Skip to content

Das Zuck-ital: The Economics of Social Networks and the Collapse of Privacy

June 4, 2010


After the wave of anger following instant personalization, Facebook has since issued an odd, self-justifying blog post and withdrawn to a kinda-sorta seeming compromise on its privacy policy. But generally the lingering question seems to be not if, but when the next attempt by a fallen social media darling to violate privacy goes down.

But even beyond this question, I think there’s an interesting deeper issue lurking here behind all the media hubblaboo of the Facepalm saga. A question that has more to do with the entire ecosystem of social media generally, rather than the specifics of who or what is going to pose a threat to personal information. Namely:

Is Facebook just an isolated case of a company gone wrong? Or are increasing violations of privacy just the typical behavior of a mature for-profit social network?

And, more generally, are privacy violations on the part of traditional social networks just a unique case of social media gone wrong? Or are they just the first in a broad and growing trend among all social media services (thinking here of microblogging, etc etc)?

The key here, might be to examine the rawest motivation of these companies, which, like all companies, is simply the need to stay solvent. And, if advertising revenue is the cornerstone of the business model for social networks, then one of the fundamental engines determining their activity is a simple formula — the formula for advertising revenue:

(Total Impressions of an Ad) X (Clickthrough Rate of the Ad) X (Cost Per Click) = Total Advertising Revenue

From this formula, I’ve been thinking that there’s a way to derive a pretty neat little economic model. A model which interestingly seems to suggest that as businesses based primarily on the revenue from online advertising continue to expand, they inevitably will try more and more intrusive strategies to acquire data about users. In short, that invasions of privacy are just part and parcel of the mature behavior of a certain type of business that makes online advertising a cornerstone. What’s worrisome, of course, is just how many “businesses that make online advertising a cornerstone” implies in the web startup world.

This argument why is a bit involved, so it’s worth going through it step by step (geekery follows).

One of the great natural advantages of social networks is that they enjoy network benefits as they grow. More active users increases the value of the social network, so growth tends to accelerate exponentially if a service is successful. From the impressions point of view, this is a great story. In the initial phase, a virally growing social network provides sufficient impressions to drive a situation where online advertising covers most if not all costs and then some.

Sadly, betting on the side of entropy, impressions are generally doomed over time. Existing users get bored — they stop coming back, or come back less frequently. After reaching sufficient size, there’s also just less potential users to acquire, which slows the rate of growth. As a large multinational company, you can do some things to shore up this inevitable decline — expand internationally, for example, but in any case the story remains the same: acquiring impressions gets more difficult over time. As a result the long-term growth of a successful social network over its lifetime looks something like the “S” curve shown below, with exponential growth in the beginning tapering out to equilibrium later on.

Without any particular efforts to improve cost per click or clickthrough rates (meaning that they’re held roughly constant, then, this simply means that revenue generally stalls over time. This problem gets compounded since the cost side of a social network doesn’t stay constant against revenue. Far from it, beyond maintaining the cost of many millions of users for a free service, factoring in the development and deployment of more (and higher bandwidth) features to keep users interested and impressions inflated tends to actually accelerate costs over time. Barring some unexpectedly improbable generosity from media buyers to shore up social network revenue by agreeing to substantial increases in cost per click (unlikely since declining or stagnant impressions lowers the value gained from an advertisement), the only remaining factor to keep the cash flow strong is the key variable of clickthrough rates.

Clickthrough rates are the function of two things. We have data, on one hand, essentially determining how much we know about users. And “technology,” broadly speaking — which pulls from the data in an effort to get people to click on an advertisement. This embraces everything from using hard Google-style quantitative scripting to target ads based on the interests of users, to the qualitative design work that gets done to create appealing designs. Technology is a machine for converting the value within the data into actual money through users clicking through ads. Knowing my music tastes and the music tastes of my friends, for example, is the data. The fetching felt-owl-Etsy designs and the strategic placement in front of lovers of similar bands on Facebook when I log on, is the Technology.

