Sir Mark Walport, speaking at the Oxford Martin School. 3rd December 2013.
The focus was on privacy and personal identity. One argument was that out of all the information we have control of we choose exactly what and how much we disclose in a given situation. Given that selectivity is not a fundamental problem, should we complain when governments and companies choose what information they wish to represent each of us by? Should we complain when they choose to cross check what we tell them using independent sources? The idea that we are in command of the representation we go by is a fragile one.
It was pointed out that not sharing information could present risks to an individual. Obsessive secrecy is not the optimum, complete openness is not the optimum. Somehow a balance must be struck; each case to be judged on its merits.
practical problems:
- Consent to data sharing in exchange for access to services is now a common practice. Does the person really consent or is there coercion?
- What happens to Facebook data when Facebook goes bust? Who owns it?
- Who would insure the person who judges the right balance between secrecy and openness in a given situation?
- Is it reasonable to object when governments act on their representation of us as individuals if they have obtained that representation by combining diverse data sets but without actually asking us?
- How do you filter out false correlations? Should you even try to do so if the damage done costs less than the cost of filtering the insights?
- Is a false positive by itself harmful or is it only harmful when acted upon?
One idea for removing the harmful effect of information about yourself is to swamp the system with so much (preferred) information about yourself that no-one would ever notice the thing you want to hide. Hiding in a crowd of your own making.
Sir Mark saw no reason to worry when government institutions create a representation of you through diverse data collations and modelling. It is bound to happen. There should be some effort to validate the information they act upon. If we don’t like the result then we don’t have to vote for them.
Comment
The private citizen has no real mechanism for correcting or choosing the imputed information derived from multiple proxies and usually would have no idea that such deductions had been created, sold and re-sold.
It is only a matter of time before profile fabrication/bolstering services become widely available simply to bias the deductive modelling in your favour. It could have a big impact on your premium for e.g. life insurance, and wouldn’t actually be a deception. The insurer deceives himself.
