10 ms·

Totally disagree. No system of rules could ever handle all the obscure edge-cases that humans can. Just look to all the Google account automated-flagging issues

Totally disagree. No system of rules could ever handle all the obscure edge-cases that humans can. Just look to all the Google account automated-flagging issues that pop up on HN, for reference. What could maybe work is an automated system for the "easy 90%", paired with an appeals system well-staffed by human beings. But that isn't Amazon's plan (or Google's, etc), because that 10% will cost much more than the first 90%, so it's easier for them to just write it off as a loss. And at their scale, that margin of error translates to thousands of potentially-ruined lives.

kjhughes 5y ago
If no system of rules can ever cover the policy space, then the policy itself is necessarily ill-defined. That alone is a problem before automation considerations. I'm arguing that automation provides the opportunity to make good, open assessment part of a process of improvement rather than opaquely driving firings.
- brundolf 5y ago
  > then the policy itself is necessarily ill-defined I'm saying no policy could ever be defined well enough to cover all possible contingencies of a messy, human world. The conceit that it could is one of the biggest driving factors of our nascent tech-dystopia. An open policy would of course be better than a closed policy, if only because it would lay bare all of its flaws, but even then it could never be completely fixed.
  - neuronexmachina 5y ago
    > I'm saying no policy could ever be defined well enough to cover all possible contingencies of a messy, human world. "All models are wrong, but some are useful."
    - version_five 5y ago
      Not when people are punished as a result of them. It's fine to make a generalization that works on average for some low stakes, benign decision, but when it materially affects real people, it is not right to generalize. This is the core argument imo against algorithmic decision making of this kind.
      - julienb_sea 5y ago
        You can make the exact same argument in reverse. When decisions materially affect real people, it is unacceptable and fundamentally unfair to rely on human judgement, along with its biases and capriciousness. Clear, transparent policy applied evenly and without bias is a more fair approach for people.
        version_five 5y ago
        No, it doesnt work that way. By projecting real, complex decisions down to some low dimensional set of criteria, you automatically throw away information and generalize. It's only fair in the sense that the criteria is the same, but it's necessarily based on some kind of average outcome and so effectively sacrifices outliers. Its is still extremely unfair to inputs (who are actual people in the cases we mean) that exhibit characteristics that reliably fit the prediction on average, but that don't conform to the prediction.
- hnick 5y ago
  This seems like a restatement of the argument that if we just code enough 'if..then..else' statements, we'd get a well-behaved general AI. That approach hasn't worked, and complicating rules and procedures to cover all edge cases will fall short too. That's why judges are still humans. The list of edge cases, the number of factors that can go into a decision, is correlated to the number of things possible in reality. It's basically unbounded. If I've misunderstood and you are arguing instead for a clear and minimal set of rules to flag and trigger a human review, that sounds sensible to me. In my experience though when companies find an exception they update the policy, and before you know it it becomes a tangled mess.
  - kjhughes 5y ago
    Such a hybrid, automanual system is certainly viable, and your analogy to laws being rules and judges being interpreters is reasonable. In that context, consider that a single law, "do the right thing," might require too many judges and might produce too inconsistent of outcomes to work well. "The right thing" may have to be better defined. Similarly for company policy, it is not the automation used to express bad, ill-defined, or secretive policy that's the problem. The problem is the policy itself.
gjs278 5y ago
google doesn’t publish their rules. you didn’t understand a single thing written in the user’s comment you replied to.
bradleyjg 5y ago
> No system of rules could ever handle all the obscure edge-cases that humans can. In the US federal criminal justice system during the 1970s and early 80s judges had a ton of discretion over sentencing. Some were “hanging judges” that gave really harsh sentences. Some never gave any harsh sentences. Some gave harsh sentences to black people but not to white people, and so on. The legal profession and Congress looked at this situation and decided it wasn’t fair. The difference between probation and ten years in prison shouldn’t be whether or not the judge had indigestion. The federal sentencing guidelines were put in place—-a very detailed system that considered the crime, mitigating and aggregating circumstances, the background of the convict, and several other relevant factors. Judges were supposed to use this system to impose uniform, fair sentences. Judges hated the sentencing guidelines and kept on trying to depart from them “in the interest of justice”. For a while the Supreme Court said no—-it forced them to follow the system. But in the mid aughts a new majority reversed these precedents. The sentencing guidelines were declared to be advisory. It took a while for the guidelines era habits to fade but now we are back to a situation where which judge you get is a significant factor in how harsh a sentence you get. In theory the magic answer is guided discretion. That allows for mostly uniform treatment and a safety valve for unusual situations. But how do you enforce that on a nuts and bolts level? Everyone thinks his particular case is special. Thoughtful empathetic judges are likely to be persuaded that they are. I don’t know what the answer is but I do know there isn’t an easy one.
- raxxorrax 5y ago
  Ha, I get what you are saying, because workers are prisoners too!
powerapple 5y ago
Machine may screw up 10%, human will screw up another 10%. The difference is that machine will be fair for 90% cases, and human will not. I doubt google flagging issue will be better with humans. Yes, it is not perfect, just as road accidents do happen a lot even though human drivers are in those cars.
- raxxorrax 5y ago
  You can talk to a human, you cannot do that with a machine. Even with those that emulate conversation. It lacks reason.

