July 04, 2011

On Asset Classes and Relationship Portfolios

As the wretched few who have been with me for far too many years know, there are moments, thankfully few and far in between, when I emerge from a haze of existential angst and school- or work-induced stupor to direct my attention and incoherence upon some subject other than the futility of life. One such phenomenon with which I have occupied myself over the years is social networking, or more specifically, the proper mechanisms by which to sort the disparate groups of individuals—value-added, commodity, and deadweight—that comprise one's relationship portfolio to protect what little semblance of privacy still exists on the web.

And while I have in the past touched upon this topic, first, in analyzing the state of relationships on Facebook back in 2005, and then in advocating for asset liquidation in 2008, I must admit that I have never fully thought through how to best implement what is fundamentally a rather complicated access control problem rife with real world social considerations. The advent of Google Plus last week and this holiday weekend presented an excellent opportunity, however, to sit down and think (alone and in the dark, obviously); what follows are some initial thoughts on the subject.

At the outset, I will note that though I will employ examples from both Facebook and Google Plus, this is not principally a comparison of the two; I will leave it to others to hammer out the differences, although I doubt any will do it better than XKCD already did last week.

Disclaimer specified, let's move onto some (incoherent) analysis. The first observation to make here, which I think is too often not made explicit in these sorts of discussions, is that access control on social networks is fundamentally at least a two dimensional problem. That is to say, when deciding what to share and with whom, at least two variables will come into play—how well you know the person (a credit rating of sorts) and the context in which you met him or her (an asset class of sorts). The first is easier to think about than the second, and I will therefore take each of those concepts in turn.

Level of Friendship

The level of friendship, I would think, is what people naturally think of when the prospect of sorting relationship portfolio assets is brought up. Is this particular asset a family member? A friend? An acquaintance? A head nod? Someone I don't even know?

It is also probably the easier problem to solve when it comes to sorting and assigning levels of access. Perhaps some exceptions exist for family members, but generally speaking, a relationship portfolio asset won't be both a friend and an acquaintance simultaneously. And since an asset can only fall into one category, sorting becomes easy. You simply place them in the right bucket (or circle in the parlance of Google Plus) and you're all set. Since one class subsumes the other (again, with family perhaps the exception), access control in this dimension is rather simply implemented: friends have more access than acquaintances who have more access than head nods and so forth and so on.

Context

Unfortunately, however, that's not the end of the story in our modern and transient existence and that's precisely where the trouble with implementing access control on social networks arises. Say I want to share a recent photograph of myself taken with Oski at a Cal football game. With which subset of assets in my relationship portfolio should I share this photo? Level of friendship is easy to determine: I only want friends and family, not acquaintances or the like to see this photo. Problem solved? No!

Given that I have acquired friend level assets in all sorts of contexts in the past decade or two, blasting off a photo of me and Oski to all these folks would likely be overkill. Many may not know who Oski is or care to see me decked out in Cal gear with a college mascot. In other words, relationship context (asset class), not just level of friendship (credit rating), matters when thinking about this problem.

The subset of assets I really want in that example is friends whom I know from Cal. And to get that list, I would have to have sorted all assets in two dimensions upon acquisition: each would need to have been assigned both a level and a context to get the subset that I need.

That, however, is when things start to get messy. Context, unlike level of friendship, is far more fluid of a concept. Perhaps it might be easy to sort people based on discrete periods in one's life such as high school or college or graduate school. But such discrete periods are hardly the only places where one acquires assets (I mean, meets people) in so-called real life. What about friends of friends met while living in some particular place? What about coworkers at various different jobs? What about friends of exes or exes of friends?

And that's to say nothing of people who may fall into multiple contexts. High school and college? College and grad school? Grad school and high school? High school and New York? I've got assets in my relationship portfolio that meet all those various categories and many more. What to do with all those folks? Assign multiple contexts?

Towards a Solution

Actually, maybe. I haven't had a chance to play around much with Google Plus yet, but the circles concept unlike Facebook's haphazard list implementation, seems like it might be just the thing to capture both level of friendship and context. The default setup takes care of the former and creating circles for various contexts shouldn't be too difficult. I wasn't immediately able to tell whether adding two circles to a share takes the union or the intersection, but at least conceptually, it should be possible to choose between the two.

As for addressing the context fluidity problem, a reasonable compromise might be to use discrete periods of life plus geographic locations as contexts. So, for example, I might have separate contexts for high school, Cal, UChicago, my various jobs over the years, San Francisco, Chicago, and New York. Overlap could be addressed by assigning multiple contexts and the truly random assets could be placed in a catch-all miscellaneous bucket.

Is It Worth the Effort?

Based on the foregoing discussion, perhaps then it is possible to solve this problem in some reasonable amount of time. So, why haven't I yet on a place like Facebook—still the principal social network on which most activity arises? Why are my lists half-assed, incomplete, and nonfunctional, and my shares essentially binary—no one or everyone?

Part of the reason, of course, is that Facebook does not make it easy to manage privacy or implement granular access control. The list mechanism and the privacy settings generally are constantly in flux, making it difficult to use and potentially costly (in terms of data exposure) if some setting gets messed up. And with Facebook constantly opting in its users without their permission into new and more egregious shit, the safe bet is to share very little of consequence (as I observed last year).

But a more significant part of the reason might be that it's an gargantuan task that offers little, if any, benefit. I have somewhere around 750 friends on Facebook (though that number is constantly fluctuating through no action on my part for reasons that escape me entirely). Sorting that list into one dimension, let alone two, is not something I will ever find time for, I don't think. And if I, someone who is about as obsessive-compulsive when it comes to data as it gets, can't be bothered, how many others will be?

Perhaps if Google Plus hits a critical mass, some combination of sorting upon addition and a more discriminating approach to asset acquisition can remedy the tedium of sorting. But perhaps not even then. As Heavy-D observed earlier this weekend, the advent of Google Plus has really crystallized the pointlessness of social networking. Is any of this shit really worth the effort?

Maybe we should actually be spending time with some of these people we know (insofar as anyone can know anybody else in this miserable charade known as life) rather than sorting them into increasingly granular categories like so many stocks and bonds on a fucking website. I mean, has it really come to that? We really are doomed.