Comment by capableweb | Hacker News Reader

Comment by capableweb | Hacker News Reader

Parent Full thread

> Five people who have worked on Neuralink’s animal experiments told Reuters they had raised concerns internally. They said they had advocated for a more traditional testing approach, in which researchers would test one element at a time in an animal study and draw relevant conclusions before moving on to more animal tests. Instead, these people said, Neuralink launches tests in quick succession before fixing issues in earlier tests or drawing complete conclusions. The result: more animals overall are tested and killed, in part because the approach leads to repeated tests.

> On several occasions over the years, Musk has told employees to imagine they had a bomb strapped to their heads in an effort to get them to move faster, according to three sources who repeatedly heard the comment

> filed a complaint with the USDA accusing the Neuralink-UC Davis project of botching surgeries that killed monkeys, and publicly released its findings. The group alleged that surgeons used the wrong surgical glue twice, which led to two monkeys suffering and dying

> included one instance in 2021 when 25 out of 60 pigs in a study had devices that were the wrong size implanted in their heads, an error that could have been avoided with more preparation

> staff accidentally implanted Neuralink’s device on the wrong vertebra of two different pigs during two separate surgeries

Surely they could have a more scientific approach that would avoid having to kill so many animals and also make progress faster because the environment would be more stable. If it's borderline illegal to accidentally kill animals or making them suffer when it could prevented, I'm not sure.

As it seems right now, they have something of a cowboy approach to science which will lead to unnecessary pain and death of animals, and walking in circles in their progress as they don't seem to know what they are doing.

I have a better idea: instead to run tests concurrently(kinda what Musk is doing) let's run them sequentially so that less animals are used(even if it takes way more time). We can use the saved animals for some tasty burgers. Nevermind that some people need treatments now. We can give them the saved burgers.

I find it interesting how quickly people resort to a "the ends justify the means" attitude about this stuff when, as pointed out in the article, there are others in the same space doing the same thing who HAVE managed to perform the due diligence required to get the proper approval. Is the assumption their efforts won't lead to anything because they've managed not to screw up the implant sizes or glue?

I think the argument is that they're taking too long. Actual humans continue to suffer with problems that could be addressed by this technology while some people are "hand-wringing" about animal life in a society that raises and kills animals inhumanely because they're tasty.

We can only assume that this is taking too long if we assume that Neuralink is somehow able to skip the fundamentals and be successful. As I pointed out above Synchron is further ahead, they are making more progress because they were able to meet the required standard, and I think we can all agree standards are important for human trials (or is a few extra human deaths ok if it gets to market sooner?), so I'd suggest it's Neuralink's own fault they can't meet the advancement criteria and not just concerns about animal welfare that are getting in the way.

Synchron was also started 4 years earlier, so it's not necessarily because they are doing a better job. Neuralink is playing catch up in a sense.

> Neuralink is playing catch up in a sense.

Yes, in areas like testing discipline. I don't accept the argument that you can skip past the fundamentals and it'll all just work out because you may be able to help some people quicker. That's everyone's goal in that space - to get implants onto the market as fast as possible.

I know you don't find it plausible, just saying that that doesn't necessarily mean it's incorrect.

It is wholly incorrect that you can perform sloppy experiments that lead to excess costs of your research and expect this to be at all respectable or acceptable among your peers just because your research is particularly exciting or potentially useful.

If anything, being sloppy is a point against you when you’re working in particularly impactful stuff, because it shows your skills as a scientist aren’t up to the standards of the field you’re researching.

Aren't Synchron doing a much easier task?

I'd have more sympathy for this argument if Neuralink was genuinely the fastest or more efficient way to solve these problems, but it's a far shot attempting to build a technology that is wildly ambitious where there's no guarantee that it will provide the revolution people claim (or that it will even be commercially and economically available at all to the vast majority of people who need it).

It's not just that the research could be done with less suffering/death, it's also that this is also not really the highest-priority research if we want to improve the lives of differently-abled and disabled people. But we're being asked to excuse bad research practices that result in unnecessary death because somehow Neuralink specifically is an emergency that needs to have market domination as soon as possible to save lives.

I have no objection to Neuralink responsibly conducting research in this direction, but it's far from the most effective or optimistic way that we could be spending that time and money, so I don't see why they specifically should get a pass to violate research ethics. They're already devoting resources in a way that is sub-optimal if you just want to improve the lives of disabled people as quickly as possible. And that's fine, but at least they can devote those resources responsibly without causing additional suffering.

