The Pretty Good way to calculate a user's influence within your web of trust
how to escape the tyranny of the social media "influencer"
The Grapevine is the name given to the implementation of the web of trust in the Pretty Good Project. Its purpose is to help you find the high quality content you seek, even when this means finding a needle in a haystack. To this end, the Grapevine calculates how much Influence any given user should have for the curation of another user’s content in some given context. This post presents an overview of precisely how the Influence score is calculated and how this method will allow us to escape the soul-killing tyranny of today’s social media so-called “influencer.”
Overview
The Influence of a user on your Grapevine, in some given context, is a function of two variables: 1) the Average Score, which is a weighted average of the relevant trust rating (“Alice trusts Bob a certain amount to filter content in the given context”), and 2) the Input, which is the sum of the weight for each individual rating. Think of Input as the “number of ratings,” where more is better, except that in the Grapevine, not every rating is weighted equally. Influence scales linearly with Average Score, as it should, because a high Average Score is a good thing. But the key to the Grapevine is that Influence does not scale linearly with Input. This is one of the central problems with legacy social media: the follower count has no upper limit, and “influencers” are rewarded for high follower count. But the goal of the Grapevine is to seek quality, not chase followers. Input, like follower count, has no upper limit, so we shouldn’t make Influence proportional to Input. The Pretty Good solution is to replace Input with Certainty, which has an upper limit of 100%. As Figure 1 shows, Certainty starts at 0% and increases quickly at first as Input increases, but then levels off at 100% no matter how high Input gets. Influence then equals that user’s Average Score multiplied by the Certainty. Using this system, the key to Influence is not a high follower count, but a high Average Score. The number of trusted ratings is taken into account but does not dominate the Influence score. Users are rewarded, not for their follower count or number of likes or for attracting more and more ratings, but for gaining the respect of the small handful of people who know them best. We now have a tool to ditch the tyranny of the follower count; to cut through the noise of the social media influencer and to find the high-quality needles in the haystacks that we have been looking for.
Motivation
The rise of the social media “influencer” encapsulates much of what is pathologic about modern day social media. Tech companies, funded by advertising, will do anything to generate clicks and attract attention. This, by definition, is what a social media influencer is good at. Our tech companies spend billions of dollars on platforms and algorithms that cater to the MrBeasts and Kim Kardashians of the world. These people are brilliant in their own way and in what they do — which is what, exactly? Attract followers. Generate clicks and likes. Figure out better ways to trigger dopamine hits. In the process, the world’s ADHD only gets worse and worse.
Arguably, this is very bad for society. Our attention is focused on frivolous topics and people. But attention is limited. The influencer economy exists to the detriment of topics, ideas and users who might be better deserving of our attention. One wonders what would happen if the next Einstein, a genius in a meaningful topic like math or physics but perhaps not a marketing genius with the skill, time or desire build a follower count, were to post a theory of everything on Twitter/X. Would you see it? Probably not. The algorithms would never even notice it. You could very well have no idea it ever existed.
The Pretty Good way to calculate influence
The Pretty Good approach to web of trust is designed to change all of that. We are building the Grapevine in the hopes that it will allow you to sort the wheat from the chaff: to see through the social media influencers who sap you attention and distract you from what’s important, and to give you a mechanism to find the Einsteins of the world. Or the Picassos, or whatever or whomever you are looking for. Almost 8 billion people out there in the world: there are many who deserve your attention but are not getting it. Lots of needles in lots of haystacks.
One of the primary functions of the Grapevine is to calculate a quantity called Influence. It is context dependent and is the primary determinant of how much attention Alice pays to Bob in some given context.
Influence is the product of two numbers: the Average Score, which is a weighted average of trust ratings by other users, and a number called the Certainty in that average score.
Influence = Average Score * Certainty
Certainty is a number between 0 and 1 (i.e., 0% and 100%), and is designed around the idea that a higher number of ratings from trusted users gives you greater confidence, or “certainty,” that the average score is meaningful. This is an idea that all of us are already used to. Consider the rating of a product on Amazon. If two products each have an average score of 4 out of 5 stars, but one is based on one review and the other on one hundred, then most of us would probably be more inclined to purchase the one with one hundred ratings. The reason, of course, is that more ratings gives us greater confidence in the average score.
In the Grapevine, the variable Input is introduced and plays a role similar to the number of ratings on Amazon. But in the Grapevine, not every rating is weighted equally; so Input is defined as the sum of the weights of each individual rating, with the weight of each individual rating determined (primarily) by the relevant (context-appropriate) Influence of the rater.
But the key is the realization that Influence should not be proportional to Input. One hundred quality ratings is meaningful improvement compared to just one; but increasing from 100 to 200? Or even 100 to 1000? At some point, the addition of more ratings becomes only marginally more meaningful.
So we require an equation to map Input into the new variable, Certainty. There are probably more than one equation that would work for this purpose, but an exponential decay seemed to us to be a pretty good solution:
In this equation, alpha is a user-controlled scaling parameter that determines how quickly or slowly Certainty approaches 100% as Input increases towards infinity. This equation is depicted in Fig 1.
As Figure 1 shows, a higher number of ratings (by trusted individuals) does translate into greater Certainty and therefore greater Influence, which makes sense. But there are diminishing returns. The incremental gains gradually level off. In this manner, the Grapevine seeks to avoid a recreation of what is arguably the central pathology of legacy social media.
Summary
The Average Score is contextual and is a weighted average of each context-based trust rating (“Alice trusts Bob 95 out of 100 to rate movies”).
The Input is a sum of the weight of each individual trust rating.
Certainty is calculated from Input using the equation above.
Influence = Average Score * Certainty.
Influence is the primary determinant of a rating’s weight.
Conclusion
The Pretty Good Project envisions a world where platforms can be managed by your web of trust rather than big tech companies. There is a lot of work to be done before this goal is reached. But once it is reached, there will be no more need for advertising dollars. Which means algorithms and platforms that cater to influencers and advertisers can fade into the background and be replaced by algorithms that are designed to serve your needs. To find the content you want.
Are you content to ransom your most precious commodity — your time and your attention — to the social media influencer? Or would you rather take control of your time and attention? Focus on the topics, ideas and people who deserve it? The simple solution described in this post may not be perfect. In the future, your web of trust will certainly find ways to improve it. But it’s good enough to get things started.