Word Clouds of Buffett’s Letters

Here are word clouds I created to visualize words Warren Buffett used most frequently in two of his famous letters to Berkshire Hathaway shareholders, the first based on his newest letter (2012, released Friday) and the second based on his oldest (1977).

The word “billion” has (inevitably) replaced the word “million;” Charlie (Munger) has assumed a preeminent position; acquisitions matter greatly now but not then; insurance float matters more in 2012 while insurance underwriting mattered more in 1977; BNSF and GEICO are big today, along with newspapers, not the textile company or trading stamp business as was true back then.  Quite a few other changes should be obvious as well.  Among the similarities: the centrality of earnings to discussions of corporate performance (particularly as compared to cash flow or dividends).


Buffett 2012 Letter Word Cloud



Buffett 1977 Letter Word Cloud














Over the past 18 years, I have prepared several editions of The Essays of Warren Buffett: Lessons for Corporate America, the latest of which will be released later this month.  In combing through the raw letters, which now add up to 400,000 words, I identified organizing principles and fundamental themes.  I tried to provide a collection, now numbering 120,000 words, that is faithful to the original.

In addition to substantive assessments, I sometimes conducted word counts using primitive word processing tools.  I also used as aids a grid showing where in my collection various original excerpts appear (and always put a variation on that tool, called a disposition table, at the end of the book).*  

This year, for the first time, I could also play around with word clouds, applying software to documents to visualize word count.   The visualized word density can be amazing when looking across the whole collection, comparing decades, and slicing in other ways.  The 1977 to 2012 comparison is one example. 



* The disposition table in the 2013 edition appears as follows:

You may also like...