Data Mining for good - SRI Pipe Dream or Pipe Nightmare?
Sergey Brin knows where I hide my toe nail clippingsÂ and Jeff Bezos can tell me what page I'm reading in the copy of SRI magazine next to the toilet.Â Does this scare me?Â Sometimes.Â ButÂ I wonder, could the massive data bases developed by companies like Google and Amazon, and the amazing capabilities companies like these have created to mine their data, be used to foster SRI,Â support CSR and generally promote good (or at least not evil in the case of Google)?
Define âgoodâ, thinks the discerning reader, a task far beyond my capability, but some examples might be a start.Â Data mining and socially oriented marketing are already highly visible in product sales.Â If I order a green lawn fertilizer from an Internet supplier, I will soonÂ be seeing ads for environmentally friendly insect repellents.Â However, socially conscious product marketing generally seems to be limited to a handful of claims: green, recyclable, healthy, all natural, locally made, union made, or in some cases a sunnier version of a claim that the product was not made in a sweatshop.Â
A seller with information culled from Google's data could do a lot more.Â For example, anyone who tracks my visits to charitable sites (Are you paying attention Sergei? Always, all right then, I won't ask again) knows I prefer the âteach a man to fishâ approach, except in emergencies.Â Right now you could batter me with ads for âSave Haitiâ t-shirts promising a portion of the sale proceeds would go to Haitian relief and get no reaction.Â I don't need a t-shirt, just giving cash made more sense.Â An ad for a t-shirt actually made in Haiti byÂ people working in safe conditions and earning a living wage might get a very different reaction.Â More generally, a seller seeking to market a product based on a socially beneficial attribute could find buyers whose Internet habits (not just purchases, but all the data) demonstrate an interest in that attribute.Â One result?Â If more products can be marketed more effectively based on their socially beneficial qualities then more âgoodâ products will be made and used.
Data mining isn't just about product and service sales, think SRI and venture capital connections.Â Consider Goodco, a start-upÂ looking for venture capital.Â The business plan shows a few years of shaky returns, but the product is vital to the health and well being of an under served population.Â If the product is a new anti-malaria salve, GoodcoÂ can probably find Bill Gates on its own.Â If the product is small scale desalination systems that will serve remote, water short AfricanÂ populations cost effectively â well maybe Sergei can tell Goodco that Larry Ellison is a prospect.Â Larry might not be remote (well, actually Larry is remote), but Sergei's data shows he sails past a lot of tropical islets that need water and he's interested in desalination.Â Now suppose Goodco needs to recruit a new CFO who really understands how to deal with AfricanÂ governments, NGOs and engineering suppliers.Â Sergei, might you be the world's most knowledgeable headhunter in waiting?
Right now much of this type of data mining for social good is a pipe dream, orÂ a pipe nightmare.Â A thicket of national and subnational laws that protect privacy and protect consumers and investors from fraud would often get in the way, and generally for good reason.Â But Sergei has the data, he knows how to mine it and he'd probably like to be allowed to do more with these capabilities, because he could ring much value, and profit,Â from them.Â Sergei, why not take the initiative, develop your ownÂ proposals to data mine for some not-for-profit good, create enough privacy and fraud safeguards to convince a few regulators that your plansÂ are workable andÂ then put your algorithms to good use.Â If it all works, maybe you can eventually convince the regulators that the same safeguards would justify expandedÂ for profit usage of mined data, in the meantime we'll all benefit.