What makes a human important? Their humanity, sure. But what makes you REALLY important? What would a balanced jury of your peers pull out about your life?
Maybe you try really hard as a parent. Or were part of the making of a product. Maybe you shook up the world in public or maybe just had some particularly happy moments with a few.
Emily Dickinson hardly left her house. And spent the last two decades of her life refusing visitors. But in the end left an indelible mark on literary history.
*Inherent importance of life aside, there’s a potential underlying structure to what we value in all these scenarios. And it’s readable by an AI. *
Semantic triples follow the structure of subject — predicate — object.
“Steve graduated from Harvard”
“Sam is 37”
“Marissa Mayer was the CEO of Yahoo!”
“My mother is a skilled flautist”
These are all semantic triples. And the act of drawing inferences from them is a veritable gold mine of linked data when done at scale.
This structure is what provides the underlying organization of a knowledge graph. For simplicity’s sake, you can think of knowledge graphs like a relational database. But basically they’re comprised of nodes (entities), and edges (relationships between entities).
Where most databases historically have been structured to retain the structure of each individual entry (think a row a spreadsheet), knowledge graphs are structured around the relationships between entities. This relationship-first
structure has long been coveted as a cornerstone of the semantic web. And today we’re just seeing these fruits bear out at large through tools like Siri, richer search results, data enrichment tools, and more.
There are probably two public knowledge graphs of particular note. Google’s Knowledge Graph is perhaps the most well known and commonly used. Diffbot’s Knowledge Graph is the largest and most accurate knowledge graph sourced from the public web.
There’s no public end point for consuming all of the relationships in Google’s KG data. So for the purposes of this exploration we used the data from Diffbot’s KG.
*So what does this have to do with importance? *
As previously mentioned, we tend to think individuals are more or less important based on how many lives or entities they’ve touched. And in turn how important those lives and entities are. Whether by proxy (making a product, or a poem), or in person (being a boss, or a friend, or attending something).
The relationship-first nature of knowledge graphs does a good job at representing the way we actually view the world. And one factor present in Diffbot’s Knowledge Graph is an “importance” score for each entity. This is basically used to determine who you’re likelier to mean if you inquire about apple. Do you mean Apple Inc. or the fruit?
Apple Inc. has millions of connections (“edges” in knowledge graph speak). News mentions, many employees, investors, products, reviews. Sure apples are popular. But in the context of a Knowledge Graph centered around organizations and
people, you‘re probably after Apple Inc.
And keep in mind that the Knowledge Graph is sourced from the public web. In essence an AI built to read web pages and infer facts. Surely there are many books out there about apple farming. But that’s not a huge portion of the web.
*So what can we learn from the 10k most important people (“MIPs”)? *
*No names are named here. But what does it take to have more connections than nearly anyone in the world? *
Education
As one would likely expect, certain schools are outsized pipelines to influence.
Looking at the most commonly attended schools in this cohort, the following are likely to be present more than once in every 200 MIPs.
In particular:
- Harvard University — 1 in 14 MIPs
- Stanford University — 1 in 28 MIPs
- University of California Berkeley — 1 in 32 MIPs
- Massachusetts Institute of Technology — 1 in 52 MIPs
- University of Pennsylvania — 1 in 64 MIPs
- Columbia University — 1 in 85 MIPs
- Yale University — 1 in 88 MIPs
- University of Chicago — 1 in 110 MIPs
- University of Cambridge — 1 in 124 MIPs
- Northwestern University — 1 in 124 MIPs
- University of Oxford — 1 in 127 MIPs
- Cornell University — 1 in 162 MIPs
- University of Illinois — 1 in 165 MIPs
- UCLA — 1 in 191 MIPs
- Brown University — 1 in 196 MIPs
Our 10,000 MIPs attended a total of slightly over 3,000 schools. 65 of these schools were attended by over 30 MIPs each. And the top handful attended by hundreds of MIPs.
65% of total MIPs did not attend these 65 premier schools, however. And a small handful did not attend higher education.
A cluster of pre-collegiate schools also surfaced. For individuals where their pre-collegiate training is listed online.
**Roughly 1 in 200 **of our MIPs attended Eton College (British prep school).
Roughly 1 in 375 of our MIPs attended the Bronx High School of Science.
Roughly 1 in 1000 of our MIPs attended the following high schools:
- Phillips Academy
- Horace Mann School
- Berkeley High School
- Phillips Exeter Academy
- Gaithersburg High School
And roughly 1 in 3000 of our MIPs attended the following:
- Stuyvesant High School
- Greeley Central High School
- Towson High School
- Horace Greeley High School
- Saint Ignatius High School
- Beverly Hills High School
Internationally, clusters were less extreme. But the most common non-American universities attended by our MIPs included:
- Cambridge University
- Oxford University
- INSEAD
- London School of Economics
- Imperial College London
- Hebrew University of Jerusalem
- University of the Witwatersrand
- Tel Aviv University
- University of Western Ontario
- University of British Columbia
- Indian Institute of Technology
- London Business School
- University of Waterloo
- National University of Singapore
- HEC Paris
- University of London
- McGill University
- University of Manchester
- University of Capetown
- University of Taiwan
- University College London
- King’s College London
Skills
At the end of the day, education will only get you so far. In our
hyper-specialized economies there are many ways to get ahead. And many problems worth solving. Let’s take a look at the most common skills our MIPs possess.
