Data Science Texts

Diversions

NB: We may earn a commission if you buy something via an affiliate link.

The Largest US Companies Have Highly Connected Boards.

Click a company to see its closest connections; click the background to unselect; you can use regular text search to find a company stock ticker.

Shameless plug: If you enjoyed this graphic, you might also enjoy The Wizard of Lies: Bernie Madoff and the Death of Trust, a best-selling account of Bernie Madoff's Ponzi scheme.

Data Notes

Synopsis

This visualization depicts the board connections of SP500 companies with market capitalizations of more than $25 billion; circles are companies and lines are board members. Market capitalization is proportional to the area of company circles. Data is current as of August 15, 2021. Some potential data issues are described below.

Visualization Issues

There were a lot of problems putting this vis together.

Initially it was supposed to be all the S&P 500 companies, but this leads to problems immediately.

  • Some companies are in the S&P "500" multiple times because the index includes multiple share classes. This causes problems because the the board is the same for both classes so one gets weird clumps in the graph. Duplicate share classes have been removed.
  • The vis is too cluttered (particularly on a phone) with 500 companies. That's why it's filtered to those with market caps larger than $25 billion. It's amazing how small a $25 billion dollar company looks when thrown into the same chart with Apple, Google, etc.
  • Even with the filtered companies, it can still be hard to see which is connected to which. That's why the click mechanic was added. It could use some improvement though. Companies/company names are hidden if they are not connected to the clicked node, but the node is still there (it's just transparent). That's why one may click on a node and an 'overlapping' node gets selected instead.

There are some potential missing/extraneous links in the graph. Links are created by matching company board member names. This means that if a different version of someone's name is present for different boards, a link could be missing. There could also be an incorrect link if two people share the same name. This shouldn't affect the overall graph too much, but could lead to errors in some individual cases. Alas, I was unable to find board members' DNA sequences to perform this match properly. Even if such were available, there is still the possibility of clones….

Some companies do not have a connection within the large companies of course. They're kept in the vis for the sake of those searching, but it is annoying that they tend to clump in the corners.

Data Sources

Public companies publish their board members and market caps are widely available. This is the data the chart uses.

Acknowledgments

This visualization was created using R and D3. The following packages were particularly helpful: