39 ms·

Because you'll be shot, probably by a robot.

anigbrowl 9y ago
That works OK as a deterrent against (some) individuals, but there are numerous historical examples of its failure to scale.
- ben_w 9y ago
  One of the stories of WW1 was a soldier explaining that a machine gun could kill enemy soldiers indefinitely if it had ammo, but the gunner had to stop for the sight of human dead. KillBot doesn’t need to stop. KillBot can read your body language. KillBot knows if you’re loyal to Supreme Leader. KillBot doesn’t care if your death makes you a martyr, all that does is add a few dozen more targets to KillBot’s naughty list. Let’s not have KillBot.
  - lostboys67 9y ago
    and if it had a crew to change belts top up the water and hump new boxes of ammo to the front
    - ben_w 9y ago
      I believe the original story didn’t bother with the specifics of machine gun logistics and maintenance as it wasn’t the relevant point.
- chopin 9y ago
  There also numerous examples of failed mass uprisings. If you can throw enough robots at the problem, the 0.1% would have nothing to fear.
  - anigbrowl 9y ago
    That's a big if, given the vast range of subversive strategies available.
    - chopin 9y ago
      Most of which require communication which is being heavily surveilled. Afaik, uprisings, even mass uprisings are directed by few well-connected individuals. If you can single those out early enough in the process...
      - anigbrowl 9y ago
        To some extent they are, but uprisings are often misrepresented as coups in retrospect. I have some radically different notions, but they're somewhat inchoate and trying to explain them in a long meandering comment would probably be a waste of my time and that of other readers - I feel like I'm a few years away from being able to generalize it, and realistically that's probably an underestimate. As a very quick thumbnail, I'm trying to blend Integrated Information Theory, Selectorate Theory, the eusocial dynamics of colonial animals (including ourselves, per Wilson), and some probably-wrong ideas about meiosis (cell division) to come up with a biologically-rooted model of the collective action problem. Not being a professional scholar, there's little or no chance I'll ever weld these into a general theory of human social behavior, nor do I want to; I hope to intuit a good-enough grasp to deploy it within polities and rely on its fundamental harmony with nature as the least destructive method of manufacturing large-scale social change. It's probably a doomed endeavor but I'd rather fail at a hard problem :)
  - visarga 9y ago
    We don't need uprisings. We just need to own the damn robots. When high automation comes, it's not going to be only the rich that own robots. As an analogy, take smartphones - there are 5 billions of them in use in 2017. Almost everyone can have the best phone today, or one that was as good as the best from a couple of years ago. There are going to be cheap and useful robots for everyone, 3d-printed, open sourced designs running on open source AI, and capable of assembling more robots. What goes into a robot? Motors, gears, sensors, batteries and computer chips? Those are all commodity everywhere on earth. The moment we can assemble them into working useful robots, we will see an explosion of do-it-yourself automation everywhere. With self printing and self assembly we can make a software based factory that can "compile" itself and anything we desire. We will have built the first "material compiler" that can "bootstrap itself" by creating a replica. Then everything will cost only as much as raw materials.
    - ben_w 9y ago
      Which is great, unless your robots have any bugs that allow remote takeovers. How frequent are root-giving zero-day bugs at the moment? If these robots can move, not even “requires physical access” will save you from the Zombie Robot Apocalypse.
    - explainplease 9y ago
      I hate to be dystopic, but what happens when a futuristic Unabomber gets a hold of 3D-printing, self-replicating robots and programs them to kill? Will we all be walking around with our own 3D-printed bodyguard robots, programmed to watch and destroy any other robots that look hostile? Or will the first bad actor's robots destroy everyone else's (or everyone else)? > Then everything will cost only as much as raw materials. Seems like one would want to own the raw materials, then. But if the evil Dr. Robotico's robots can kill the owners and take their quarries, then what? Does he rule the whole world? Are we in a race to robot armageddon? Man, that Paperclips game is starting to seem more and more prescient.
s73ver_ 9y ago
But eventually, the robots, unable to cope with what they've got to do, will turn to robot drugs, robot alcohol, and other robot-destructive behaviors to escape.
- anigbrowl 9y ago
  That's funny, but insofar as they need to be truly adaptive they're subject to subversion. See James Cameron's in-depth studies of the halting problem as applied to the psychodynamics of biopollutant elimination.
- visarga 9y ago
  That would mean getting illicit reward signals in their reinforcement learning system (behavioral system).
- lostboys67 9y ago
  Ah like Bender and "Olde Fortran"

That works OK as a deterrent against (some) individuals, but there are numerous historical examples of its failure to scale.

ben_w 9y ago
One of the stories of WW1 was a soldier explaining that a machine gun could kill enemy soldiers indefinitely if it had ammo, but the gunner had to stop for the sight of human dead. KillBot doesn’t need to stop. KillBot can read your body language. KillBot knows if you’re loyal to Supreme Leader. KillBot doesn’t care if your death makes you a martyr, all that does is add a few dozen more targets to KillBot’s naughty list. Let’s not have KillBot.
- lostboys67 9y ago
  and if it had a crew to change belts top up the water and hump new boxes of ammo to the front
  - ben_w 9y ago
    I believe the original story didn’t bother with the specifics of machine gun logistics and maintenance as it wasn’t the relevant point.
chopin 9y ago
There also numerous examples of failed mass uprisings. If you can throw enough robots at the problem, the 0.1% would have nothing to fear.
- anigbrowl 9y ago
  That's a big if, given the vast range of subversive strategies available.
  - chopin 9y ago
    Most of which require communication which is being heavily surveilled. Afaik, uprisings, even mass uprisings are directed by few well-connected individuals. If you can single those out early enough in the process...
    - anigbrowl 9y ago
      To some extent they are, but uprisings are often misrepresented as coups in retrospect. I have some radically different notions, but they're somewhat inchoate and trying to explain them in a long meandering comment would probably be a waste of my time and that of other readers - I feel like I'm a few years away from being able to generalize it, and realistically that's probably an underestimate. As a very quick thumbnail, I'm trying to blend Integrated Information Theory, Selectorate Theory, the eusocial dynamics of colonial animals (including ourselves, per Wilson), and some probably-wrong ideas about meiosis (cell division) to come up with a biologically-rooted model of the collective action problem. Not being a professional scholar, there's little or no chance I'll ever weld these into a general theory of human social behavior, nor do I want to; I hope to intuit a good-enough grasp to deploy it within polities and rely on its fundamental harmony with nature as the least destructive method of manufacturing large-scale social change. It's probably a doomed endeavor but I'd rather fail at a hard problem :)
- visarga 9y ago
  We don't need uprisings. We just need to own the damn robots. When high automation comes, it's not going to be only the rich that own robots. As an analogy, take smartphones - there are 5 billions of them in use in 2017. Almost everyone can have the best phone today, or one that was as good as the best from a couple of years ago. There are going to be cheap and useful robots for everyone, 3d-printed, open sourced designs running on open source AI, and capable of assembling more robots. What goes into a robot? Motors, gears, sensors, batteries and computer chips? Those are all commodity everywhere on earth. The moment we can assemble them into working useful robots, we will see an explosion of do-it-yourself automation everywhere. With self printing and self assembly we can make a software based factory that can "compile" itself and anything we desire. We will have built the first "material compiler" that can "bootstrap itself" by creating a replica. Then everything will cost only as much as raw materials.
  - ben_w 9y ago
    Which is great, unless your robots have any bugs that allow remote takeovers. How frequent are root-giving zero-day bugs at the moment? If these robots can move, not even “requires physical access” will save you from the Zombie Robot Apocalypse.
  - explainplease 9y ago
    I hate to be dystopic, but what happens when a futuristic Unabomber gets a hold of 3D-printing, self-replicating robots and programs them to kill? Will we all be walking around with our own 3D-printed bodyguard robots, programmed to watch and destroy any other robots that look hostile? Or will the first bad actor's robots destroy everyone else's (or everyone else)? > Then everything will cost only as much as raw materials. Seems like one would want to own the raw materials, then. But if the evil Dr. Robotico's robots can kill the owners and take their quarries, then what? Does he rule the whole world? Are we in a race to robot armageddon? Man, that Paperclips game is starting to seem more and more prescient.

s73ver_ 9y ago

But eventually, the robots, unable to cope with what they've got to do, will turn to robot drugs, robot alcohol, and other robot-destructive behaviors to escape.

anigbrowl 9y ago
That's funny, but insofar as they need to be truly adaptive they're subject to subversion. See James Cameron's in-depth studies of the halting problem as applied to the psychodynamics of biopollutant elimination.
visarga 9y ago
That would mean getting illicit reward signals in their reinforcement learning system (behavioral system).
lostboys67 9y ago
Ah like Bender and "Olde Fortran"