In total, our 10k MIPs have listed or attested to roughly 6,000 unique skillsets, suggesting a sizable amount of overlap.
If you had to guess one single skill that is most prevalent among these individuals, you probably wouldn’t get it. Not even on a multiple choice test.
*The single most common skill attributed to our 10,000 MIPs is teaching. *
Of every skill attributed to the MIPs, one out of 55 is teaching. That might not be quite what you expect from our empire-creating cadre. But in a larger cluster of human-related skills it starts to make more sense: teaching, management, leadership, human resources management.
Add to that that a large portion of the individuals in question hold advanced degrees and at one point were university TAs, and perhaps the number isn’t that surprising.
In descending order, the 50 most common skills attributed to our MIPs include:
- Teaching
- Economics
- Management
- Marketing
- Supply Chain Management
- Start-ups
- Strategy
- Sales
- Entrepreneurship
- Leadership
- Law
- Mass Media
- Human Resources Management
- Software Development
- Business Development
- Cloud Technologies
- Strategic Partnerships
- Product Management
- Content Management Systems
- Writing
- Public Speaking
- Advertising
- Mathematics
- Social Media
- Venture Capital
- Mergers and Acquisitions
- Research
- Mobile Technologies
- User Interface
- Ecommerce
Working through the entire list of skills, three clusters appear:
finance-related skills, engineering-related skills, and marketing or public-facing skills.
The top finance-related skills include:
- Economics
- Venture Capital
- Mergers and Acquisitions
- Investing
- And Fundraising
The top engineering-related skills include:
- Cloud Technologies
- Mobile Technologies
- Enterprise Software
- Networking Technologies
- And Robotics
The top public-facing skills:
- Marketing
- Sales
- Mass Media
- Public Speaking
- And Online Advertising
**A large majority of MIPs also specialize. **While a cluster of skills are shared by many MIPs (as in the illustration above), a majority of skills are one-offs, shared by no or very few other MIPs.
While there are too many specializations to list, to exemplify the range of industries and competency areas represented, a random sample is presented below.
- Union negotiations
- eSports
- Phytochemicals
- Quorum Sensing
- Essential Oils
- Federal Budget Management
- Printing Solutions
Location
While we’ve just witnessed the year of remote work, location still matters. Particularly in networking-heavy, governmental, research, and capital-intensive industries like manufacturing, MIPs tend to cluster.
In fact, while many of these individuals have undoubtedly worked remotely for at least part of 2020, only 1 in 100 have listed remote working as a current or past job location.
Our 10k MIPs are listed as working in a total of 1,800 locations throughout their lives. Considering there are over 4,000 mid-sized cities in the world, this suggests a definite clustering. The most recent location listed for each of our 10k MIPs lowers this number to around 600 cities, with only 36 cities hosting more than 1 in 250 of our MIPs.
*Of MIPs located in the top 100 MIP-hosting locations in the US, 1 in 3 are cities in California, 1 in 6 are in New York, and one in 15 in D.C. No other locations come close. *
Beyond large financial, research governmental, and technical hubs, noteworthy small clusters include well-known university towns throughout the United States and Europe.
Additionally, there are definite “stepping stone” locations among MIPs. These are past locations associated with MIPs. And this range of locations pulls in a range of university towns with the leading few including:
- Cambridge, MA
- Stanford, CA
- Berkeley, CA
- Princeton, NJ
- Oxford, UK
- New Haven, CT
- Boulder, CO
- Ann Arbor, MI
- Evanston, IL
Job Titles
Most large scale impact by MIPs is derived from their work. And while MIP work is at the end of the day very wide ranging, definite clusters appear.
- More than 1 in 8 MIPs work in computing or information science roles
- More than 1 in 8 MIPs work in finance-related industries
- More than 1 in 10 MIPs work in software-related industries
- More than 1 in 20 MIPs work in health care-related industries
For job titles, many MIPs have accumulated quite a number through the years, and
hold several simultaneously.
**The single most common job title of our MIPs was board member. **Though many of these individuals also lead or help lead their own enterprise.
As one might expect, the top handful of job titles for MIPs
include:
- Board member
- Chairman of the board
- CEO
- Founder / Co-Founder
- Owner
- Executive Director
- Chancellor
- And Partner
Roughly half of all current jobs held by MIPs were some derivation of the above titles. For the other half, an exceedingly diverse range of titles emerges. A sampling includes:
- Angel Investor
- Lobbyist
- Chief negotiator
- Journalist
- Philosopher
- Governor
- Attorney General
- General
- Bass Player
- Chief Scientist
- Author
- Producer
- Senator
- Rector
- Evangelist
- Bishop
- Head Coach
So what have we learned?
On one level the public (in this case facts from the public web) visibility of individuals will never capture a truly holistic vision of “important” people. Importance is subjective in and of itself.
But the ability to structure and quantify relationships at scale is new. Particularly from otherwise unstructured natural language and visuals from around the web.
This quick illustration validates many things one may have already known. Power and influence cluster. Education matters. There are a few ways to gain large levels of influence, and they tend to revolve around public service, being the best in a particular niche, building a company, or owning things. And this seems
to align with a common sense view of who would realistically be able to change a large number of lives. Or have more “touch points” with the world.
Top comments (0)