Trust is all. I listened yesterday to some very smart people talk about transparency, data, government and even how predictive analytics might replace investigative journalism. Each of the very smart talking heads (writer Steven Baker, technology expert Stephen Brobst, transparency expert Steve Horn from Dow Jones) acknowledged that ultimately to be able to effectively use this data, the keepers of the data, or The Numerati as Baker calls them, will also need to deliver trust. (This is at a luncheon sponsored by our client, Teradata, at their annual users’ conference, in Washington, DC.) Trust drives data-based decision-making, not only in being able to believe the information that accumulated data delivers but more importantly, in gaining the acceptance of real human beings to accept the conclusions that data predict.
Baker discussed what impact business intelligence, when used pro-actively, could have when applied to solving large-scale problems like the current crisis in health care… a huge topic these days in Washington. But he also ventured, somewhat off-handedly, a view on his personal situation as a long-term and respected journalist now facing the sale of his magazine, BusinessWeek (for whom we at SMT have had a client relationship) to Bloomberg. He sees the kind of predictive quality of intense data analysis eventually taking the place of investigative journalism.
Which brings us to government, and the odd position that people like our panel take when they espouse the belief that data mining will also shape government policy, the frequent target of the best journalist watch-dogs. I’m not one to defend traditional media business models, but who will be the watchdogs if the watchdog is an algorithm? Steve Horn partially addressed this issue by speaking about how comments to existing data, if it is transparent, could act as a curb to legislative blunders and excesses.
The panelists were particularly focused on how healthcare decision-making could be made more specific to individual behaviors and thus more predictable, and without the redundancies that lead to cost. Fair enough. But I couldn’t help but wonder how the whole mistaken controversy over “death panels” could be twisted into some new form of propaganda when machines determine such things as my likelihood to contract a lifestyle disease and consequently, some greater tax penalty. The problem, dear Brutus, is not in our data but in ourselves, and that brings me to the trust issue. While in theory reinforcing positive behaviors based on data works with selling you botox if you’ve indicated to facebook that you’re over 50, I’m not sure, yet, what incentives to good behavior might work in health care if they are in turn associated with penalties for bad. The Oregon case study conducted with a grant from Intel of the “healthy aged” was mentioned as an example of pro-active health modeling, and Brobst has obviously thought through a “roll-out strategy” involving influential(s) (Cleveland Clinic, Mayo — a thoughtful approach for whatever health “plan” predictive analytics might suggest.
Steve Horn admitted that currently there is no infrastructure in place to build “trust” — which he defines as getting information that is accurate, timely and complete, although he favors passage of HR 1242, better known as the Transparency Act, which aims to make TARP data more available faster. It’s easy to use shopworn clichés about genies and bottles when it comes to making data-based decisions for life and death issues, but it’s safe to assume that no amount of transparency alone can address the kind of breakdown in public trust that characterizes the current debate.
