August 17, 2010

No the title does not refer to my fillips in their early versions but to the Gospels. In Saint John's, Peter and some friends, all professional fishermen, had worked the whole night in vain. Someone on the shore shouts to them to cast their net over to the right side of the boat. They comply and make a catch you wouldn't believe, a sign by which they recognized the Lord.

Notice this miracle does not conflict with science. By schooling, fish can easily elude a few straight folks in the vastness of what we call the Sea of Galilee. Yet they are there, somewhere and in abundance for who happens on the right spot at the right time. As a matter of fact the resulting probability distribution is quite common, though it follows neither Gauss nor Poisson.

This improbable success theory is for instance behind the business model of venture capitalists. Invest in one hundred companies and hope one of them will make you deservedly rich and famous, like Compaq did to Ben Rosen (1). Traditionally, the pharmaceutical industry has used the same approach to find new drugs. Let R&D study a thousand molecules, hoping to find the one blockbuster product to pay the bills and then some more.

In the past decade this theory has been adopted by the forces whose responsibility it is to keep the world safe from terrorism. In his passionate defense of the United Arab Emirates' planned Blackberry ban, Richard A. Falkenrath declares "monitoring electronic communications in real time and retrieving stored electronic data are the most important counterterrorism techniques available to governments today" (*).

Since terrorists try to hide within the surrounding population like fish in the water, this is not so surprising. But notice how the theory is now streched to its logical limit. To catch the one suicide bomber, one will spy on each and every resident. To some, this will smack of an Orwellian hell.

Still had Ms Dunn, the CEO of Hewlett Packard who spied on her board members to find the source of a leak, sat down together with Mr Perkins, the famed venture capitalist who resigned from her board in public disgust, they would have realized they had more in common than met the eye.

When science mimics miracles, no blasphemy is to be found. But it is bad science for its practitioners to ignore, if not cover up, some significant costs. The improbable success theory has two flaws which must be acknowledged.

The first downside has to do with the successful outcome itself. As it is not a miracle, only envy would refuse to attribute it to the superior qualities of those who so obviously beat the odds. Unfortunately the latter all too easily fall prey to hubris and use their newly found power to abuse who stands in their way, casually at first, willfully once entitlement kicks in.

Prudence alone can temper the temptation to squeeze as much profit as possible while it lasts, but who can afford to listen to prudence? Were the successful entrepreneur too frugal, her shareholders, be they private backers or public raiders, would insist she needs to help them recoup all their past investment losses and provide them with the wherewithal to multiply future failures. This model makes it inevitable that excess attends success.

This is a question, not a proof but, if its members did not depend on turning their most promising drugs into blockbusters, would the pharmaceutical industry be the target today of so many probes for unethical marketing? Mired in conflicts of interest, they increase sales by harvesting physicians' prescriptions and profiling them without their consent. They ghostwrite their scientific contributions to better control the outcome. Now Stephanie Kirchgaessner reports "[DOJ's] investigators would be focusing on international corruption in the pharmaceuticals industry for "years"" (**).

The second flaw concerns the many failures which accompany the search for an improbable success and whose associated costs are often born in part by third parties. In the case of surveillance, failures are false positives which turn honest citizens into potential criminals or terrorists. To individual costs, profiling add considerable social costs. Originally for instance all assassins were muslims (1). To hold therefore all muslims, whether then or now, to be assassins and behave accordingly is neither logically nor socially productive, to say the least.

It does not follow that failures should never occur. Errors are indeed unavoidable and even profiling has its use. The flaw is to forget to account for all costs. John Kay has just penned a column showing how a muddled understanding of economics can lead to dreadfully dramatic conclusions (***). In the case at hand the consequence will be to underfund other avenues to fish for information which may show a greater respect for privacy.

For example listening to conversations can be as deceptive as it is wasteful. Can Richard A. Falkenrath swear the Chertoff Group will never raise to the bait of a Major Martin, the World War Two embodiment of disinformation, while failing to catch "Mincemeat swallowed whole", the coded message which signaled the success of this daring plot (2)? As Paul J Davies illustrates (****), relationship mapping methods may be more efficient.

In pursuit of an improbable success, far too many justify as many failures as needed with a sleigh of hands. In one case, no matter the cost of failure to the community, it is taken to be lower than the acceptable threshold of sensitivity, a practical way of equating it to zero. On the other case, whatever the cost of failure to the investor, no success will be declared satisfactory unless it proves to be bigger still, the very definition of infinity.

Cap the rewards of success, compensate individuals for false positives. My fillips do not shy away from solutions, even if they seem utopian.

More humbly mathematics tells us multiplying zero by infinity has no scientific meaning, a gap which opens the world to other sources of truth.

Philippe Coueignoux

PS: As I was writing, Joe Nocera denounced the recent sacking by the board of Hewlett Packard of the CEO, Mark Hurd, as another masterpiece of make believe. Under some flimsy pretense, "they got rid of an executive they didn't trust and employees didn't like" (*****). While in no position to judge this analysis of H.P. board games, I notice Joe Nocera accuses Mark Hurd of "putting up dazzling short-term numbers [...] while robbing H.P.'s future". If true, this would amount to a third, clearly non sustainable approach to catch the fish, i.e. drying out the sea like the Aral one (3).

• (*) ......... Texting With Terrorists, by Richard A. Falkenrath (New York Times) - August 10, 2010
• (**) ....... US probes corruption in big pharma, by Stephanie Kirchgaessner (Financial Times) - August 13, 2010
• (***) ..... A good economist knows the true value of the arts, by John Kay (Financial Times) - August 11, 2010
• (****) ... A web of fraudulent details, by Paul J Davies (Financial Times) - August 9, 2010
• (*****) . Real Reason For Ousting H.P.'s Chief, by Joe Nocera (New York Times) - August 14, 2010
• (1) for more details, read about Assassins in the wikipedia
• (2) read The Man Who Never Was by Ewen Montagu
• (3) for more details, read about the Aral Sea in the wikipedia