• 3 Posts
  • 42 Comments
Joined 1 year ago
cake
Cake day: June 20th, 2023

help-circle


  • I think it’s worth thinking about this in a technical sense, not just in a political or capitalist sense: Yes, car companies want self driving cars, but self driving cars are immensely dangerous, and there’s no evidence that self driving cars will make roads safer. As such, legislation should be pushing very hard to stop self driving cars.

    Also, the same technology used for self driving is used for AEB. This actually makes self-driving more likely, in that the car companies have to pay for all that equipment anyway, they may as well try and shoehorn in self driving. On top of this, I have no confidence that the odds of an error in the system (eg: a dirty sensor, software getting confused) is not higher than the odds of a system correctly braking when it needs to.

    This means someone can get into a situation where they are:

    • in a car, on a road, nothing of interest in front of them
    • the software determines that there is an imminent crash
    • Car brakes hard (even at 90mph), perhaps losing traction depending on road conditions
    • may be hit from behind or may hit an object
    • Driver is liable even though they never actually pressed the brakes.

    This is unacceptable on its face. Yes, cars are dangerous, yes we need to make them safer, but we should use better policies like slower speeds, safer roads, and transitioning to smaller lighter weight cars, not this AI automation bullshit.





  • I’m just using it as a space heater for my study, which is also where I work from. While using the computer in Winter I just switch on f@h for both CPU and GPU (AMD 5700x and 6700xt), and this heats up the room. It’s a good 300-400W. I have home assistant telling me the temperature in the room and it bugs me to turn it off if it’s too hot. That’s my “temperature control”. I didn’t build anything, the computer is just under my desk and it heats up my room.

    Originally my plan was to have F@H automatically turn on and off based on temperature, but it turns out the power is low enough and the lag is high enough that you switch it on in the morning, and then once the room is upto temperature you can just switch it off and the room will stay warm the rest of the day.


  • I do this. If you want to actually want to use or donate the processing power, this is kind of a good thing. However, there are a lot of downsides:

    • Computers are generally much lower power than a heater. This makes them very slow to “react” to heating needs. Heating a small room, even with a 500W PC, could take an hour or maybe more.
    • Heaters have a thermostat, which computers don’t, so even though they are very laggy, they also don’t stop heating when the temperature is right. This means they can overshoot and make the room uncomfortably hot.
    • You could set up an external thermostat but then you need a load which can be switched on and off.
    • I was using folding@home, but the work items take a long time, and switching them on and off will increase the time taken to resolve the work item, which in turn means the system could get annoyed and use someone else’s computer to resolve the work item faster, or worse, blacklist your computer.
    • Using your PC to generate heat will use up its maximum lifetime. The fans aren’t built to be running at max speed all the time, the CPU & GPU could wear out, and the power systems will also wear as time goes on. You sort of have to align that lifetime against usage. This is likely fine if you see the computation as a donation or if you have important stuff to compute, but it’s probably not worth just wasting the cycles.













  • This is probably even true in the philosophy sense. Basically instead of a single lever, each of us gets a lever which might change something or might not, or it might do something unrelated. This means that everyone’s responsibility for that decision is dithered. This sort of rewrites the trolley problem. How does it change the philosophy? No idea.