email, though I did choose to open it. And (as expected), the adverts are all tailored to the context of the email. This is good advertising, this is targeted advertising, and this is only the beginning.

Choice:
Choice is always nice. Currently in my bio, I can choose to omit certain bits of data that I feel are too personal. On Facebook, if I had chosen to not associate myself with my high school in Toronto, I would simply remove it from my personal information. With that one simple choice/action, it would eliminate any explicit way of connecting me to the school I went to before University. On Google+, I am almost certain that people have started to add me to the “Northern Secondary High” circle, because that makes perfect sense, because that is exactly what circles were designed for. I have no choice in the matter, I have been labeled. I can choose to not use Google+ of course, but I enjoy using Google products. I could choose to not use circles at all, but the current UX requires me to put everyone into at least one circle.

Parallels:
Some have described “Circles” as being analogous to Facebook lists, for which I would disagree. While the concept is similar, the privacy model (in the context of this article) is very different. I can only add people who are my friends to lists. People can only be my friend if either they or I have permitted the connection. Therefore, the people that may be adding me to lists are within my accepted/trusted circle. Some have also drawn parallels between circles and Twitter lists. Again there are differences. While the asymmetric nature of the following graph is similar to Google circles (anyone can add me to a list), the difference is that these lists are all public to me.

Lists I have been added to. Lists created by people I know/people I don't know...

I’ve been added to ~100 or so lists, and I can check them out and see what people have identified me as. Nothing too surprising, but the information is all there. If I found a list I did not like being in, I could simply remove myself from said list*.

To me, Circles are not like either Facebook/Twitter lists. I see closer parallels in other places on the web:

delicious.com: tags associated with a specific bookmarked url

stackoverflow.com: Tags associated with a specific question

This “tagging” model is much closer to circles in my opinion than anything else I’ve seen on the web. And as I said before, it feels weird being tagged/indexed like a book, a question, a url, especially when I have no control over it.

“Ian, you are a rambling lunatic…..”
I’ve chosen to *really* nitpick a single feature. I really don’t think there is some super secret master plan here. The purpose of this entire post is to just be a talking point about privacy, the social web and ad targeting in general. I think that Google+ is great. I really find the circle metaphor an interesting approach and concept. I do not believe that they built it as some kind of secret way of mining data about people, and truly hold the opinion that it was designed to empower users. However I acknowledge that creating a rich, strong taxonomy/database of people and characteristics would be incredibly hard to do purely automated. I kind of feel like the cirlces metaphor is the greatest Mechanical Turk task ever created which is organizing and labeling the people of the world. What circles do you think you have been added to? If you decide to add me to a circle, maybe just add me to your “following” circle. And while your at it, you might as well just follow me on Twitter, I’ll probably follow you back….

Editor’s note: This is obviously an expression of my own opinions and not my employer. While my employer is publicly being considered a competitor to G+, I would hold the same thoughts and unbiased opinion regardless of where I worked. Also, thanks to @michaelcvet for the feedback and revisions on this post.

[*] On the difficulty of automation: If you disagree, stop reading my blog, make the product and make millions!

[**] On Twitter lists: I acknowledge that there are some complications with private users/lists, but that is far from the average use case.