- Algorithms to Live By
- The Big Short
- Fragile by Design
- Too Big to Fail
Wednesday, May 18, 2016
Tuesday, February 2, 2016
The magic of the original PageRank algorithm was that it extracted a list of elements order ranked by importance from an unordered graph.
There is - I think - a similarity with the economy at large. One of the things I want from a market, one of the the important things that markets do that, say the Soviet system, could not is represent the cumulative 'worth' of goods and services by allowing the market participants to 'vote' with their purchasing decisions for the things they think are 'valuable'.
Here's the Algorithm analogy: Imagine that you're trying to extract a list of elements order ranked by importance from an unordered graph - in this case the elements (nodes) aren't web pages, but providers of goods and services in an economy, and the edges aren't links, they're purchasing decisions.
Value (money) flows along these edges (in both graphs) and pools in the 'highest value' nodes in the system.
Here's where most attempts to use graph-based algorithms to handle search before PageRank stopped, and where the genius of PageRank stepped in to start the fortune of what is now the most valuable company on Earth. Once value pooled at a few 'high value' nodes the older algorithms stopped. It was a winner-take-all approach to search if you will. The granularity of opinion and diversity of value was lost.
But PageRank did a tiny thing that made a huge difference: along the way, as value is flowing around the graph, PageRank introduces what in the research is called a 'damping factor' a tiny little bit of value to each node in the network, a basic income of value.
The effects are profound. Instead of winner-take-all, PageRank - because the ability of nodes to 'vote' on what's actually important is never allowed to drop to zero - is able to create a highly granular representation of the importance of everything in the system, not just identify what the single most important thing is.
Let me say that again: in these two systems, the web and the economy, 'worth' is calculated by allowing 'value' to pass between nodes (providers of goods and services), and if nodes do not have a way to 'vote' by passing value (because their own value reserves are allowed to drop to zero, or they aren't allowed to choose freely where value they control is directed), the information you can extract out of the system about what's important and what's not breaks down.
This is why I'm a supporter of a basic income, not because I think everyone deserves it (maybe they do, but that's not my reason), but because I appreciate what the market provides (information about what's actually worth spending your time on), and as a software engineer I see this system as a directed graph.
We know how to handle directed graphs.
Basic income (as damping factor) would improve the quality and granularity of information the market can provide, by giving a vote to people (nodes) who don't have the ability to signal worth right now - it actually doesn't need to be a living income, really, just something to spend on what those people actually consider worthwhile (which distinguishes it from the social safety net of America in 2016, a system that give no additional information about what's actually important).
There's a related conversation to have here, one about the two potentially incompatible systems of democracy that America has at the moment (a political one where everyone has one vote, and an economic one where the votes are dollar bills), but that's a conversation for another day.