Wine tasting usually rhymes with friendship, memorable moments and harmony. It is hard not to remember the deep purple color, the unusual nose and blended flavor of a 1985 Côte Rotie Côte Brune. Wine is distinctly magic. It is a subtle hyphen between people.
Wine is also an investment. You buy wine to drink it a later. You hope that at the time you sip your wine it has become a great wine, an unforgettable one. You may by the way either drink it or sell it at an auction. If the price has gone up, you may even drink one case, sell the other and end up with more money than you initially invested. But nothing is sure. Only times will tell it seems. Winemakers are in the same uncertain situation. After the crop, (Bordeaux or Burgundy) wine has to age in oak casks for eighteen to twenty-four months. It is difficult to figure out from tasting wines in their early stage whether a great vintage is in the making or not. Hard to tell whether the business will be good or not.
Since 1978 one man has held the wine forecast podium. His name is Robert M. Parker, Jr. He is a wine guru, maybe the one and only of his kind. He spends tree months of every year tasting in vineyards. He then publishes grades for each wine he has tasted. His swishing and spitting approach has made him world famous. Every year wine amateur and wine professionals eagerly wait for Parker's verdict. Robert Parker publishes his grades in a newsletter entitled The Wine Advocate with more than 50,000 subscribers. In the late eighties Parker's dominance was challenged by one man who claimed that one cannot predict how well a certain vintage will age just by tasting it.
Princeton University economist and former editor of the American Economic Review Orley C. Ashenfelter , heavyweight of the economics profession, is the man who dared to defy Parker. He publishes a newsletter entitled Liquid Assets. For Ashenfelter, finding a good bottle of wine is as easy as an econometrics equation that basically says:
Wine quality = 12.145 + 0.00117*winter rainfall + 0.0614*average growing season temperature – 0.00386*harvest rainfall.
Mind-boggling, isn't it! According to Ashenfelter weather statistics are enough to predict the quality of any vintage. Needless to say that the exchanges between the two experts have been rather harsh: Nose against data, palate against computer.
Ashenfelter decried what he perceived to be a sense of elitism in the wine industry. He says:
“A more fundamental question arises about the motives of the early purchasers of the wines. Why have they ignored the evidence that the weather during a grape growing season is a fundamental and easily measured determinant of the quality of the mature wines? And will they continue to do so as the evidence for the predictability of the quality of new vintages accumulates.”
Robert Parker called the professor's methods "Neanderthal," "ludicrous and absurd". He once said that Ashenfelter was like “a movie critic who never goes to see the movie but tells you how good it is based on the actors and the director.” So, should you swish and spit or should you data crunch to know the future of your vintage?
The 1989 Bordeaux deeply divided the two men. Ashenfelter predicted it would be the wine of the century. Parker could not disagree more. A year after Ashenfelter predicted that the 1990 Bordeaux would be even better. Parker disagreed again. The verdict was given by wine auctions. The 89s sold for more than twice the price of Parker cherished 86s. The 90s sold for even higher prices! Crunched weather data outperformed Parker's nose and palate.
A third man has recently invited himself to the debate. Tristan Fletcher, an Artificial Intelligence researcher at University College London and co-founder of Invinio, brings support to Ashenfelter's faith in the data. He is too convinced that machines can understand the value of fine wines better than any human. He even thinks that machine learning techniques can do better than Ashenfelter's regression techniques. His preliminary results published in The Journal of Wine Economics seem to suggest that his techniques are indeed able to improve the accuracy of vintage predictions. I am sure that we will soon see other data scientists turned wine lovers promoting alternative algorithms to crunch wine data. These algorithms will be applied not only to wine but also to art, to collectibles, to any items that can be quantified one way or the other. This is not necessarily bad news. A French proverb says that when one loves one does not count. In that case one does love and one does count! One counts because it has value to know beyond Parker's nose and palate how vintages will fare in the future. In vino veritas and in data veritas!
Ashenfelter's seminal work shows how data and algorithms may leverage our nose and palate and improve our wine choices. It is very often used as an example of the merits of big data in daily life. Indeed, the very fact that wine veritas may be found in data provides evidence that data are precious, that algorithms are valuable. Businesses like Ashenfelter's or Fletcher's are born out of data. They do make money. These delightful examples should however alert us on the value we dissipate when we agree to give our personal data for free on the Internet. Yes we do get some services in return but I am not sure the value of this services matches that of our data. The data collection has reached a scale that most of us are unaware of. Personal data are no longer captured through our browsing of Internet only. Apps and sensors proliferate everywhere from our smartphones to our homes and cars. Digital vacuum cleaners silently siphon our data wherever we happen to be, whatever we happen to do, inside or outside. We have become data obedient. Swapping our data in exchange of the free use of search engines, e-mail services, calendars, applications of all sorts has become the rule to which we abide. No questions asked.