But, really, a given piece of data (or a big bundle of data) has only so much exploitable value for this purpose. Technology can only go so far in extracting useful information from a set of data. This is due to the limitations of how much design can add to the attractiveness of something, and how good predictive algorithms are at delivering advertising to susceptible people. This problem is particularly exacerbated since nearly all studies show that users are becoming increasingly desensitized to online advertising. Despite wheelbarrows of cold hard cash flowing into research and online advertising coming into its own as a huge industry, we’ve actually seen a decline in clickthrough rates over time. We’ve gone from 2% in the 1990s to 0.28% by 2003, and by 2010, it’s good to get something in the 0.1 – 0.2% range. This is another good reason for Cost to Click to stay steady or even decline as well, since the quality of clicks gained from an ad buy is more likely to contain a bigger proportion of mistakes and other noise. This makes it a less valuable proposition to buyers.

As technology attempts to squeeze an ever declining bit of value from the total number of impressions hitting a site, social networks face the big crunch of increasing costs and slowing revenues. This produces an increasing pressure to find ways of boosting advertising revenue again.

And, conveniently enough, one other unleveraged door remains open: data. That is, rather than expand the technology to pull ever smaller marginal values from data or crudely try methods to boost impressions, it’s possible to just attempt to acquire more complete information about users. Indeed, as the other components determining advertising revenue fall down, the incentives for taking this route get higher and higher.

It’s not controversial at this point to state the obvious: features on a social network or social media are essentially gateways for a service to collect data about users. In addition to improving clickthrough rates by garnering more data, refactoring these features to acquire deeper and broader information about users provides a universal push upwards to all variables in the advertising revenue equation. Specifically, it raises impressions by giving more places for users to see ads and it raises cost per click by potentially opening up some more valuable on-screen real estate.

The conclusion? Maturing social networks, particularly with slowing user base growth and with advertising as a primary business model, become increasingly hungry for data about their users as they age. As a result, they inevitably tend towards invasions of privacy. The incentives increase to find ways to gaining more data about users to increase revenue, at first through non-intrusive and less publicity costly ways. But there’s only so many options to do so, particularly in cases when a user base is unwilling to give up more information more quickly through opt-in features.

As the benign invasions of privacy become less easy to find, the remaining options are to implement increasingly creepier and more intrusive “Instant Personalization”-type strategies. Stated simply, any promises for user privacy are increasingly at odds with the financial incentives of the social network, particularly as it faces higher costs and lower returns. For some companies who become storehouses of information (i.e. Twitter) but don’t want to implement large-scale advertising, incentives also exist to just provide all the data wholesale at a price (i.e. to Google), often to other data-hungry, advertising-centered companies.

There’s three considerations to this logic which make this basic picture more nuanced, but leave the conclusion essentially the same:

1) The Venture Capital Delay: Assuming that a social networking company really has a strong internal commitment to user privacy and autonomy, sufficient influxes of venture capital could stave off this effect in the short run, since it delays the need for resources with all the other factors falling. However, this is merely a delay insofar as major investors are willing to believe that other options for revenue exist, and the collapse of this artificial flow of funds puts a company back into the dilemma. It’s a time bomb for privacy based on investor confidence. Perhaps worst of all: this period of good behavior is habit-forming for users of a social network. They come to develop expectations that the service will be free and non-intrusive, and the social network becomes more “sticky” and hard to leave as more friends join. This sets up a natural conflict when this artificial infusion of money declines, as the business has incentives to behave differently.

2) The Google Exception: Here’s also another possibility: users willingly give social networks their data, which also takes companies out of this flow. So long as they are willing, the engine still runs since the data flows in without ever being truly “intrusive.” This is essentially the Google model, where control over the search market gives an enormously large, regular base for impressions — and where search and other services provide the company with the lifeblood of data to keep clickthrough rates high(er). Given the dominance of Google over key data sources that users are strongly willing to give up information to (Gmail, Google Reader, search itself), however, there’s a good argument that it’s an outlier case in the dilemma. For most other services, willingness to give up data is more limited to the purpose of the site so that the willingness to give up information only exists until the site steps “out of bounds.” For instance, we’re comfortable with giving information about ourselves and our friends to Facebook, but are less comfortable as soon as it starts attempting to seep into every other major site in our online lives.

3) Other Business Models: Obviously this dire situation is made less dire in the case where the company has a different business model. Indeed, sites that survive off a mixture of product sales and advertising can largely avoid some of these incentives to subvert privacy. Regardless, with regards to advertising, the economics remain the same — forcing businesses to attempt to acquire more data or to change to alternative revenue sources over time to grow.

“Traditional” social networks are probably the first place where we seen this phenomena emerge since they, out of all services, fit the pattern of this theory the closest. They confront a universe where online advertising is the overwhelmingly important revenue component, they face the rising costs of maintaining users and features, and the declining factors with regard to clickthrough rates and impressions. In order to continue expansion, there’s an increasing attractiveness to become data hungry and privacy indifferent.

What potentially worrying about this logic is that the problem it gestures at is systemic: in the fifteen years of the popularity of the open web, advertising still emerges as one of only a few reliable business models online (that, and, like, selling t-shirts and books). To that end, all businesses that are founded on the advertising engine and roughly fit the situation described face these incentives, which should make the threats to privacy and personal data numerous as the entire Web 2 ecosystem ages. It’s also worrying because it’s difficult to push hard against these businesses on a regulatory front (as has been suggested by some) due to their dependence on the revenue from online advertising : doing so could deflate cashflow and produce a “big crunch” slowdown of social platforms online and their ability to expand.

The picture that’s drawn here rethinks Facebook and is prescriptive about what should be done. Rather than a single company hellbent on disregarding user privacy, it reframes the social network as just the first player that’s reached sufficient scale and age to face the difficulties of being a late-stage Web 2 service. And, if anything, this model suggests that data hunger on the part of social platform companies is something that we’ll be seeing generally more of into the future, rather than less. If you’re a strong believer in privacy, Facebook isn’t the enemy — but its business model is. Ultimately, like the fears that a massive dependence on fossil fuels negatively impacts the decisions of a society, there’s room to perhaps advocate for a diversification in the revenue engines that play a role in fueling the internet’s social infrastructure.

[Thanks to the very good folks of Web Ecology Camp IV that talked through and critiqued a early version of this idea]

Advertisements
4 Comments leave one →
  1. Zip permalink
    June 5, 2010 4:12 am

    I see the core difference between Google and Facebook privacy intrusions as being a case of public vs. private intrusions. I am much less worried about private intrusions then I am about public intrusions. What I mean by private intrusions is the Google variety- the advertisements on the side of a Google search or my gmail are tailored to my search history or conversations. These types of privacy intrusions are acceptable to me. If facebook were to intrude like this I would be fine- that is, if facebook were to just tailor adds to me personally based on my specific conversations and preferences, this would, not only be tolerable, but preferred.

    I think most people would rather have advertising tailored to their interests. When your watching TV, how many guys like seeing Tampon adds and how many girls like seeing Axe body spray adds. If I have to see advertisements (and I accept the fact that I DO have to see some advertising or else live under a rock) then I would rather see advertising for things that interest me. As an advertiser, I would rather reach my core market as directly as possible then to through up some expensive mass market campaign and hope some hit my buyers. In my eyes, targeted advertising is a ‘win-win’ for sellers and potential buyers.

    What troubles me about facebook is two things- 1) the illusion of privacy and ii) identity linking.

    To the first point- I’m not sure who to blame. I’m sure at some point I thought of facebook as being ‘safe’- maybe it was when my Mom and then rest of family ‘friended” me that I got worried that facebook was no longer ‘safe’, that it was no longer a spot for me an my friends, sort of like if Zack’s parents came and started to hang out at Max’s in ‘Saved By the Bell’- maybe it was when my co-workers ‘friended’ me- maybe it was when that girl that sat two rows up from me in algebra sophomore year of high school ‘friended’ me- maybe it was when I realized that after that one night stand in collage I ‘friended’ whatshername on fb and now she is connected in some web of friends to my grandmother that it all clicked. Facebook is not a private network- its far, far less private then my private life. But I think a lot, if not all, of this is on the user- I was the one that allowed all these ‘friend’ connections.

    So now I’m looking for a good reason to cut the ties with FB- my RL ID is too connected to my avatar. How do i get out of this. Maybe I should look towards FB and find something that pisses me off- this will be the justification I need to pull the plug?

