5 ms·

Let's not forget that what's even more profitable than 'pricing insurance according to actual activity' (which is a nicely sounding PR cover up) is 'dropping pe

Let's not forget that what's even more profitable than 'pricing insurance according to actual activity' (which is a nicely sounding PR cover up) is 'dropping people with early symptoms of a disease before it gets found by their doctor'. Which is unethical, but legal (right?) and extremely, extremely profitable. Keep all premiums, drop people with cancer before they find out about it. And yes, if it's possible to infer pregnancy from shopping bills, then inferring a early health problem from a detailed 24/7 tracker data will be much easier. And if you want to opt out of this? 'Here are our rates without a fitness tracker discount, they cost 5 times more, ok?'

m_mueller 9y ago
.. which is why a government guaranteed insurance with good overall coverage is necessary. severe market defects in health care furthermore make dictated prices for services/prescriptions necessary. Adjusting Obamacare for that you end up with the Swiss system - still expensive bur at least effective and IMO fair. Thinking about it, increasing information asymmetry due to machine learning applied to marketing may make heavy regulations necessary in most other sectors as well. Imagine software that tracks every person and figures out what they need when, identifies a monopoly of opportunity for any given good and them jacks up the price just for that person individually - in real time, while the person walks up to it. Maybe that's how capitalism ends. Slavery 2.0.
- specialist 9y ago
  "Imagine software that tracks every person and figures out what they need when..." aka the status quo. People are profiting from my data. I want my cut. Screw privacy. I want my cut. Any one storing or using my data for any reason owes me money. Sell my address to a direct mail campaign. I want my cut. Need my health history for some longitudinal study that may ultimately save lives. Awesome. Pay me. Use my purchase history to recommend a new pair of shoes or sell ads. Woot! Pay me. Micro target my snowflake demographic for targeted political agitprop (aka campaigning). That's worth at least $50. Pay me.
  - m_mueller 9y ago
    Have you looked at the BAT project? Seems to me that something along these lines could bring such a vision closer. As in, your browser blocks tracking / identification by default (integrated TOR?). Sites can offer you a bit of tokens if you agree to share your identity / if you let them track what you're looking at. It's a bit of an extension of what BAT aims for, but I think their vision goes in such a direction. I could see such a thing taking hold, just like AdBlock Plus and co. have become a thing that advertisers / publishers need to work around.
    - specialist 9y ago
      Do you mean Basic Attention Token? https://basicattentiontoken.org https://basicattentiontoken.org Looks interesting. Thank you.
      - m_mueller 9y ago
        Yep, that's the one. Disclaimer: I hold some of those ;).
laydn 9y ago
This should quickly be declared illegal, if not already. With the amount of data that is being gathered on us, coupled with advances in machine learning and statistical analysis, in the near future, "they" will be able to figure out high risk scenarios before we even know about it. If they drop coverage or increase premiums based on this, they are no longer an "insurance company". I don't know what we would call them.
- chimeracoder 9y ago
  > This should quickly be declared illegal, if not already. Dropping coverage like that is illegal already. > If they ... increase premiums based on this, they are no longer an "insurance company". I don't know what we would call them. Quite the contrary - an insurance company, by definition, is expected to price insured entities based on the expected future payouts (ie, the levels of risk). More specifically, the sum of all expected future premiums that a patient pays should be slightly more than the sum of all expected future claims for that patient. If you're expecting it to behave otherwise, then you're looking for something that isn't insurance - insurance is, by definition, a model that relates risk and price.
  - laydn 9y ago
    Shouldn't that read: "" More specifically, the sum of all expected future premiums that all patients pay should be slightly more than the sum of all expected future claims for all patients "" You can not reduce the calculations to a single patient. If an insurer finds out that you will develop a certain type of cancer for which median treatment cost is 1M USD and lifetime support costs will be 30K USD/year, they should not be allowed to increase your premium accordingly.
    - chimeracoder 9y ago
      > Shouldn't that read: "" More specifically, the sum of all expected future premiums that all patients pay should be slightly more than the sum of all expected future claims for all patients "" No, it shouldn't. I mean, the thing you wrote is a true statement too, because of the distributive property of summation, but the point is that insurance does work on a per-plan level. If any individual's expected future premiums are less than their expected future payouts, then it's not insurance. There's a common misconception that insurance is about pooling risk between patients, which is not really the case - it's true at an accounting level, but only as a secondary effect. The fundamental purpose of insurance is to smooth risk across states of the world, which is a completely different thing. Pretend for the moment that spontaneous combustion of humans exists, can be predicted with some understood probability model, and is an insurable event. In that case, "spontaneous combustion insurance" is not about smoothing risk between Alice and Bob - it's about smoothing risk between universe A (in which Bob spontaneously combusts tomorrow) and universe B (in which Bob does spontaneously combust). Spontaneous combustion is an easy example to use illustratively because it's a binary event, terminal (you can only combust once), and intuitively independent (by definition of spontaneity, Bob's combustion does not impact Alice's chances of combustion). This is a simplified version of the general insurance model, however, which is capable of handling non-binary events, non-terminal events, and interdependent events. In all of those cases, the risk smoothing happens between states-of-the-world, not between individuals. > If an insurer finds out that you will develop a certain type of cancer for which median treatment cost is 1M USD and lifetime support costs will be 30K USD/year, they should not be allowed to increase your premium accordingly. That's a value judgement that you're free to make, but at that point, we're no longer talking about insurance in anything but name.

