Narrative as Analysis, Analysis as Narrative
Karl T. Muth
CEO, FRST Corporation
Long Ago at UChicago
One of the courses that changed the path of my life was 33111, Microeconomic Analysis of Policy Issues, with Gary Becker and Kevin Murphy at the University of Chicago. Becker had recently published a now well-known op-ed about organ donations and began the lecture with something I’ve always remembered and will paraphrase here: We [economists] draw conclusions from analysis, but most people draw conclusions from stories.
In other words, no matter how rich, or compelling, or unexpected the conclusion the data endorses, it won’t connect with most audiences (even sophisticated audiences) without good storytelling.
Much of what we do at FRST involves taking data from blockchains like Ethereum, exchanges like BitMEX, and partners like Crunchbase and feeding it all into a FRST narrative loom that allows our user to weave this data into a story. Narrative adds to our customer’s insights but also his or her persuasive efforts. What Becker’s nugget of wisdom has hidden within it is yet another nugget: your conclusions, even if groundbreaking, are meaningless if you can’t communicate them. You must use the data to discover insights, but you then must use those insights to persuade one’s boss, colleagues, readers, and constituents. An uncommunicated insight is not impactful, and stories are more impactful ways to communicate than hard-to-digest datasets or single observations.
If you ask an economist about the Great Depression or the Great Recession, he or she will likely begin with an anecdote or story. This is not because the person is technically illiterate or unfamiliar with the events; some of the statistics at the core of the events are either unapproachable technical (8% reduction in real dollars in government expenditure during the Hoover Administration) or simply unimaginable in the modern context (24.9% unemployment rate in 1933, a GDP drop from $104B to $56B between 1929 and 1933). Good analysis — whether economic, financial, anthropological, ethnographic, or historical — involves storytelling.
And what most of our customers do is use data to tell a story they hope stretches past t0 and into the future. They hope the past holds predictive power for the future.
LeBron as Storyteller
To explain the enormous difference between data from which a conclusion can be drawn and a narrative the reader/listener/recipient can digest as a story, consider the first few minutes of the fourth quarter of Game 1 of the 2018 NBA Eastern Conference Finals. Don’t recall? No problem, have a look at the data below.
https://www.basketball-reference.com/boxscores/plus-minus/201805130BOS.html
Or, watch a minute of this video starting at 6m44s (link already cued up to 6m44s mark):
https://youtu.be/VyASvbL_FdA?t=404
Why is LeBron’s narrative of what happened so much more compelling? It’s not just because his memory is impressive or because he treats basketball like he’s engaged in writing a participant ethnography. It’s because LeBron strings together key data-points, events, and descriptors into an approachable narrative.
To develop this deep narrative at FRST, we have to establish the relationships between actors. Like the series of passes that LeBron describes, FRST’s graph database shows who sent value to whom, in what order, and in what amount. To understand why this is such a revolutionary visualization, we have to look back in time to Boston in the late 1930s when a Harvard student named William Whyte — working less than 100 yards from what would decades later be Mark Zuckerberg’s dorm room — coined the term “social network.”
Below is a graph diagram Whyte created in 1937 to illustrate who in Boston’s North End Norton gang enjoyed influence over, or trusted who.
Above: An early social network graph diagram. From W.F. Whyte, Street Corner Society (University of Chicago Press 4th ed. 1993).
The early graph database approaches of FRST in 2016 and early 2017 were not much different from this 1937 illustration. We looked at the personification of wallet addresses on the network and then attempted to establish which parties dealt with one another, who enjoyed the most influence/trust/credibility on the network, and how these differences in position and status influenced financial transactions.
Over time, FRST’s graph database grew to encompass a hot copy of the entire blockchain (every transaction, every block), and the annotations grew more nuanced, eventually including categorical personifications like “this is a trading floor” or “this is a mining site” or “this is an exchange.”
The reason Whyte’s work and our work are similar in that both provide rich context that allows its reader to interpret complex data quickly, discovering insights. Imagine walking around the streets of 1930s Boston; if I told you “that’s Joe,” and “that’s Angelo,” you’d know something about those people.
For example, simply looking at the diagram above, if I told you “Joe ordered Angelo to kill Danny,” you’d question the veracity of my story.
Both Danny and Angelo are senior to Joe in the gang, and Joe has no direct line of communication or influence to either Danny or Angelo. Further, Danny enjoys direct communication with, and hierarchical influence over, Angelo; it seems unlikely Angelo would obey this order, and it seems Joe is in no position to give it. Despite knowing nothing about Boston’s North End street gangs in the 1930s, armed with this simple social diagram you can quickly and accurately challenge the Joe-orders-Angelo-to-kill-Danny narrative.
FRST’s Offering
Our clients are searching for relationships between people and actionable insights, but on a high-performance trading floor rather than in a depression-era Boston slum. By personifying network locations and quickly illustrating the history of relationships between wallets, exchanges, mining operations, and other key landmarks on the blockchain landscape, we empower our customers to understand the blockchain as a narrative rather than as a collection of scattered, hard-to-unite data points. Whether event analysis, backtesting, or scenario work, looking retrospectively at transactions among parties is really about testing stories.
A story that makes sense can be an insight. That insight can lead to a strategy, which can then help inform a trader or suggest a group of assets deserves a closer look. That group of assets has a biography that can be built, tested, and understood; some subset of those assets may be thrown into a bucket worth testing against actual or hypothetical market events — alongside dozens of other buckets. The buckets that survive might mature into portfolios, and further testing may reveal or refine which rules should apply to those portfolios’ management.
This is what people do at funds across the world every day, and there is no reason it should be harder when the assets are in a digital environment rather than an agricultural or petrochemical one. In the case of digital asset networks, we show customers the “what,” the “where,” and the “who” and give them tools that can help them discover the “how,” and, eventually, the “why” behind every transaction.
We are proud to help our customers discover narrative in the noise.
Find the story, test the story, tell the story. Do it with FRST.
Want to Learn More?
To learn more about our capabilities and where we fit into the blockchain ecosystem, please see my recent interview at MIT Technology Review or my discussion of these issues at IBM. To see what wallet-level analysis looks like using our tool, please see an excellent Medium post from Patrick Doyle, who is on our data science team. To see what event-level analysis looks like, contact me at karl@frst.com for a copy of our EOS-ETH rolling crowd-sale case study. To learn more about what’s going on with me and what I’m excited about, say hi on Twitter at www.twitter.com/karlmuth or @karlmuth.
