In 2007 The Economist had a special briefing on business schools. Among other things they argues that ethics classes were becoming more important, and that the programs have more drinking than learning (and used a pub crawl at Wharton as an example). I wrote a letter that was lucky enough to by published a few issues later (Economist March 31st 2007, Letters). Here is the short letter:
SIR – The Walnut Walk (which you called the “Wharton Walk”) is much more of an occasion than you described. Students do not just visit “ten bars in a single night”. The revellers are required to wear suits above the waist and boxer shorts below while being stared at by passers-by. If that does not reduce the arrogance that leads to Enron-sized scandals, then I am not sure the trend towards ethics classes you reported will help.
Yesterday I wrote my second letter to the newspaper.
The article in question talked about how effective Big Data is at making many many small changes that add up to big effects. I generally disagree. Here was the letter:
SIR-
Schumpeter repeats the standard optimism around the latest management buzzword: “Big Data” (“Little things that mean a lot”, July 19th, 2014). In my experience that optimism needs to be tempered by three often ignored drawbacks.
First, while Big Data may allow companies to target the last “20” of the 80/20, most companies have not yet finished with the 80. By going after Big Data they risk missing the core drivers of their company performance.
Second, when shooting for 1% improvements there is a significant risk of chasing noise. Normally a 95% chance of “Significance” is considered ‘real’, but it means there is a one-in-twenty chance your results will not be ‘true’. With the size of Big Data, thousands if not millions of interactions can be tested, leading to a plethora of false positives.
Finally, Schumpter makes the assumption that big effects can be achieved by layering on many tiny changes. This assumes the effects are additive. In practice most things in business interact with each other. These “interaction effects” can cause the 1% effects seen in isolation to change significantly in magnitude (or even direction). This can be overlooked when the effect is big, but when it is tiny you may find yourself optimizing for a world that doesn’t exist anymore.
Big Data is a wonderful marketing slogan, but most companies would be well served to invest in something more old fashioned. I call it Analytics.
I will expand on all three of those arguments going forward. In the meantime, if you haven’t already, you might want to check out my older post on the (lack of) value of Big Data.