Kirk Klasson

Social Subsidization and Diminishing Returns …

Google’s recent admission that Youtube hasn’t turned a profit must have put a whole bunch of VC’s, and their vaunted valuations, on full pucker.

The generally accepted premise behind most social media business models is a rudimentary production function. The simple idea that users generate content that drives activity that drives traffic that in turn causes advertisers to pay for the audience so captured. What this suggests under the usual economic rubric is that user-generated content is a variable input (aka labor) and for the most part the technology required to sustain the model is a fixed component (aka capital). This simple model hit a speed bump recently when Google sort of acknowledged that Youtube had yet to turn a profit. Doubtless a whole bunch of VC’s took note since most of the billion dollar valuations being tossed around in the social media space, like Snapchat at $15b, are grounded on the supposition that the user activity on social sites, including the posting of videos, pictures, conversations, etc., amounts to freely supplied, monetizable content, even if that content spontaneously disappears. In truth, this content is not without calculable economic value in the form of both capital and labor; capital in the form of personal devices and the networks they attach to and labor in the form of user-dedicated time to generate content.  One way to determine the exact economic value of free socially supplied content would be to come up with a substitute and determine how much cost would have to be incurred in order to generate the same amount of revenue normally obtained from user generated content if it were to be removed from the equation. Therefore, when compared to traditional paid-for media models, social media content represents a substantial user supplied subsidization with respect to the content used to generate an audience and sustain an advertising model.

Content is King

In fact, it wasn’t long ago that industry analysts, in an effort to rationalize social media business models invented metrics like content added per user or revenue per content added. (See Trouble in Paradise – May 2012) What Google’s Youtube admission amounted to was the suggestion that the engine of social media profitability, the efficient monetization of freely provided content, might not be as easily achieved or sustained as originally thought. The underlying assumption being that, all other factors being equal and essentially fixed with respect to content publication and curation, user access and management, etc., freely provided “social” content was the primary factor driving increases in revenue and profit. Implicit in this thinking is that there is a inherent network effect that exists between users and content and more of each essentially begets more of the same creating a form of positive feedback and an endless wall of eyeballs pursuing an endless well of content. Best of all, the content behind this positive network effect was substantially free.


To date there has been little academic research dedicated to examining the economic implications of user provided content subsidies to social media business models. Most studies on the impact of subsidization to production functions have been focused on the impact that government subsides in the form of tax and tariff relief has on given markets or industries. These findings suggest, as you might suspect, that the impact from subsidization is to produce changes to the cost frontiers of the companies or industry in question, improving their ability to compete or generate a profit by lowering their overall cost in the markets that they serve.


In the example above, the cost frontier for a hypothetical industry is lowered from point A to point B due to the application of a government subsidy.  Here the subsidy is essentially cost relief in the form of the removal of government burdens over and above the normal cost of production. Since government burdens (aka regulations) amount to trillions of dollars each year this may be an area more thoroughly explored in future research. By removing these burdens the cost frontiers shift. Here the cost frontier is the average unit cost at specified volumes of input. Under other circumstances, and in a slightly different context, a cost frontier can be considered a roughly curved space inhabited by angry shareholders armed with slings and arrows who lurk behind trees in the form of stacks of old 10-k’s and ask difficult and sometimes embarrassing questions. Something that managers of publically traded social media sites might want to start getting used to.

If at first you don’t succeed…

Turns out several of the assumptions underlying the rudimentary social media production function are now in question. First, it seems like the most sought out content on Youtube, leaving cat videos aside for the moment, turns out to be commercially produced music videos of pop performances that appeal to the preternaturally pubescent as, according to the WSJ, a weirdly asymmetric distribution has emerge with respect to the content viewed vs. the content contributed to the site. It seems that only 9% of Youtube’s total audience and a lower percentage of contributors is driving 85% of the content views and those views are predominantly of Lady Gaga, Justin Beaver and Felix “PewDiePie” Kjellberg. Go figure. However, the unrelenting increase in content and viewers means that the technology that supports an acceptable user experience cannot remain fixed and must keep pace or risk losing its primary audience along with any future potential audience that might find a reason to hit Youtube’s content.  Recent estimates suggest that Youtube viewing has continued to grow at roughly 50% year over year even in the face of stiff competition from the likes of Facebook. So the notion that the technology infrastructure costs of an ever-increasing content base can remain relatively fixed doesn’t seem to bear up either. So, it would appear that our original take on the social media production function needs to be revised, possibly inverted, where socially provided content becomes the fixed component with respect to cost and ever increasing additions to technology becomes the variable component. But there may be more to it than that.

…try, try again

In light of this, established social media players are beginning to tinker at the margin of the production function and there are a number of obvious options to chose from.

