Epsilon Theory Discovery Map

January 4, 2019·5 comments·epsilon theory archive

Please Note: For now, this will likely only be a desktop-only feature, so please access next time you're at your laptop or workstation.

We subject all other publishers and media to the Narrative Machine, but what about us? Yes, we are missionaries, too. In that spirit, we've created a Narrative Map of all the notes and briefs we've ever written, going deep through our full six-year archive.

If you've been reading certain notes and want to go down the rabbit hole to find some notes telling similar stories in different ways to help grasp a concept, this will show you the way. If you want to see the notes which connect the various civic, political and investment topics we cover, this will help you do that, too. Want to see what the core Epsilon Theory narratives are? Want to see how we connect the ideas of individual sovereignty, big data, political polarization and missionary behaviors in media? Look for the intersection points and connections (lines) between clusters.

Just click on any node and it will take you to one of our 300+ notes or briefs.

As you navigate through this map, you'll find how faded narratives return with a similar rhyme. See that Game Theory node embedded in the midst of the Central Bank Omnipotence cluster? That's Ben's recent "Lord Make me Chaste...But Not Yet" brief.

If these maps we produce using software from our friends at Quid is unfamiliar to you, a couple of quick reminders.

Up, down, left and right have no objective significance here. It's all about connectedness and proximity.
Proximity is a good guide for similarity in the use of language. This is true within clusters, across clusters, and between the clusters themselves.
Adjacency, which is expressed with a connected line within or between clusters, is usually an indication of multiple identical uses of language. Look for these to find the most closely related topics.
Colors reflect clusters, a range of articles that are most closely connected to one another

We'll be shooting to update this every month or so. You'll find that the clusters themselves begin to shift meaningfully as even a few new pieces of content are added for context. But for now, enjoy going down a few rabbit holes.

epsilon theory archive

Comments

JaneyVeeabout 7 years ago

This data visualization makes my right brain very, very happy. Thanks, Rusty.

rguinnabout 7 years ago

Glad to hear it, Jane!

mataylabout 7 years ago

Thanks for the explanation of the narrative map. There is something I don’t understand. You mentioned “adjacency”, “proximity” and “color”. Are these variables all orthogonal to each other? In other words, is each of the three variables independent of the other two?

I assume the label for each cluster (color) is defined by a human, not the machine. Is that right?

You mentioned the “Game Theory node embedded in the midst of the Central Bank Omnipotence cluster”. I assume that that node is not connected to any Central Bank Omnipotence node? Given that it is embedded in the Central Bank Omnipotence cluster, what is it specifically that makes it part of the Game Theory cluster?

bhuntabout 7 years ago

Thanks for the good questions, Matthew.

“adjacency” and “proximity” are roughly the same concept, although adjacency is a matrix term and proximity is a visualization term. In this particular case, node color is also a visualization quality used to communicate cluster identity (so yes, similar again to adjacency and proximity), but color can be used (and is probably better used) to visualize a totally orthogonal dimension like recency or sentiment.
The latest version of Quid labels clusters based on an analysis of shared most-common words, but yes, these cluster labels are our own interpretation of the cluster meaning.
All nodes (or almost all nodes) are connected to every other node (or almost every other node) in the underlying matrix-space. The visualization you’re seeing is a flattening of all those matrix dimensions into two dimensions, which sometimes results in a node from one cluster overlapping or sitting within nodes from another cluster. It’s an artifact of how the visualization algo compresses multiple dimensions of connectivity.

mataylabout 7 years ago

Thanks Ben.
Sorry to keep pestering you. Ignore me if it gets to be too much.

I looked up the Adjacency Matrix on Wikipedia, so I do understand adjacency now.

You said that the underlying matrix space is N-dimensional, where N is very large. I presume that each dimension represents a [shared common] word? Then a link between node A and node B in dimension n of N would represent A and B both using word n, right?

Does the distance between node A and B in dimension n somehow represent the similarity in the number of times word n is used in A wrt B? So that if A and B both use word n the same number of times then there is zero distance between them in dimension n. Just a guess. Is that right?

What does the diameter of node A represent?

Thanks in advance.