On blockchains, ‘fake Trump news’ and uberization of Facebook.
In the digital world of virtual economies and soon virtual realities distinction between ‘fake vs real’ loses its ordinary meaning. The most important metric of the attention economy is Return-On-Attention.
Introduction
The history of the Internet is shaped by the evolutionary process called ‘unbundling’. What ‘unbudling’ means is that components which previously used to go together are broken into pieces and sold separately on the market.
Newspapers were bundles of written word and paper. The Internet unbundled words from paper and changed the entire industry. Previously the economics of paper defined the way we consume written words. Newspapers, books, magazines were shaped by the underlying economics of paper. With digital distribution of written words through the Web, the economics changed, and we got blogs, tweets along with entirely new business models. These business models were enabled by near zero cost of digital distribution.
Universities used to be (and still are) bundles of location, knowledge access, community, parties & status signalling with your future degree. The Internet unbundled access to information from geographical location, and now you can acquire the university level knowledge for free but without other perks.
Everyone can be CNN on YouTube
With unbundling, you always get new market entrants who can separately offer previously bundled components at the lower price/better quality. And for incumbents, it means having to restructure their business models and reinvent themselves in the new reality. What was previously a high margin product becomes a commodity. The New York Times competes for your attention on Facebook with your uncle’s alternative news blog. On YouTube, it doesn’t matter whether you are CNN or a 15-year old vlogger. You get the same treatment by the algorithm.
This recurring pattern is evident in hindsight but very hard to foresee in advance. Just try explaining to someone in 1994 that Instagram will unbundle fashion & beauty magazines and their friends will act like models. And yet that’s what happened.
In 2016 most people don’t realise that intelligence is being unbundled from consciousness with developments in AI. You can now get the ‘intelligence’ component without the accompanying human.
Currency issuance in particular & trust, in general, are unbundled from central banks and other traditionally reputable institutions thanks to blockchains. You can now get certainty and trust regarding financial asset transfers using maths and cryptography without the institutional component.
Algorithmic Content Ranking Needs to Get Unbundled From Google/FB/Twitter
The mechanism of unbundling is sometimes called ‘uberization’. What was previously limited to the privileged few, is now available to everyone.
What most people don’t realise is that services they rely on every day- Google, Facebook, Instagram — are also bundles of components. Just like newspapers were in the past.
But in their case, bundles consists of data layer with your identity and curation graph (what you liked, clicked, starred etc) and all sorts of proprietary ranking and reputation systems on top of them. What you see is as a user is an interface displaying curated information from the lower levels, accessible only to the platform owner.
So while everyone can publish on Facebook, only Facebook can rank this information and create a newsfeed out of it. No one else can build alternative newsfeed by accessing Facebook’s data layer. You get few APIs here and there, but the raw data allowing you to build the alternative version of Facebook is strongly protected.
Extraction Of Attention
Because Facebook’s business relies on trading your attention for dollars, it’s important that they keep your data to themselves and construct algorithms that keep you coming back. More attention equals more inventory for advertisers which then translates to revenues. This model applies to all platforms but let’s just keep using Facebook as an example here.
The most efficient way to keep you engaged is to display things relevant to you — your friends, your mom’s birthday or a friend’s cat. Once you give Facebook more data, it starts feeding your more information based on this data. Your preferences start influencing the information you see because that’s the easiest way to keep you engaged. More attention equals more dollars.
That’s all fine as long as the content is about birthday photos shared by 100 million Facebook users. It starts to get tricky once you move to 1.6 billion users discussing economic and political matters under the condition of permanent information overload. You get filter bubbles and echo chambers.
The US presidential campaign made people realise the threat coming from proprietary algorithms optimised for pure attention extraction:
And yet there’s nothing really that Facebook can do about it given its current business model.
What Ranking Algorithms Optimise For?
Facebook’s algorithms optimise for making you click and not for making you better informed.
Their entire business model relies on driving more engagement for Facebook and its advertisers.
Making you better informed means serving you information that sometimes contradicts your beliefs and might result in less engagement. In the long run, exposure to that information might make you better off but in the short term, it’s not going to generate as many clicks as ‘clickbait’ content.
But Facebook has quarterly revenue goals to meet, and it’s KPIs are not going to go up by having ‘less engaged but better informed’ users. The numbers will improve with ‘more engaged but less informed’ user base.
It’s a classic problem where short term benefits are internalised (engagement, ad profits) but long term costs externalised (misinformation, social media fatigue, filter bubbles).
