8 ms·

I think the more important point here isn't about ethics and punishment, but incentives. The larger a system is and the stronger it incentivises something, the

by ATsch 7y ago

I think the more important point here isn't about ethics and punishment, but incentives. The larger a system is and the stronger it incentivises something, the more effort you're going to have to put into enforcing the status. This is why, in general, ideas that involve lots of actors changing their behavior contrary to incentives, instead of changing the incentives, are fundamentally flawed.

rpastuszak 7y ago
+1 Also, I feel like every system can be abused if the incentivised metric is defined poorly. Example: almost every advertising KPI. 1. A Brand wants to sell a product (incentive: sales). 2. Agency sales people want to reach a particular KPI—budget in this case, the higher, the better. 3. An Ad Tech company will serve ads based on a KPI agreed with the Agency (e.g. CPM, impressions). It's very easy to satisfy 2 and 3 without bringing any additional value to 1., for example by applying less strict viewability metrics (is 50% in view "visible" or 1%? should we pause the video ad below the fold? what is an impression, exactly?). The original intent (sell more units of product x) is gone. This is due to systematic issues with the industry, not a single specific company. But mainly, this is due to the fact that the established incentives represent only a chunk of the reality and can be followed without focusing on the actual problem. The overlap between actors' goals is poorly defined, small and often doesn't exist. In a nutshell: if we define a metric serving as a shortcut for our goal, people will find a way of improving the metric, but rarely reaching the goal. A broadly defined incentive is better than punishment.
- maccard 7y ago
  That has a name, the Cobra Effect [0] [0] https://en.wikipedia.org/wiki/Cobra_effect https://en.wikipedia.org/wiki/Cobra_effect
  - jefftk 7y ago
    The Cobra Effect is when you're trying to make there be fewer of things (cobras) but the incentives you set up actually lead to there being more of them (it's now worth it for people to be cobra breeders). Perhaps you're thinking of Goodheart's law? https://en.wikipedia.org/wiki/Goodhart%27s_law https://en.wikipedia.org/wiki/Goodhart%27s_law
- scarejunba 7y ago
  That's why attribution is such a big story in advertising. With semi-persistent identifiers, you can learn the story of whether someone was: 1) Visited your store before they saw your ad 2) Only visited your store after seeing your ad You can do this online or for physical store locations. There are lots of players there because big marketing teams are fairly sophisticated now: they want cross-channel identity (across mobile, web, TV), attribution, and measurement.
- ec109685 7y ago
  A short term metric is necessary to allow campaigns to optimize their spend without a long multi-month turn around as conversions are fully tracked and users act upon the ad. While there is risk that the wrong short term metric is chosen (that is not indicative of the real goal the advertiser is trying to achieve), it is not an invalid approach. It is untrue that setting a short term goal will _rarely_ achieve the long term value. Otherwise, the advertiser will throw out that short-term goal since it is flawed.
danans 7y ago
> This is why, in general, ideas that involve lots of actors changing their behavior contrary to incentives, instead of changing the incentives, are fundamentally flawed. Indeed, this is what drives the Tragedy of the Commons - everyone acting according to their individual incentives, maximizing their individual benefit, but while exploiting an unregulated limited common resource, and externalizing the cost of doing so upon the commons. There are two solutions paths which are not mutually exclusive: put a higher price on the resource use or externality, and find a more efficient way to achieve meet needs. In many cases, the current incentives and resulting approaches keep us stuck in local maxima. An example is commuting in single occupancy personal vehicles in urban areas in the US. When you consider the layout of our cities and the limited mass transit availability, it very often makes more individual sense to drive when you consider factors like transporting children to school/activities, grocery shopping, appointments, etc. Sometimes the costs are actually to the individual rather than the commons, or are delayed in impact to the commons. In the previous example hours spent sitting in a vehicle instead of walking deprives individuals of opportunities to burn calories to control their weight, resulting in a variety of health issues. The problem is that there are a powerful set of interests who are interested in maintaining the current approaches and incentives at all cost. Again, in the transportation example above, these would include the fossil fuel extraction and highway expansion industries.
- TeMPOraL 7y ago
  > There are two solutions paths which are not mutually exclusive: put a higher price on the resource use or externality, and find a more efficient way to achieve meet needs. I think the second one isn't really a solution. If you find a more efficient way to achieve the needs, the needs will expand to compensate. Even if everyone executes restraint, an entrepreneur will show up and start selling the remaining resources to people who don't have access to those resources. As far as I know, the only way to solve tragedies of the commons is for everyone to agree on a set of rules, and punishments that disincentivise breaking these rules. Which, in real life, is achievable only through centralizing the power that sets the rules and executes punishment - ongoing coordination is an unstable state.
  - 7y ago
SkyBelow 7y ago
I sometimes like to think of ethics as an incentive system that is (generally) stronger than social norms but weaker than laws. Once you reduce it to incentives, it is a matter of comparing incentives. If the incentive to behave ethically is much lower than the incentive to not, then people will pick the better option. That some people seem to be more ethical than others can be viewed as the extent they estimate the incentives from different behaviors (importance placed in the thoughts of others, if they believe some greater power judges them for their actions, etc.).

