Re:Senators push to ditch social security numbers in light o...
by FireBeyond on Thursday November 09, 2017 @07:33PM
(#15667016)
> Insurance is acquiring similar issues, though with very different mechanisms. As statistical inference improves, insurance ceases to be insurance... Risk gets pooled across ever smaller groups. The whole purpose of insurance is to pool risk widely. IMO, this is the problem the US' health policy architects underestimated. Insurance companies know too much to be insurance companies.
Insurance companies typically want the best of both worlds, to -know enough- to pool narrowly (or often, exclude high risk), but glean the benefits from wide pooling.
In many ways, credit scores are an early example of data related problems that will become more common. Some will become serious issues in the next few years.
Credit scores are basically an estimate of credit-worthiness. The system is designed to solve the banks' problems, not consumers'. For banks, some number of false positives (creditworthy person with a bad score) is acceptable. It may narrow the market slightly, but it improves the quality of the remaining majority of customers.
This is true of a lot of statistical/algorithmic decision making. Paypal flagged me as "high risk" when I moved countries, so I can't use paypal now. For paypal, losing 1 of me to avoid 2 malicious users is an acceptable trade-off. For me, it's annoying but not a big deal because paypal is rarely the only option.
If paypal's "danger" flag were shared across the financial system, this would become a serious problem for me. It could freeze me out of online transactions entirely, maybe other things. If landlords had access to paypal's "high risk" filter, would it be worth using to filter out bad tenants? Seems possible.
The false positive problem goes from an annoyance to a fundamental rights vioation quickly, if data is shared.
Insurance is acquiring similar issues, though with very different mechanisms. As statistical inference improves, insurance ceases to be insurance... Risk gets pooled across ever smaller groups. The whole purpose of insurance is to pool risk widely. IMO, this is the problem the US' health policy architects underestimated. Insurance companies know too much to be insurance companies.
More such issues are coming soon. China seems to want a credit score system, with alarmingly ambituous and and political breadth... The Social Credit System.
Adtech companies and conglomerates (especially FB) are really hitting their stride. I wouldn't be surprised if they can score credit worthiness, insurance risk and other things using their web browser and app datasets.
Should adtech be allowed to service banks, insurance companies, real estate agencies and employers? If not, we should probably decide this now.
I don't know what the answers are, but credit score may be the place to start. I think approaching the problemfrom and identity/data protection perspective is fundamentaly flawed.