Is Facebook like Coca-Cola of the Information Age?
It’s a far-reaching metaphor, but I believe it’s accurate. Both FB and Coca Cola sell you quick hits of dopamine caused by ‘sugar’.
In the case of Facebook, this ‘sugar’ is informational, in the form of novelty, confirmation of your beliefs, social validation, etc. Small amounts of sugar can be good or neutral to you.
The problem is that we’re evolutionary wired to consume more sugar than we need. Our biological responses evolved in low carbohydrate and low information ecosystems where the environment itself created natural limits preventing overconsumption. In the 20th century, this natural cap for carbohydrates availability was removed with the explosion of processed foods and the supermarket. The 21st century removes this natural cap for information availability . And just as overconsumption of sugar results in unhealthy bodies, overconsumption of information will lead to unhealthy minds.
In biology, there’s a term ‘hormesis’, and it describes how biological organisms respond to stressors. The basic point is that ‘dosage’ matters. What’s beneficial in small doses can be harmful in large doses. The chart below explains why you feel the way you feel after consuming too many sweets or too much social media.
Source: http://www.iqmindware.com/wp-content/uploads/2014/06/hormesis-response-1.jpg
Both Coca-Cola and Facebook are economically incentivised to push you beyond your sweet spot of sugary or informational intake into the dangerous territory. To maintain their quarterly revenue growth they have to sell more cans of liquid or attract more attention/engagement to sell to advertisers.
Coca Cola hires the best people to craft recipes and design ads that’ll make people drink more Coke. I highly recommend this book if you want to learn about the extent of their practices.
Similarly, Facebook can hire the best people to fine tune the algorithms and improve the product to make people pay more attention to its Newsfeed. But more Facebook engagement doesn’t mean users are ‘better informed’, just as higher Coca-Cola consumption doesn’t automatically translate to better nutrition for consumers.
We Need ‘Uber for Ranking Algorithms’ Optimised for Return On Attention.
At Userfeeds.io, we’ve been thinking a lot how to design the economic and technological infrastructure for the attention economy. Our vision for the future is based on open data layers and user-controlled identities with marketplaces of ranking algorithms on top of them.
Imagine unbundling of Facebook’s ranking algorithms from Facebook’s data layer and letting different ranking providers compete for users attention. In this scenario, third parties could serve their version of the Newsfeed offering alternative views on Facebook essentially creating a multiverse of realities. The currency of this ecosystem would be attention.
Some of these Newsfeeds could sell advertising, and other could charge users for providing superior filtering customised to their needs. It already happens today with all sorts of social media analytics and monitoring tools, but they don’t get access to Facebook’s raw data to replicate the newsfeed functionality.
I believe this evolution from centralised content ranking monopolies into the decentralised marketplace for ranking algorithms is inevitable and necessary. You should be able to pick and choose your content filtering algorithms in the same way you select products on Amazon.
This design wouldn’t automatically fix the filter bubble problem and in the short term, it would probably even exacerbate it. ‘Trump-only’ or ‘Clinton-only’ versions of Facebook would become possible in the short term. But in the long run, new ranking methods would emerge and operate on business models other than advertising.
Return-On-Attention as a metric for the Attention Economy.
In such world, the distinction between ‘fake vs real’ would slowly lose its meaning. The emphasis would move to effective allocation of attention by maximising Return-On-Attention.
We can define Return-On-Attention as a measure of the increase in one’s economic, mental or physical well-being per unit of attention in a given context. This context is defined by one’s skills, abilities, physical location, etc.
Would Bitcoin pass the ‘fake vs real’ test in 2011? To most people at the time, it was a ‘fake’ virtual currency then. Today it’s still virtual, but it became more ‘real’ to people as bitcoin prices increased. In hindsight, we know that paying attention to Bitcoin in 2011 gave people huge returns. One might say that you had to buy bitcoins to realise these gains. Just paying attention was not enough. However, putting just a few lattes worth of dollars in early days would make you wealthy today. So money wasn’t the scarce asset back then. It was attention.
With virtual and augmented realities coming who will define what’s ‘true’?
Today our perceptions of reality are shaped by algorithms and algorithms are shaped by economic incentives behind them. So to change our perceptions we have to change economic incentives behind algorithms.
Blockchains give us the open layer that is necessary for storing data relationships currently locked up in closed silos. Crypto-tokens make it possible to program incentives and design new reward structures for people’s behaviours.
At Userfeeds.io we want to experiment with these new discovery paradigms, starting with the cryptocurrency and blockchain communities.