This “free against data” swap we are forced into is unfair. The data currency with which we pay is a strong currency but it trades as if it were a weak one. As Ethan Zuckerman reminds us,
“we’ve been taught that this is simply how the Internet works: If we open ourselves to ever-increasing surveillance—whether from corporations or governments—the tools and content we want will remain free of cost.”
It is unfair because nobody is willing to openly tell us how much our data are worth. The sheer size of advertising revenues is an indication of money at stake, but this is the tip of the iceberg. Nobody is willing to tell us the true price of data for fear that we may revise our opinion towards the free for data swap. University of Carolina Chapel Hill Professor Zeynep Tufekci recently made an interesting offer to Facebook founder, Mark Zuckerberg in a New York Times article entitled “Mark Zuckerberg, Let Me Pay for Facebook”. Facebook makes billions of dollars in advertising. In the second quarter of 2014, Facebook released the following quarterly numbers: 1.32 billion users, $ 2.91 billion revenue, $ 0.791 billion profit. This is a profit of about 20 cents per month per user. Tufekci suggests that she would be willing to pay 20 cents per month for the full privacy of her data1. She adds that these 20 cents could eventually subsidize Internet access in poorer countries. But guess what, this is not something that Facebook and the like want to hear. Why? Because our data are worth a lot more to them than their current ad (meagre) value per head. This is where opacity kicks in.
On top of being unfair, the swap is opaque as nobody is willing to openly tell us what uses beyond advertising will be made out of our data, by whom and for how long. Weather data can be used by a lot of different actors including wine investors and insurers. The same holds true for our personal data. They can be processed by banks, insurers, employers, schools, governments etc. and, of course, not only for advertising purposes. The monetization options of our personal data are numerous. We give all these options, known and unknown, for free. These options however are worth a lot of money. Each of us is indeed worth more than 20 cents per month and that's where it becomes scary. We are now short of our options.
Useful insights can be drawn from data. There is no doubt about that, but not at any cost. We shall not blindly consent to data despotism. Despotism might indeed be the right word. If big data delivers veritas, it is fair to ask whose veritas it is and for what purpose. In the US FTC (Federal Trade Commission) Chairwoman Edith Ramirez is wondering where big data is driving us. The FTC is in particular concerned with low-income communities and black people being discriminated against. Research by Latanya Sweeney from Harvard University shows that indeed “advertisements for arrest records were 25 percent more likely to show up on a search for a list of distinctively black names than white names.” The discriminatory bias identified in this study is only advertising related but it does raise serious issues. The problem may become worse when discrimination will affect healthcare access, insurance provision, job selection etc.
Internet used to be synonymous of greater transparency, shortening of distances and easy access to a wide choice of content. There is a significant risk that what we believed in was just the model apartment! The Internet is on data steroids. Massive virtual warehouses accommodate the steady stream of information worldwide. Data are no longer stored on our devices: they are saved in the so-called cloud. This cloud is sold to us as a personal digital attic where we can easily store and access our data. But we have to realize that this attic unlike our attic at home is not ours. Its content is no longer ours either. Ownership has been replaced more often than not by license agreements which we do not even read. We have become cargo cultists hoping that more and more bounty will freely fall from the cloud in exchange of our data. Eric Schmidt, Google's ex CEO, candidly summarizes the situation we are in : “We know where you are, We know where you’ve been. We can more or less know what you’re thinking about.”
Scary is not it? What is even more scary is that the manager of a private company is proud of this dominance and praises it in public. These days the words of Estienne de la Boétie, a 16th century French intellectual and a dear friend of French author Montaigne, resonate rather strangely:
“It is indeed the nature of the populace, whose density is always greater in the cities, to be suspicious toward one who has their welfare at heart, and gullible toward one who fools them. Do not imagine that there is any bird more easily caught by decoy, nor any fish sooner fixed on the hook by wormy bait, than all these poor fools neatly tricked into servitude by the slightest feather passed, so to speak, before their mouths.”
Estienne de La Boétie wrote this warning in 1576. His book is titled “The Discourse of Voluntary Servitude”!
Unfairness and opacity rule the Internet game: the
depth of gathered personal data and the sheer scale of their uses remain inaccessible to the ordinary person. We fool ourselves in believing that we are fishermen navigating the digital ocean but the odds are high that we have become the baits.
Free has become awfully expensive.
The “free for your data” Faustian swap is odious.