by ATsch 7y ago

rpastuszak 7y ago

+1 Also, I feel like every system can be abused if the incentivised metric is defined poorly. Example: almost every advertising KPI. 1. A Brand wants to sell a product (incentive: sales). 2. Agency sales people want to reach a particular KPI—budget in this case, the higher, the better. 3. An Ad Tech company will serve ads based on a KPI agreed with the Agency (e.g. CPM, impressions). It's very easy to satisfy 2 and 3 without bringing any additional value to 1., for example by applying less strict viewability metrics (is 50% in view "visible" or 1%? should we pause the video ad below the fold? what is an impression, exactly?). The original intent (sell more units of product x) is gone. This is due to systematic issues with the industry, not a single specific company. But mainly, this is due to the fact that the established incentives represent only a chunk of the reality and can be followed without focusing on the actual problem. The overlap between actors' goals is poorly defined, small and often doesn't exist. In a nutshell: if we define a metric serving as a shortcut for our goal, people will find a way of improving the metric, but rarely reaching the goal. A broadly defined incentive is better than punishment.

maccard 7y ago
That has a name, the Cobra Effect [0] [0] https://en.wikipedia.org/wiki/Cobra_effect https://en.wikipedia.org/wiki/Cobra_effect
- jefftk 7y ago
  The Cobra Effect is when you're trying to make there be fewer of things (cobras) but the incentives you set up actually lead to there being more of them (it's now worth it for people to be cobra breeders). Perhaps you're thinking of Goodheart's law? https://en.wikipedia.org/wiki/Goodhart%27s_law https://en.wikipedia.org/wiki/Goodhart%27s_law
scarejunba 7y ago
That's why attribution is such a big story in advertising. With semi-persistent identifiers, you can learn the story of whether someone was: 1) Visited your store before they saw your ad 2) Only visited your store after seeing your ad You can do this online or for physical store locations. There are lots of players there because big marketing teams are fairly sophisticated now: they want cross-channel identity (across mobile, web, TV), attribution, and measurement.
ec109685 7y ago
A short term metric is necessary to allow campaigns to optimize their spend without a long multi-month turn around as conversions are fully tracked and users act upon the ad. While there is risk that the wrong short term metric is chosen (that is not indicative of the real goal the advertiser is trying to achieve), it is not an invalid approach. It is untrue that setting a short term goal will _rarely_ achieve the long term value. Otherwise, the advertiser will throw out that short-term goal since it is flawed.

danans 7y ago

> This is why, in general, ideas that involve lots of actors changing their behavior contrary to incentives, instead of changing the incentives, are fundamentally flawed. Indeed, this is what drives the Tragedy of the Commons - everyone acting according to their individual incentives, maximizing their individual benefit, but while exploiting an unregulated limited common resource, and externalizing the cost of doing so upon the commons. There are two solutions paths which are not mutually exclusive: put a higher price on the resource use or externality, and find a more efficient way to achieve meet needs. In many cases, the current incentives and resulting approaches keep us stuck in local maxima. An example is commuting in single occupancy personal vehicles in urban areas in the US. When you consider the layout of our cities and the limited mass transit availability, it very often makes more individual sense to drive when you consider factors like transporting children to school/activities, grocery shopping, appointments, etc. Sometimes the costs are actually to the individual rather than the commons, or are delayed in impact to the commons. In the previous example hours spent sitting in a vehicle instead of walking deprives individuals of opportunities to burn calories to control their weight, resulting in a variety of health issues. The problem is that there are a powerful set of interests who are interested in maintaining the current approaches and incentives at all cost. Again, in the transportation example above, these would include the fossil fuel extraction and highway expansion industries.

TeMPOraL 7y ago
> There are two solutions paths which are not mutually exclusive: put a higher price on the resource use or externality, and find a more efficient way to achieve meet needs. I think the second one isn't really a solution. If you find a more efficient way to achieve the needs, the needs will expand to compensate. Even if everyone executes restraint, an entrepreneur will show up and start selling the remaining resources to people who don't have access to those resources. As far as I know, the only way to solve tragedies of the commons is for everyone to agree on a set of rules, and punishments that disincentivise breaking these rules. Which, in real life, is achievable only through centralizing the power that sets the rules and executes punishment - ongoing coordination is an unstable state.
- 7y ago

SkyBelow 7y ago

I sometimes like to think of ethics as an incentive system that is (generally) stronger than social norms but weaker than laws. Once you reduce it to incentives, it is a matter of comparing incentives. If the incentive to behave ethically is much lower than the incentive to not, then people will pick the better option. That some people seem to be more ethical than others can be viewed as the extent they estimate the incentives from different behaviors (importance placed in the thoughts of others, if they believe some greater power judges them for their actions, etc.).