m_mueller 9y ago

.. which is why a government guaranteed insurance with good overall coverage is necessary. severe market defects in health care furthermore make dictated prices for services/prescriptions necessary. Adjusting Obamacare for that you end up with the Swiss system - still expensive bur at least effective and IMO fair. Thinking about it, increasing information asymmetry due to machine learning applied to marketing may make heavy regulations necessary in most other sectors as well. Imagine software that tracks every person and figures out what they need when, identifies a monopoly of opportunity for any given good and them jacks up the price just for that person individually - in real time, while the person walks up to it. Maybe that's how capitalism ends. Slavery 2.0.

specialist 9y ago
"Imagine software that tracks every person and figures out what they need when..." aka the status quo. People are profiting from my data. I want my cut. Screw privacy. I want my cut. Any one storing or using my data for any reason owes me money. Sell my address to a direct mail campaign. I want my cut. Need my health history for some longitudinal study that may ultimately save lives. Awesome. Pay me. Use my purchase history to recommend a new pair of shoes or sell ads. Woot! Pay me. Micro target my snowflake demographic for targeted political agitprop (aka campaigning). That's worth at least $50. Pay me.
- m_mueller 9y ago
  Have you looked at the BAT project? Seems to me that something along these lines could bring such a vision closer. As in, your browser blocks tracking / identification by default (integrated TOR?). Sites can offer you a bit of tokens if you agree to share your identity / if you let them track what you're looking at. It's a bit of an extension of what BAT aims for, but I think their vision goes in such a direction. I could see such a thing taking hold, just like AdBlock Plus and co. have become a thing that advertisers / publishers need to work around.
  - specialist 9y ago
    Do you mean Basic Attention Token? https://basicattentiontoken.org https://basicattentiontoken.org Looks interesting. Thank you.
    - m_mueller 9y ago
      Yep, that's the one. Disclaimer: I hold some of those ;).

laydn 9y ago

This should quickly be declared illegal, if not already. With the amount of data that is being gathered on us, coupled with advances in machine learning and statistical analysis, in the near future, "they" will be able to figure out high risk scenarios before we even know about it. If they drop coverage or increase premiums based on this, they are no longer an "insurance company". I don't know what we would call them.

chimeracoder 9y ago
> This should quickly be declared illegal, if not already. Dropping coverage like that is illegal already. > If they ... increase premiums based on this, they are no longer an "insurance company". I don't know what we would call them. Quite the contrary - an insurance company, by definition, is expected to price insured entities based on the expected future payouts (ie, the levels of risk). More specifically, the sum of all expected future premiums that a patient pays should be slightly more than the sum of all expected future claims for that patient. If you're expecting it to behave otherwise, then you're looking for something that isn't insurance - insurance is, by definition, a model that relates risk and price.
- laydn 9y ago
  Shouldn't that read: "" More specifically, the sum of all expected future premiums that all patients pay should be slightly more than the sum of all expected future claims for all patients "" You can not reduce the calculations to a single patient. If an insurer finds out that you will develop a certain type of cancer for which median treatment cost is 1M USD and lifetime support costs will be 30K USD/year, they should not be allowed to increase your premium accordingly.
  - chimeracoder 9y ago
    > Shouldn't that read: "" More specifically, the sum of all expected future premiums that all patients pay should be slightly more than the sum of all expected future claims for all patients "" No, it shouldn't. I mean, the thing you wrote is a true statement too, because of the distributive property of summation, but the point is that insurance does work on a per-plan level. If any individual's expected future premiums are less than their expected future payouts, then it's not insurance. There's a common misconception that insurance is about pooling risk between patients, which is not really the case - it's true at an accounting level, but only as a secondary effect. The fundamental purpose of insurance is to smooth risk across states of the world, which is a completely different thing. Pretend for the moment that spontaneous combustion of humans exists, can be predicted with some understood probability model, and is an insurable event. In that case, "spontaneous combustion insurance" is not about smoothing risk between Alice and Bob - it's about smoothing risk between universe A (in which Bob spontaneously combusts tomorrow) and universe B (in which Bob does spontaneously combust). Spontaneous combustion is an easy example to use illustratively because it's a binary event, terminal (you can only combust once), and intuitively independent (by definition of spontaneity, Bob's combustion does not impact Alice's chances of combustion). This is a simplified version of the general insurance model, however, which is capable of handling non-binary events, non-terminal events, and interdependent events. In all of those cases, the risk smoothing happens between states-of-the-world, not between individuals. > If an insurer finds out that you will develop a certain type of cancer for which median treatment cost is 1M USD and lifetime support costs will be 30K USD/year, they should not be allowed to increase your premium accordingly. That's a value judgement that you're free to make, but at that point, we're no longer talking about insurance in anything but name.