Userfeeds got funded! Here’s how we plan to bring ‘skin in the game’ back to discovery algorithms.

This is a longer, follow-up post to the Techcrunch article with our funding announcement.

In 2017, no one doubts anymore that the ‘age of algorithms’ has arrived. The economic and political power of content ranking algorithms was brought to the mainstream attention during the last presidential election in the US. Issues like ‘fake news’, the opacity of discovery algorithms and centralisation of advertising revenues among dominant platforms are now being publicly discussed even in the non-tech circles.

At Userfeeds, we look at these issues in the context of a wider systemic economic change that’s happening now in the digital world. With the exponential growth of information sources and free digital distribution, human attention became the scarcest and most desired asset of the online economy.

Aggregation and control of global attention flow are the #1 online businesses because wherever attention flows money follows. The multi-billion dollar valuations of companies such as Google or Facebook result from their ability to attract, package and sell attention for dollars.

Incentives Behind Algorithms Shape Our Reality

The abundance of choices forces us to rely on algorithms to determine what’s relevant and ‘true’. It’s inevitable that these algorithms will become even more important in shaping our perceptions in the coming years.

The issue is that economic incentives behind these algorithms are opaque and not well understood.

In our view, such diverse problems as ‘fake news’, spam filtering, ad targeting, ad-blocking, newsfeed censorship and moderation are essentially information ranking problems. But companies operating algorithms behind these rankings are not neutral and have their own economic goals and incentives they optimise for.

We Have To Bring ‘Skin In The Game’ back to the Attention Economy.

Competition for attention in the current environment has a strong asymmetry that favours platforms and content producers at the expense of the users.

What do we mean by that? We already pointed out that there’s a fixed supply of attention that everyone competes for. For us users, whatever we pay attention to has a cost — we won’t get that attention back. In addition to that, once we let information enter our minds it will stay there and affect our thoughts and future actions.

Even if we try to evaluate information and decide it’s untrustworthy or misleading it is still a mental cost we have to absorb. The only way to avoid that cost is not to see that information in the first place.

The ranking algorithm should optimise users attention and prevent them from seeing the misleading information.

The question is: what the algorithms optimise for?

Unfortunately, the current ecosystem has a built-in ‘principal-agent’ problem. Platforms and content producers benefit from showing you things and putting them into your mind. The attention you provide and the data trail you leave will immediately benefit Facebook and the content producer that got your attention through Facebook.

Your attention is immediately exchanged to dollars through advertising and the data you leave behind improves Facebook. But the costs of seeing bad quality information will be absorbed by you. Potential consequences of that exposure will often be hidden and delayed in time.

In a way, incentives of Facebook (and other media platforms) are similar to incentives of Coca-Cola. Coca-Cola generates cash from sugary water sales but has no downside of absorbing health costs from overconsumption of sugar. In fact, for many years the risks of sugar consumption were not recognised, and because they were delayed in time, it was hard to link them to sugary drinks.

As an early information society, we still don’t recognise dangers of letting other people with opposing incentives to put images and thoughts in our heads at no cost.

And in the current environment, producing toxic information that is viral is essentially a free option on attention. By free option, we mean a situation where information is free or almost free to produce but has a potential to generate lots of attention dollars if it spreads virally.

Everyone benefits apart from the ‘attention suppliers’. The platform gets the data. Content producers get eyeballs and traffic. Both of which are instantly monetized. Users, however, receive a short-term dopamine boost from consuming low-quality content (which is often designed to by emotionally engaging, confirm their previous beliefs, stimulate outrage) but don’t recognise the long-term opportunity costs associated with it.

The concept of ‘skin in the game’ as explained by Nassim Taleb means that if you can expose others to harm you also should be exposed to some of this harm yourself. It’s all about symmetry. If there’s no ‘skin in the game’ in the system, it will blow up at some point. Any system where one party keeps all the upside, and another absorbs the downside is doomed to fail.

The best recent example of that is the financial crisis of 2008, where the upside from risky instruments was kept by the bank managers, but losses were transferred to the public.

We think that the current economics of the online media landscape is set up in the similar fashion.

