James Lind thought that scurvy was due to putrefaction of the body which could be helped by acids, and thus included a dietary supplement of an acidic quality in the experiment. This began after two months at sea when the ship was afflicted with scurvy. He divided twelve scorbutic sailors into six groups of two. They all received the same diet but, in addition, group one was given a quart of cider daily, group two twenty-five drops of elixir of vitriol (sulfuric acid), group three six spoonfuls of vinegar, group four half a pint of seawater, group five received two oranges and one lemon, and the last group a spicy paste plus a drink of barley water. The treatment of group five stopped after six days when they ran out of fruit, but by that time one sailor was fit for duty while the other had almost recovered. Apart from that, only group one also showed some effect of its treatment.
“You have to kiss a
lot of frogs to find one prince. So how can you find your prince faster? By finding more frogs and kissing them faster and faster.” Mike Moran Do It Wrong Quickly: How the Web Changes the Old Marketing Rules, 2007 Cited by Kohavi in Online Controlled Experiments at Large Scale, 2013
“The cost of experimentation is
now the same or less than the cost of analysis. You can get more value…by doing a quick experiment than from doing a sophisticated analysis.” Michael Schrage Value-Creation, Experiments, and Why IT Does Matter, 2010
“with massive data, this approach
to science — hypothesize, model, test — is becoming obsolete… Petabytes allow us to say: "Correlation is enough." We can stop looking for models…analyze the data without hypotheses…throw the numbers into the biggest computing clusters the world…and let… algorithms find patterns where science cannot.” Chris Anderson The End of Theory, 2008
But we pay the price.
Example: search engine improvements in batch evaluations don’t always predict real user benefits. [Hersh et al, 2000] Do Batch and User Evaluations Give the Same Results? [Turpin & Hersh, 2001] Why Batch and User Evaluations do not Give the Same Results [Turpin, Scholer, 2006] User Performance versus Precision Measures for Simple Search Tasks But also see… [Smucker & Jethani, 2010] Human Performance and Retrieval Precision Revisited
To summarize: how is web
science different? • Online testing is cheaper and scalable. • Data exploration tools make hypothesis generation cheaper and easier. • But the experiments that are easy and cheap aren’t always the most valuable. • Easy to forget our biases as scientists.
Take-Aways • The scientific method
is alive and well. Big data has just changes the economics. • Cheaper hypothesis testing and generation has already been transformative.That’s why big data matters. • But we neglect the human side of scientific experimentation at our peril.