Tuesday, December 18, 2018

Incentivize Social Media to Starve Disinformation, Not Promote It

Connect the dots:
  • The reports of social media enabling disinformation, evading responsibility.*
  • The problem of motivated reasoning.*
Social media will not manage disinformation effectively until they no longer profit from it. We are trying to stop the tide of disinformation, when all we can do is limit its spread and impact. Tides cannot be stopped, but they can be managed -- if the managers are motivated to do so.

Follow the money. It is well established that
  • Social media are optimized for engagement, so they can sell ads.
  • Disinformation enhances engagement.
  • Therefore, social media profit from enabling disinformation to spread; they lose money by limiting harmful engagement.
The only systemic solution is to change their incentives: their salary (and stock value) must depend on revenue from users, not advertisers.

That may seem impractical, given where we are now, but there are powerful tools for changing that:
An Open Letter to Influencers Concerned About Facebook and Other Platforms
(*Quoting Congressional report, as reported in NY Times, and Sinclair Lewis)

Monday, December 3, 2018

Reverse the Biz Model! -- Undo the Faustian Bargain for Ads and Data

It is time to reverse the fundamental premise. Many now see that the long-popular model of free digital content (or other services) -- in exchange for advertising and personal data -- has become a Faustian bargain with the devil. It is bad for both users and their service providers. We are losing our souls to empty but addictive engagement -- and to destructive disinformation. Journalism is failing, music is reeling, video is struggling to find viable subscription models, and Facebook is poisoning our democracy.

The public barely took note when newspapers could no longer live off the fat of classified ads. Then "digital pennies" replaced "analog dollars" more widely, and, still, few cared. But now the devil has come for all of us. The increasing price of this deal with the devil has reached crisis levels. The recent PBS Frontline documentary, The Facebook Dilemma, reports in depth on how Facebook sold its soul and still seems to only barely realize it. Or, as the NY Times reports, maybe they do. Why should they care, when they are making billions?

The press and government may investigate, but what can anyone do? Governance, regulation, or breakup do not get to the root of the problem. What we have here is a business model problem. What we need is a business model solution. That is not as hard as it seems. 
This problem goes far beyond Facebook. Most ad-supported business models suffer from mis-aligned incentives. This post was first written with emphasis on Facebook and other social media, but has been lightly edited [4/16/19] to make it more clear that it applies broadly. 
The next two sections focus on social media, but the rest is applicable to making any ad-supported service more win-win for customers, advertisers, and publishers/platforms.
Getting to the heart of the problem ... and some alternative paths to a solution

In "A Blueprint For A Better Digital Society" (in HBR), Jaron Lanier and E. Glen Weyl provide a thorough analysis of why these ad-supported services have proven so harmful -- and offer their blueprint for a much better model.

Here, I draw on that as background, to outline a simpler and more immediate path -- one that enables individual businesses act on their own to credit consumers for the value of their data. Starting there would shift incentives to better-enable the wider market in data that they propose.

Lanier and Weyl provide an excellent primer on the problem:
...the dominant model of targeted advertising derived from data surveillance and used to fund free-to-the-public services like social media and search is increasingly viewed as unsustainable and undesirable.
Today, internet giants finance contact between people by charging third parties who wish to influence those who are connecting. The result is an internet — and, indeed, a society — built on injected manipulation instead of consensual discourse. A system optimized for influencing unwitting people has flooded the digital world with perverse incentives that lead to violations of privacy, manipulated elections, personal anxiety, and social strife. 
They set the stage for a proposed solution (emphasis added to points I will address) :
As we wait helplessly for more elections to be compromised, for more nasty social divisions to be enflamed, for more invasive data surveillance, and for more workers to become insecure, the widespread assumption that no other models are possible leads to a state of despair.
But there is an alternative: an emerging class of business models in which internet users are also the customers and the sellers. Data creators directly trade on the value of their data in an information-centric future economy. Direct buying and selling of information-based value between primary parties could replace the selling of surveillance and persuasion to third parties. Platforms would not shrivel in this economy; rather, they would thrive and grow dramatically, although their profit margins would likely fall as more value was returned to data creators. Most important, a market for data would restore dignity to data creators, who would become central to a dignified information economy.
These models have been discussed widely for years. Here, we describe a future based on them by exploring the business and societal structures that will be required to bring them to life. In the process, we will advocate for a more coherent marketplace. Without one, no corrective measure stands a chance.
...A coherent marketplace is a true market economy coupled with a diverse, open society online. People will be paid for their data and will pay for services that require data from others. Individuals’ attention will be guided by their self-defined interests rather than by manipulative platforms beholden to advertisers or other third parties. Platforms will receive higher-quality data with which to train their machine learning systems and thus will be able to earn greater revenue selling higher-quality services to businesses and individuals to boost their productivity. The quality of services will be judged and valued by users in a marketplace instead of by third parties who wish to influence users. An open market will become more aligned with an open society when the customer and the user are the same person.
They refer to this kind of "market economy for information" as providing "data dignity" and note some important challenges:
The foremost challenge in implementing data dignity is the yawning gap between big tech platforms and the individuals they harvest data from. If we asked big tech alone to make the change, it would fail: Too many conflicts of interest exist, and the inevitable concentration of power these platforms create is inimical to competitive markets and an open society.
For data dignity to work, we need an additional layer of organizations of intermediate size to bridge the gap. We call these organizations “mediators of individual data,” or MIDs. A MID is a group of volunteers with its own rules that represents its members in a wide range of ways. It will negotiate data royalties or wages, to bring the power of collective bargaining to the people who are the sources of valuable data. It will also promote standards and build a brand based on the unique quality and identity of the data producers they represent. MIDs will often perform routine accounting, legal, and payment duties but might also engage in training and coaching. They will help focus the scarce attention of their members in the interest of those members rather than for an ulterior motive, such as targeted advertising. 
Boiling the ocean of two-sided markets -- Faust wins the world (for a time)

This vision of MIDs is a worthy one, and one I hope will succeed. But I suggest a path that traverses its way up this hill in a more indirect path might be more feasible.  That still faces challenges, but they may be far more easily overcome.

Lanier and Weyl point out that MIDs are not a new concept, referring to pre-Internet examples. But I find a more concerning case in point to be the still-thought-provoking proposal for "infomediaries" in the 1999 book, Net Worth, by John Hagel and Marc Singer of McKinsey. That drew attention when published, but got little traction in practice. That history seems largely forgotten by those now proposing similar ideas. Their infomediaries seem to be much the same as Lanier and Weyl's MIDs.

I have been wondering for years why this vision did not come to be. I have seen no clear answer, other than that no one found a path to achieve the critical mass needed to establish such a multi-sided market for consumer data. MIDs face the same problem of critical mass.

Instead, the path taken was that the ad model proved wildly successful for Facebook and Google. That gave them the critical mass of users, and huge financial clout. That now makes it even more challenging to introduce infomediaries or MIDs -- whether by convincing the dominant platforms to enable that, or by competing with them.

The "reverse meter" as the essence of a market economy for information

FairPay suggests an alternative path toward a market economy for information -- one that may not go as far as infomediaries or MIDs, but which can be pursued unilaterally by individual businesses, in direct cooperation with their customers. That could set the stage for more customer-driven solutions -- for any ad-supported business.

Infomediaries and MIDs are, in essence, a way to create a "meter" for the value of data and attention:
  • Data and attention go from the consumer to the businesses that want to use it, and in exchange, funds go back to the consumer. That reverses the normal "metering" of service to the consumer, in exchange for funds to the business.
  • But we don't need infomediaries or MIDs to do that. A more basic kind of reverse meter can be applied by any paid Web service business to compensate users for their data and attention. That reverse meter offers direct benefit to those businesses and their customers.
The basic idea of the reverse meter is much like reverse metering of co-generated power when it is fed back to a power company's grid -- instead of paying the power company, the consumer gets paid. (Jeff Jarvis of CUNY School of Journalism suggested using reverse metering for online newspapers in 2011, when "metered paywalls" were a new thing.)

The first businesses to offer such reverse metering have not been in social media or search, but as the model is proven effective, it can motivate similar changes in those business sectors. The easiest place to start is in businesses that already charge users -- data/attention credits can simply offset fees for service, so no funds need be paid out directly. Thus services for news, music, video or social media that now charge users -- as alternative to showing them ads -- can offset those charges using a reverse meter that meters the value of attention they provide. (Some already do, as noted below.)

The beauty of the reverse meter, much as Lanier and Weyl explain, is to make for an exchange that benefits all parties of that exchange. The consumers gets credit for their attention and data (as quantified by the meter). The advertiser or data user gets value that they are willing to pay for. The service provider profits from operating this marketplace. This has great power because it aligns the incentives of all three parties:
  • Now the deal is obscured, arbitrary, and one-sided -- "We give you free service and you surrender whatever amount of your attention and data that we extract. We hope you will just accept that."
  • With a meter, the deal is quantified -- "For X units of attention or data we give you $Y of credit against the fee for our service. The meter will quantify that."
Reverse metering can be simple, done by any business

