"250 Years of Clever Counting"
Stephen Ziliak emails:
Only moralists and economists know that Adam Smith's Theory of Moral Sentiments (1759) turned 250 years old this year.
However worthy an Adam Smith party, I thought you'd like to know about another party and sentiment, "Arthur's Day" - the Guinness Brewery's 250th birthday party - to be celebrated Thursday, September 24th, all over the world:
My article, "Great Lease, Arthur Guinness - Lovely Day for a Gosset!" (prepared for a special "Beeronomics" issue of the Journal of Wine Economics), shows how clever counting at Guinness did not stop two and a half centuries ago when Arthur signed a 9,000 year lease for the brewery, house, and land at St. James's Gate in exchange for 45 pounds a year (nominal, not inflation adjusted)!
"The great innovation in statistics in the era after Galton and Pearson was made in the private sector of the economy, between 1904 and 1937, at Guinness's Laboratory, to the end of improving, however gradually, production of a consistent beer at efficient economies of scale" (Ziliak 2009, p. 4).
Here's the article "Great Lease, Arthur Guinness—Lovely Day for a Gosset!":
Abstract: Small sample theory—the great innovation in statistical method in the period after Galton and Pearson—was ironically discovered by a brewer during routine work performed at a large brewery, Arthur Guinness, Son & Company, Ltd. For four decades William S. Gosset applied small sample experiments to the palpable end of improving, however gradually, the production and control of a consistent unpasteurized beer when packaged and sold at efficient economies of scale. Introducing, "Guinnessometrics." Annual output of stout at Guinness’s Brewery may have topped 100 million gallons but Gosset’s scientific knowledge was built one barleycorn at a time; in fact, the inventor of small sample theory worked closely with botanists and breeders. In the process, the brewer, William Sealy Gosset (1876-1937) aka "Student," an Oxford-trained chemist—though self-trained in statistics—solved a problem in the classical theory of errors which had eluded statisticians from Laplace to Pearson. In addition, though few have noticed, Gosset’s exacting theory of errors, both random and real, marked a significant advance over ambiguous reports of plant life and fermentation asserted by chemists from Priestley and Lavoisier down to Pasteur and Johannsen, working at the Carlsberg Laboratory. Central to the Guinness brewer’s success was his persistent economic interpretation of uncertainty, what Ziliak and McCloskey (2008) call the "size matters/how much" question of any series of experiments. An enlightened change in Guinness human resources policy gave an incentive structure that also seems to have nudged "Student," who rose in position to Head Brewer, to find a profit when the opportunity knocked. Beginning in 1893, Guinness vested "scientific brewers" such as Gosset with managerial authority. In fact Gosset was at times involved with price negotiations over hundreds of tons of barley and hops—perhaps hours or minutes before he ran (that is, calculated) a regression on related material. In brewing circles William Gosset is remembered less nowadays than he might be. He did not give two cents for arbitrary rules about statistical significance—at the 5% level or any level arbitrarily assumed. How the odds should be set depends on the importance of the issues at stake and the cost of getting new material, he said from 1904. Yet even in brewing journals, both academic and trade, and for the past 85 years, statistical significance at the 5% level continues to draw its arbitrary line segregating a meaningful from a non-meaningful result, a better barley from a worse.
Posted by Mark Thoma on Sunday, September 27, 2009 at 10:35 AM in Academic Papers, Economics |
You can follow this conversation by subscribing to the comment feed for this post.