First, we could have a look the presumed network nature of content producers and content consumers and determine whether or not they are symmetrical in nature. In Youtube’s case it doesn’t appear that they are. Those supplying the bulk of the content and those consuming the bulk of the content do not appear to be from the same cohort. But if content is king and the king is dead its time to find a new one even if that means paying for it so long as the content so acquired appeals a cohort that advertisers want to reach. Not a bad idea and one that Google seems ready to pursue but then the subsidization provided by free content could no longer factor into the revenue scale properties of our social media production function.

Next up would be to examine the other inputs to the production function. As content, users and transactions start to number in the billions lowering the overall cost of technology becomes a key consideration. Facebook’s Open Compute Project along with Google’s Nearline announcements suggest that both are aggressively focused on lowering the cost of massively scalable compute facilities. And the reasons for this are becoming obvious, as the volume of content and transactions nears infinite proportions, the only marginal technology cost at which profitability becomes possible is zero.

Not all laws are created equal…

Over time it appears that social media has begun to confirm that some cherished myths concerning the social impact of technology adoption, while emotionally appealing, may not bear up under closer examination. This isn’t exactly new or the first time this has been recognized.  About ten years ago Andrew Odlyzko and Benjamin Tilly raised some serious questions about what was then considered “settled science” in the form of Metcalf’s and Zipf’s laws, the former with respect to the value of network connections and the latter with respect to the value of content collections.

For starters, it turns out that the notion of “value” can be highly subjective. Odlyzko and Tilly pointed out that the fact that we can connect over a network doesn’t necessarily mean that we have anything to say or anything to add to a conversation. So, even though a network may be growing, that growth does not mean that “value”, whatever that might be, is growing exponentially along with the newly connected members. The same it seems holds true for the “value” attributable to increases in a specific collection of content.  At the time they published their research, roughly the same time that Youtube got started, many believed that there was vast potential value at the very end of “long tails”.  The problem some contested was not that the “tail” was long but rather hard to get to and removing that obstacle would increase the value of otherwise orphaned content. Odlyzko and Tilly conjectured that the value of large collections adheres to a power law such that if value was synonymous with popularity, over a collection of content made up of 1 billion items, the first 1000 items would represent 33% of the value of the collection, the next 1 million items would represent the next 33% of value and the remaining items or the greater part of a billion would only represent 33% of the value of the collection. However, the cost to maintain the most popular item would be the same as to maintain the least popular item, hence, some serious diminishing returns to the scale or volume of content, unless of course the cost to maintain any additional item is zero.

abracadabra, hocus pocus

Suddenly YouTube’s numbers begin to make sense. This seems to confirm our earlier notion that the greater the asymmetry of interest or “locality” between content producers and content consumers for a given site the faster that site will approach a threshold of diminishing returns from increases to the volume of content collected.  “Locality” was the term that Odlyzko and Tilly used to explain the affinity between networked parties. What this also suggests, although it’s only conjecture, is that Tumblr’s goose is already cooked, but it’s likely no one will ever know since those numbers are going to be more deeply buried than radioactive sludge from Three Mile Island. (See Big Bets and Long Tails – June 2012)

This might also go along way explain why purpose built social sites, including those like Facebook, with greater contribution to consumption symmetry or “locality” will likely enjoy greater revenue to content scale but over a shorter duration and serving what might relatively prove to be a smaller and narrower cohort.

For the foreseeable future, social media sites will continue to chase content volume while attempting to squeeze every penny out of the technology infrastructure supporting their sites. Not a big surprise. Now it turns out that the success or failure of any given social media proposition may not have any long term repercussions, nobody seems to be waxing nostalgic over Myspace and the same will likely be true of Tumblr and, longer term, Facebook as well. However, the pursuit of social media profits by the likes of Google and Facebook may have some profound second and third order effects.

In their efforts to lower the marginal technology cost of content management they are eliminating any perceived value from large scale, commercial integrations of technology. Not necessarily a bad thing. But what this does suggest is that the value propositions of some of the largest technology firms in the world who happen to rely on the extraction of profit from the large scale integration of technology is about to come to an end. As the marginal cost of content management to Facebook and Google relentlessly marches to zero, the margins of firms whose mission is to integrate large scale technology for non-social media customers is also vanishing.

Businesses will still need technology and whether or not they choose to manage and construct proprietary technology facilities will likely hinge on factors other than the micro-economics of their business or the macro-economics of the IT industry.

But some things are certain. The age of the titans, the behemoths of the IT industry, names synonymous with vibrancy and vision are coming to an end. Their hubris is as evident in their headquarters buildings as it is in a Mentos video gone wrong.

What exactly takes their place has yet to be sorted out but judging by recent quarterly conference calls, those who the gods wish to punish, they first make mad.




Images courtesy of United Nations FAO and The World Trade Organization

Sorry, the comment form is closed at this time.

Insights on Technology and Strategy