    Identity linking- when google looks at my search and sees ‘Asian Ass Porn’ it tells ME hey Zip, check out this great site for ‘Asian Ass Porn’, your gonna love it. And I say, wow, google good job, you pulled up the best possible site and even had click through advertisers that gave me the perfect product. Well Done! Now this interaction was just between me and the google, Comcast, the NSA and maybe the person looking over my shoulder. My grandma did not get a piece of this nor did that girl that I some how facebook ‘friended’ because I had too many Tequila shoots in Cancun nor the my coworker in the shipping department. This was just between us.

    The trouble with Facebook is that is can tell others what I’m up too, and I unless I’m careful, it will. I don’t want to tell my boss that I ‘digg’ the legalize weed video and I don’t want my girlfriend to know that I just bought that flesh light. Now that I think about it I don’t want anyone to know that I just bought that flesh light. Even you. But you don’t know who I am, right- so who cares… but some of you will find out. Anonymity is a joke. And for you who do know who I really am, all I have to say is- ‘it was a gift’….

    • Tim Hwang permalink*
      June 10, 2010 3:19 pm

      @Zip — absolutely a good point, Fleshlight and all. There’s a lot here, and trying to think through all the implications and how to fit them into the overall picture.

      I think the public/private intrusions distinction as well as the two problems with Facebook (identity linking and the illusion of privacy) are an interesting function of the kind of platforms a business has access to. Presumably, data hungry businesses fill in all opportunities to make intrusions that minimize the bad publicity and user anger they’ll confront, before moving on to the less savory stuff.

      In Google’s case, the main thing that’s stopped them from doing more worrisome stuff is that they have control over private applications — as you mention, the fact that you get targeted ads based on what you’re e-mailing on isn’t necessarily a bad thing. In Facebook’s case, their platform IS a public/sharing platform as a social network, so the kinds of intrusions they can make tend towards the two things that make you uncomfortable.

      But these things are so malleable, and once the data is acquired, it’s acquired. If Google chooses to eventually make more public intrusions or set up platforms that create social sharing (see Buzz), the consequences seem as bad, if not more bad.

  2. June 9, 2010 8:00 pm

    Why not take the next step and tie this directly to Marxist theory?

    Marx posited that the internal inconsistencies of capitalism would drive citizens to revolt and overthrow the system. Will the same happen with Facebook? Is Diaspora some sort of proletariat?

    Or, as with capitalism, will this prediction fail to be realized and people will live with the (supposed?) internal contradictions that are making people worse off?

    • Tim Hwang permalink*
      June 10, 2010 2:59 pm

      @Kevin — despite the ridiculousness of the blog image (that image shoop has been sitting in my folder for months, glad I finally found a use for that thing), was thinking through it and I’m still shaky about making the final big jump and connecting to Marxist theory.

      I think the main similarities between the two theories is the claim that the fundamental engines of a certain type of organization inevitably tend towards certain negative impacts. That being said, I think a full-blown Das Kapital connection gets tenuous for two reasons:

      1) It’s unclear what “class” means online — particularly if we’re trying to fit into the Marxist picture. When we’re beyond an industrial organization model, the separation between who’s a capitalist and who’s a bourgeois and who’s a proletariat get really fuzzy. If I’m a blogger that makes money from ads, does that make me the capitalist (in that I’m making money?) or does that make me the proletariat (in that I labor to produce products that advertisers use as a platform to sell products?)

      2) While I’m coming to believe that the business model of Web 2.0 has some inherent inconsistencies that pose a threat to users, I’m still not sure if there’s a compelling argument to claim the inevitability of revolution. That’s dependent on a huge number of factors — beyond which it’s tough to rule out that user lock-in, apathy, and an increasing willingness to give up information means that these businesses might continue indefinitely harvesting more information without contest.

      Dunno, thoughts? It occurs to me that the two considerations are themselves pretty great questions.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

  • Twitter Updates

  • Elsewhere!

  • Pages

  • June 2010
    M T W T F S S
    « Dec   Sep »
     123456
    78910111213
    14151617181920
    21222324252627
    282930  
  • Archives

  • Meta

  • Advertisements
    %d bloggers like this: