|
|
|
|
|
|
by titzer
2586 days ago
|
|
|
> where littering is not socially acceptable and happens less than other places. It's great that there are places where littering is much less due to social norms. But it's not zero. People litter. Period. To a first approximation, humans can be modeled as points wandering around on a map, and at each timestep there is some nonzero probability they will just drop garbage on the ground. This is true the world over, in every single country I've been, on essentially every hiking trail, no matter how remote or inaccessible. If you were to have a heatmap of litter the world over, it would very closely approximate the simple heatmap of where people have been. Just look at any stretch of highway. Unless it has just been cleaned up by a very thorough cleanup crew, you will find trash. And this is where people are riding around in vehicles that could as well just keep their trash in the backseat. People go out of their way to throw that shit out the window. People can't even get trash into a fricking trash can. It will literally be sitting on the ground next to a trash can. People are trash monkeys. You can knock one or two orders of magnitude off the litter problem with education, societal norms, fines, etc. But people are trash monkeys, and whereever they go, they fucking litter. It's as inescapable as nuclear decay. It's just a function of probability. |
|
|
When you talk about large populations, a lot of things normally considered individual responsibility suddenly become a function of probability. That's why I believe in many cases we should shift responsibility from individuals to organizations at the other end of a transaction.
E.g. if you know that roughly 5% of casino gamblers end up addicted and in life-ruining debt, and 0.1% of gamblers end up dying or killing people because of that, you can't open a new casino and shield yourself behind "it's the people who can't control themselves". You knew the probabilistic model, you used it to profit, you're responsible for some deaths and for ruining people's lives.
That's IMO doubly important if the probabilistic models has incentive knobs that you control.