kjhughes 5y ago

If no system of rules can ever cover the policy space, then the policy itself is necessarily ill-defined. That alone is a problem before automation considerations. I'm arguing that automation provides the opportunity to make good, open assessment part of a process of improvement rather than opaquely driving firings.

brundolf 5y ago
> then the policy itself is necessarily ill-defined I'm saying no policy could ever be defined well enough to cover all possible contingencies of a messy, human world. The conceit that it could is one of the biggest driving factors of our nascent tech-dystopia. An open policy would of course be better than a closed policy, if only because it would lay bare all of its flaws, but even then it could never be completely fixed.
- neuronexmachina 5y ago
  > I'm saying no policy could ever be defined well enough to cover all possible contingencies of a messy, human world. "All models are wrong, but some are useful."
  - version_five 5y ago
    Not when people are punished as a result of them. It's fine to make a generalization that works on average for some low stakes, benign decision, but when it materially affects real people, it is not right to generalize. This is the core argument imo against algorithmic decision making of this kind.
    - julienb_sea 5y ago
      You can make the exact same argument in reverse. When decisions materially affect real people, it is unacceptable and fundamentally unfair to rely on human judgement, along with its biases and capriciousness. Clear, transparent policy applied evenly and without bias is a more fair approach for people.
      - version_five 5y ago
        No, it doesnt work that way. By projecting real, complex decisions down to some low dimensional set of criteria, you automatically throw away information and generalize. It's only fair in the sense that the criteria is the same, but it's necessarily based on some kind of average outcome and so effectively sacrifices outliers. Its is still extremely unfair to inputs (who are actual people in the cases we mean) that exhibit characteristics that reliably fit the prediction on average, but that don't conform to the prediction.
hnick 5y ago
This seems like a restatement of the argument that if we just code enough 'if..then..else' statements, we'd get a well-behaved general AI. That approach hasn't worked, and complicating rules and procedures to cover all edge cases will fall short too. That's why judges are still humans. The list of edge cases, the number of factors that can go into a decision, is correlated to the number of things possible in reality. It's basically unbounded. If I've misunderstood and you are arguing instead for a clear and minimal set of rules to flag and trigger a human review, that sounds sensible to me. In my experience though when companies find an exception they update the policy, and before you know it it becomes a tangled mess.
- kjhughes 5y ago
  Such a hybrid, automanual system is certainly viable, and your analogy to laws being rules and judges being interpreters is reasonable. In that context, consider that a single law, "do the right thing," might require too many judges and might produce too inconsistent of outcomes to work well. "The right thing" may have to be better defined. Similarly for company policy, it is not the automation used to express bad, ill-defined, or secretive policy that's the problem. The problem is the policy itself.

gjs278 5y ago

google doesn’t publish their rules. you didn’t understand a single thing written in the user’s comment you replied to.

bradleyjg 5y ago

> No system of rules could ever handle all the obscure edge-cases that humans can. In the US federal criminal justice system during the 1970s and early 80s judges had a ton of discretion over sentencing. Some were “hanging judges” that gave really harsh sentences. Some never gave any harsh sentences. Some gave harsh sentences to black people but not to white people, and so on. The legal profession and Congress looked at this situation and decided it wasn’t fair. The difference between probation and ten years in prison shouldn’t be whether or not the judge had indigestion. The federal sentencing guidelines were put in place—-a very detailed system that considered the crime, mitigating and aggregating circumstances, the background of the convict, and several other relevant factors. Judges were supposed to use this system to impose uniform, fair sentences. Judges hated the sentencing guidelines and kept on trying to depart from them “in the interest of justice”. For a while the Supreme Court said no—-it forced them to follow the system. But in the mid aughts a new majority reversed these precedents. The sentencing guidelines were declared to be advisory. It took a while for the guidelines era habits to fade but now we are back to a situation where which judge you get is a significant factor in how harsh a sentence you get. In theory the magic answer is guided discretion. That allows for mostly uniform treatment and a safety valve for unusual situations. But how do you enforce that on a nuts and bolts level? Everyone thinks his particular case is special. Thoughtful empathetic judges are likely to be persuaded that they are. I don’t know what the answer is but I do know there isn’t an easy one.

raxxorrax 5y ago
Ha, I get what you are saying, because workers are prisoners too!

powerapple 5y ago

Machine may screw up 10%, human will screw up another 10%. The difference is that machine will be fair for 90% cases, and human will not. I doubt google flagging issue will be better with humans. Yes, it is not perfect, just as road accidents do happen a lot even though human drivers are in those cars.

raxxorrax 5y ago
You can talk to a human, you cannot do that with a machine. Even with those that emulate conversation. It lacks reason.