There are already many basic examples of reverse metering:
  • A number of services (such as Hulu, Spotify, USA Today) already offer a simple alternative to advertiser-supported "free" service:  instead, opt for ad-free service with a paid subscription. That puts a specific value on advertising, and gives consumers a basic level of choice over whether to accept that value proposition.
  • Some ad-blockers offer similar options to control ads, and some publishers are participating.
  • One clever new service, Paytime, offers a kind of reverse metering that entirely decouples the advertising from the primary service (a bit like a very simplified infomediary or MID). Consumers with more time than money can watch video ads to get credit, to use to subscribe to Netflix, Spotify, or other services. Instead of interrupting the primary service experience, the consumer can watch the ads whenever they wish, and has some choice as to which ads they watch. Equivalent functionality could also be integrated directly into an individual service business.
[Update 12/20/18: The IAB (Interactive Advertising Bureau) has recognized forms of this as "Opt-in Value Exchange advertising...an honest transaction that provides value to the Advertiser, Publisher (developer), and the Consumer," and provides detailed guidance on why this is important. Here I suggest a broader context.]

Win-win-win for consumer, provider, and advertiser

Think about how even a simple reverse meter changes advertising. Now:
  • Consumers are annoyed by intrusive and annoying ads and abuse of their personal data. 
  • Advertisers are frustrated that they cannot get their message through, even at high cost.
  • Publishers/service providers are caught in the middle, as ad rates fall, customers get angry and install ad blockers, and their business suffers on many fronts. 
With a reverse meter, consumers are compensated for their attention and their data. Just having such a meter quantifies the value of data and attention, and implies a price for that value. So once it is metered, consumers will see that, and can judge if the price is right. If the price seems fair, they will accept the ads, if not, they will pay to avoid them. (FairPay provides advanced methods for setting this price for value from the consumer, as well as the price of metered value to the consumer, but even simple methods change the game.)

Once we begin to think in terms of metering the value of attention and data, we are able to get far more efficient in maximizing that value, even within a single service business:
  • For consumers, what is the value (or cost) of the ad to the user?  Is the message relevant, timely, interesting, entertaining, useful? Or is it a just annoying. Is the delivery of that message intrusive? Can I enhance that by having some say in what messages I get, and when I get them?
  • For  advertisers, what is the value of a more balanced relationship with the consumer? Am I getting my message to a good prospect, in a way that builds my brand? Can I build a relationship with the consumer, so that they help me craft just the right message? Can I get a direct response to my ad (including simple feedback, even if there is no purchase). Or am I turning off the people I want to reach, and wasting money on the wrong ones? 
  • For  publisher/service providers, how can I maximize that shared value so that I can earn a share for myself, and make my service more popular (to get more value share from more consumers)? Do my customers feel that the ads add value or subtract it? 
Remember that advertising can be valuable when relevant and useful, or entertaining. Don't you get value from ads for products you want, or may want -- or those that make you feel good? There are many examples of valuable ads: just look at magazines for fashion, style, travel, sports, lifestyle, or hobbies. Some buy them mainly for the ads. Why do people like to watch and talk about Super Bowl ads? (some watch just for the ads). With simple reverse metering, service providers can provide a basic marketplace where their customers can interact with their advertisers to maximize value all around.

Economics and business models are all about aligning incentives. Metering data and attention enables everyone to manage incentives so they can all maximize value. That is the sustainable path to long term profit. Quietly addicting users to engagement, by spreading disinformation and sensationalized content, provides no real value -- it destroys it. Sooner or later, that model will self-destruct.

Customizing services and prices to serve all

If we can bring this kind of flexibility and economic efficiency to the reverse meter, why not the primary meter? To fully align incentives, we must customize the value proposition to optimize all forms of value for each party. Why should all consumers pay the same? Why should we all get the same level of advertising? Why should we have few choices about how our data is used, and get no compensation, whatever the level of data usage?

This is the age of "mass customization" and "one to one marketing," but we are not being creative about that. Zuckerberg admits the problem applies to Facebook, but sees no solution. He described his dilemma to the Times:
…having [Facebook] be free and have a business model that is ad-supported ends up being really important and aligned…Now, over time, might there be ways for people who can afford it to pay a different way? That’s certainly something we’ve thought about …But I don’t think the ad model is going to go away, because I think fundamentally, it’s important to have a service like this that everyone in the world can use, and the only way to do that is to have it be very cheap or free.
But the with the reverse meter, an array of variable options can be provided. A full ad load for free access, a full price for ad-free service, and a range of options in between. 

FairPay expands on this idea of tracking value and giving consumers more choice about the value propositions they are offered. It provides an architecture for metering and setting a price on value in the individual context of each customer. An example that explores use of the reverse meter is Patron-izing Journalism -- Beyond Paywalls, Meters, and Membership

Self-regulation, government mandate, or new market structures?