The upside from aggregating and trading attention is captured by a few players but the downsides and delayed risks from surfacing misleading content are externalised to the users.

Show me your tokens, and I’ll tell you what your beliefs are.

Cryptocurrency innovations like Bitcoin and Ethereum are not usually linked to online publishing and advertising but rather perceived as financial innovations.

But at their core, these technologies are built on ‘skin-in-the-game’ principles and symmetry in decentralised systems.

The entire concept of ‘mining’ in the world of cryptocurrency is nothing more than a mechanism proving that someone has sacrificed some economic good (in this case — electricity) to receive a reward from a network in the form of new coins. It’s ‘skin in the game’ rule applied to token allocation.

No one cares if you say you support Bitcoin. You prove your support with ‘proof-of-work’.

What if we applied the same idea to allocating human attention?

What if ‘attention seekers’ had to back information they promote with similar ‘proofs-of-sacrifice’ instead of easily gamed likes or votes?

What if we could bring symmetry to online relationships where every time someone wants your attention they’d have to prove what incentives they have, and they’ve had to sacrifice something to make you pay attention to them?

The concept of digital tokens on platforms like Ethereum, allows us to design systems where tokens can be used for signalling relevance of information to communities owning these tokens.

‘Tokenized’ Ranking Algorithms

At Userfeeds, we develop the engine that will support many ‘tokenized ranking algorithms’.

By incorporating token-based signals into information ranking algorithms, we can install ‘skin in the game’ filters for our attention.

Today, when you see an ad on the Web, you see it because somebody created a ranking signal by paying dollars to the advertising network which you probably don’t own.

Advertising platforms are essentially platforms for ranking algorithms that use dollars or euros as their primary ranking mechanism.

If you transfer dollars to me, I’ll boost your position in the search results — that’s the deal with the AdWords platform. Using this simple but powerful value proposition, Google makes its money.

How would the economics behind advertising change if owning a token would be a prerequisite for getting the attention of a targeted community?

Send Signals With Your Tokens.

Tokens are stored on the public ledger, visible and accessible for everyone. Token holders can already send ranking signals by signing arbitrary messages with keys corresponding to their token holding.

What we’re missing is the infrastructure that could interpret such signals, construct interaction graphs and apply algorithms on top of them to build discovery rankings.

Having such infrastructure would allow every community of token holders to allocate attention and discover content using signals coming from the inside of the network. In other words, the token would become a frame of reference for ranking information (and allocating attention) for its community.

Our first product, the Userfeeds Engine will allow everyone to generate rankings based on blockchain-based and off-chain signals.

It’s Not Only About How Many Tokens You Have

A common feedback we receive when describing the idea is that it leads to large token holders having more voice in the community.

It couldn’t be further from what we’re building. While it’s possible to create simple rankings based purely on token holdings we expect a whole range of other algorithms that will take other signals into account.

For example, due to the open nature of the blockchain, we can use signals such as duration of holding. You could create a ranking where long-term token holders, who proved commitment to the network receive boost in rankings. In other cases, recent activity of a token holder could be a key ranking factor. The algorithm design space is infinite here.

Another interesting use case could be a ‘proof-of-donation’ based ranking. Many websites display their BTC or ETH tipping addresses, but today there’s not much incentive for visitors to contribute tips. A simple application built on top of the Userfeeds Engine could allow donors to display messages on the website as a reward for a donation.

There are many other examples which we’ll explore in the future blog posts.

Why We’ve Chosen To Raise A Private Seed Round Instead of a Token Sale.

One might ask then, why did we decide to raise a private seed round instead of a public token sale if we are believers in a ‘tokenized’ digital economy.

Indeed, the token sale option was on the table, but we’ve realised that we still needed funds to build proper proof-of-concepts and fully explore the idea before running a token sale.

Fortunately, we’ve met investors who understand the space well and were excited about the problem to back us financially on our journey. We’re grateful to have them on board. Their support provides us with time and funding to make the vision more concrete before we open up to stakeholders from the community.

Once we’re confident with the early results, we will definitely explore the token sale option again.

Stay tuned!