3 ms·

Killing animals needlessly aside, if you have to repeat experiments because you're sloppy, you're not moving at the correct speed as you'll have to double-run e

Killing animals needlessly aside, if you have to repeat experiments because you're sloppy, you're not moving at the correct speed as you'll have to double-run every trial you're doing. Regardless how you feel about killing animals for humans, don't you want companies that can help people to actually move forward faster rather than slower?

treis 4y ago
They're not repeating all experiments.
- capableweb 4y ago
  I don't think anyone has claimed that either. What the claim is, is that Neuralink might have abnormally high number of repeated experiments as they continue to commit mistakes which makes them either have weak value or having to repeat the same trials. > the number of animal deaths is higher than it needs to be for reasons related to Musk’s demands to speed research. Through company discussions and documents spanning several years, along with employee interviews, Reuters identified four experiments involving 86 pigs and two monkeys that were marred in recent years by human errors. The mistakes weakened the experiments’ research value and required the tests to be repeated,
  - treis 4y ago
    >I don't think anyone has claimed that either You said they'd "have to double-run every trial you're doing". Which isn't correct. They have to double run the experiments they screwed up on. That may or may not be faster than taking the time to improve experiment quality.
radu_floricica 4y ago
That's isolated demand for rigor. You're basically saying that "yes, they're doing great work, and the average slaughterhouse is killing the same number of animals in a day or two, but they could be 10% more efficient and that makes them evil".
- danShumway 4y ago
  If the argument here is that they should be allowed to violate research ethics in order to get 10% faster at letting paralyzed people walk, I think it's fair to point out that they way they're conducting research seems to be fairly sloppy and probably isn't actually giving them the productivity/efficiency boost that their defenders claim. thefounder here argues that running experiments in parallel is a moral obligation because we need to get Neuralink out the door as quickly as possible. capableweb responds by arguing that Neuralink's current strategy isn't the fastest way to get that research done. I don't see that as an isolated demand for rigor, I think that's a reasonable objection to the idea that the need to be perfectly efficient outweighs animal ethics.
  - radu_floricica 4y ago
    Actually, that's a pretty clear tradeoff of moving faster. You can't say "ok, we want to move faster because we want to heal people faster" and have absolutely no downside. That would mean you were just inefficient before. Truly upping the pressure comes with downsides, which you accept in the name of speed. And one of the most obvious is that you accept a higher probability of things breaking, as long as overall you make better progress. Parallelizing is very similar to the way modern processors do speculative computing: they go ahead and run several branches of code at once, before knowing which one will actually need to be executed. Once they know, they just use that result and discard the others. It's a pretty standard strategy every time you want to trade money for speed. In the first push for reaching the moon, there was mass production of rockets that went a lot faster than a slow build-test-analyze-modify-build. By the time one rocket went kaboom, quite a few were already in various stages of being built or even finished.
    - danShumway 4y ago
      If your argument is that capableweb is wrong and that parallelizing experiments will make research faster, then fine? That's an argument you can make. But either way I don't really see it as an isolated demand for rigor: capableweb is disagreeing about whether Neuralink's strategy is actually optimized for speed, given that thefounder seems to be arguing that optimizing for speed is a moral obligation.

treis 4y ago

They're not repeating all experiments.

capableweb 4y ago
I don't think anyone has claimed that either. What the claim is, is that Neuralink might have abnormally high number of repeated experiments as they continue to commit mistakes which makes them either have weak value or having to repeat the same trials. > the number of animal deaths is higher than it needs to be for reasons related to Musk’s demands to speed research. Through company discussions and documents spanning several years, along with employee interviews, Reuters identified four experiments involving 86 pigs and two monkeys that were marred in recent years by human errors. The mistakes weakened the experiments’ research value and required the tests to be repeated,
- treis 4y ago
  >I don't think anyone has claimed that either You said they'd "have to double-run every trial you're doing". Which isn't correct. They have to double run the experiments they screwed up on. That may or may not be faster than taking the time to improve experiment quality.

radu_floricica 4y ago

That's isolated demand for rigor. You're basically saying that "yes, they're doing great work, and the average slaughterhouse is killing the same number of animals in a day or two, but they could be 10% more efficient and that makes them evil".

danShumway 4y ago
If the argument here is that they should be allowed to violate research ethics in order to get 10% faster at letting paralyzed people walk, I think it's fair to point out that they way they're conducting research seems to be fairly sloppy and probably isn't actually giving them the productivity/efficiency boost that their defenders claim. thefounder here argues that running experiments in parallel is a moral obligation because we need to get Neuralink out the door as quickly as possible. capableweb responds by arguing that Neuralink's current strategy isn't the fastest way to get that research done. I don't see that as an isolated demand for rigor, I think that's a reasonable objection to the idea that the need to be perfectly efficient outweighs animal ethics.
- radu_floricica 4y ago
  Actually, that's a pretty clear tradeoff of moving faster. You can't say "ok, we want to move faster because we want to heal people faster" and have absolutely no downside. That would mean you were just inefficient before. Truly upping the pressure comes with downsides, which you accept in the name of speed. And one of the most obvious is that you accept a higher probability of things breaking, as long as overall you make better progress. Parallelizing is very similar to the way modern processors do speculative computing: they go ahead and run several branches of code at once, before knowing which one will actually need to be executed. Once they know, they just use that result and discard the others. It's a pretty standard strategy every time you want to trade money for speed. In the first push for reaching the moon, there was mass production of rockets that went a lot faster than a slow build-test-analyze-modify-build. By the time one rocket went kaboom, quite a few were already in various stages of being built or even finished.
  - danShumway 4y ago
    If your argument is that capableweb is wrong and that parallelizing experiments will make research faster, then fine? That's an argument you can make. But either way I don't really see it as an isolated demand for rigor: capableweb is disagreeing about whether Neuralink's strategy is actually optimized for speed, given that thefounder seems to be arguing that optimizing for speed is a moral obligation.