The big question is how we get from where we are to "a coherent marketplace" for data and attention:
  • Introducing infomediaries/MIDs has the problem of critical mass that is inherent in any two-sided marketplace.
  • Simple, company-specific reverse meters do not require a critical mass, only enough scale to justify building the reverse metering system. Very simple forms are already finding success in practice (as noted above). FairPay will take somewhat more effort to develop, but still can be within reach for many businesses (especially if supported by SaaS providers). That can begin to establish a more level market for data and attention. If the marketplace is level for all participants, including consumers, "data dignity" will be a natural by-product.
  • Facebook (and similar consumer platforms) could voluntarily begin to experiment with reverse metering now, starting with narrow trials, then learning, and expanding. Premium services could be offered to introduce the idea of consumer fees (offset by the reverse meter). As consumer revenue increased there would be less need for ad revenue. They could start simply, ins specific segments, and then expand and add the richer functions of FairPay. Change at the scale of Facebook might have to be gradual, but it is in their interest to start somewhere. This is explored further in Who Should Pay the Piper for Facebook? (& the rest).
  • If Facebook or other consumer platforms fail to act voluntarily, a simple regulatory strategy could force that -- in a market-driven way. Instead of mandating how to fix their business model, the government could simply mandate that X% of their revenue must come from their users -- with a timetable for gradually increasing X.  This is much like how auto emissions mandates work -- don't mandate how to fix things, just mandate the result, and let the business figure out how best to achieve that. Since reverse metered ads would count as a form of reader revenue, that would provide an immediate incentive for Facebook to provide such compensation. This strategy is outlined in Privacy AND Innovation ...NOT Oligopoly -- A Market Solution to a Market Problem.)
  • All of the above partial steps would create a market for data that infomediaries/MIDs could compete in. By introducing reverse metering for the value of attention to ads and release of personal data, we would begin to establish a market value for it. Once that value is established, then we have a clear motivation to look to infomediaries or MIDs -- if they can exchange that value more efficiently. 
  • If we already have SaaS services that operate reverse meters for multiple consumer service businesses, such SaaS services could, themselves, expand to add infomediary/MIDs functionality. That provides a natural path for evolving into a multi-sided marketplace for all of the consumers and businesses (service providers and advertisers) that they serve.
FairPay is agnostic as to whether that market is directly between businesses and consumers, or includes MIDs.  But FairPay explicitly puts the value of attention and data into the overall value proposition.  Consumers will be able judge to what degree a business compensates fairly for their attention and data, and decide whether they are satisfied with that.  If so, fine -- they can work with them directly.  If not, then they will be motivated to use an infomediary or MID.  In that way a “coherent market” (or at least a more level one) can come first, and so provide fertile ground for the emergence of MIDs.

Of course there may be many paths to this goal, but this one seems to go where we want, in manageable steps. It promises to change the nature of business relationships in a way that enables a more human form of market capitalism. And, at the same time, it can lead to increased profit in the long term -- by being more economically efficient in serving a very wide range of customers with individually customized value propositions.*

*On that final point I think Lanier and Weyl may understate the business profit value of their proposal. They say (emphasis added):
Platforms would not shrivel in this economy; rather, they would thrive and grow dramatically, although their profit margins would likely fall as more value was returned to data creators.
I submit that the economic value of advertising can be increased significantly by being better targeted and better received, and using more productive and appealing formats -- all driven by aligned market incentives. That might well increase platform profit margins, since advertisers will be justified in spending more than they do now. And as Lanier and Weyl hint at, but do not emphasize, even at lower unit margins, more customers can mean higher total profit margin.

More on this theme from Lanier and Weyl

Those authors have collaborated with others in other important works that explore the transformative potential of the economics of reverse metering.
[Update 1/27/20:]
Surveillance Capitalism by Shoshan Zuboff:  For an eye-opening wake-up call on the harm of our current business models (even for those who are tuned in), see Shoshana Zuboff's NY Times op-ed summarizing her views on "Surveillance Capitalism."  But the holy grail question is “who does it serve?” Not us, as Zuboff makes so clear -- but instead of killing the growing golden goose of data (as she seems to suggest), we should require that the business model be reversed so that golden goose serves us! -- as I suggest here.

More about platform regulatory issues is on my other blog.

More about FairPay:

A concise introduction is in Techonomy"Information Wants to be Free; Consumers May Want to Pay"

For a full introduction see the Overview and the sidebar "How FairPay Works" (just to the right, if reading this at FairPayZone.com). There is also Selected items (including links to videos and decks). 

The Journal of Revenue and Pricing Management, "A Novel Architecture to Monetize Digital Offerings" provides a scholarly but readable overview. 

Or, read my highly praised book: FairPay: Adaptively Win-Win Customer Relationships.

(FairPay is an open architecture, in the public domain. My work on FairPay is pro-bono. I offer free consultation to those interested in applying FairPay, and welcome questions.)