Synchron is not nearly in the "same space". Comparing their active stents to Neuralink's implants is like comparing a current Timex G-Shock to the new Apple dive watch. (Which is to say both are cool, but one requires evolutionary tech progress and the other requires significant R&D.)

Does it matter that Apple's watch is going to eventually be better if only the Timex is able to keep time right now?

Are you claiming that the ends can never justify the means? If not, then I don't see how their arguments would be inherently devoid of merit.

You have to weigh the ends against the means. "Repeating the experiments because of sloppiness" is pretty much the bottom of the barrel when it comes to ends, in my opinion.

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?

They're not repeating all experiments.

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,

>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.

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".

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.

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.

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.

I believe it's a moral obligation not to trade speed for efficiency when you consider the raw material(animals) are almost "free" as far as ethics are concerned(i.e we kill them for the taste of their meat with almost no ethical cost). I have a different opinion on human testing of course.

It's like banning SpaceX's "sloppy" development because the failure of their rockets create pollution. Polution is bad and causes death but I think we should not get religious about polution where it matters little and the potential reward is high.

> when you consider the raw material(animals) are almost "free" as far as ethics are concerned(i.e we kill them for the taste of their meat with almost no ethical cost)

I've commented similarly elsewhere, but I really wish people would stop saying "think about the meat we eat" as an argument for more animal suffering in the world. The correct response to hypocrisy is to have consistently better morals everywhere, not to have consistently worse morals elsewhere.

Of course, people can disagree whether or not killing animals unnecessarily is immoral. But if you want to take the premise that eating meat is wrong, no vegetarian/vegan is ever going to tell you that they don't want unnecessary animal testing stopped until after everyone stops eating meat. People's hypocrisy around specific moral issues should not be treated like a license to completely abandon other standards; taking that approach to morality is a recipe for rapid societal decay.

> where it matters little and the potential reward is high.

Again, you can disagree with capableweb's argument, but I just want to characterize correctly what the argument is -- capableweb is saying that the testing here isn't actually going to significantly increase Neuralink's odds of success or their research speed. I would add to that argument that it may in fact lower Neuralink's odds of success if it encourages sloppier research or decreases public trust in the project. I don't have a strong take on that beyond that I think their testing is symptomatic of Musk imposing deadlines and pressures that are going to be toxic to a productive research environment. I don't know that excessive animal testing is going to slow down their research, but based on what the researchers in this article have said, I am pretty certain that the reason why they're doing excessive animal testing in this case is because Musk is pushing them to do sloppy testing to meet unrealistic deadlines that will lead to sloppy results and a worse product, if they get a product out at all. Fast careless testing is sometimes slower than methodical careful testing.

That's an argument that's orthogonal to the moral cost of killing animals for research purposes. If you don't think there's any moral tradeoff to killing animals at all, then the only thing that matters is how their decisions impact the end result.

And seriously, I just responded on here to clarify what capableweb was talking about and that they weren't making an isolated demand for rigor. I am not interested in getting into an argument about whether all of the research standards for animal welfare should all be dropped so that Musk can theoretically maybe monopolize a market in the future that will likely be inaccessible to most disabled people at a potentially slightly faster speed.

Once you accept that parallelizing is a legitimate strategy with a clear tradeoff, then the conversation moves towards if that tradeoff is legitimate. And here I'm saying that's isolated demand for rigor to say that research that would, at the very least, help heal seriously ill people is less important than a random celebration which ups the meat consumption to the same levels, or inefficiencies in the meat supply chain or hundreds of other sources of waste.

It's more salient, of course, but if the people commenting on this didn't feel the same desire to protect animal lives on other instances in the past few months, I kinda have to think this is less about their actual values and more about the article being written as a successful indignation pump.

> Once you accept that parallelizing is a legitimate strategy with a clear tradeoff

That's the question. Is parallelizing animal testing and ramping up the number of animals used a legitimate strategy and does it have a clear tradeoff? No one in is arguing that animal rights only matter in lab testing. Rather, there are two arguments that can be made:

- A) parallelization of animal testing is sloppy research that won't lead to better results.

- B) the benefits of parallelization don't outweigh the moral cost of animal rights.

Those are two separate arguments, and the parent comments were primarily arguing around A. B is orthogonal. That doesn't mean that B is an irrelevant argument or that there's no case to be made for animal welfare in testing, but it's a separate argument from A.

That's precisely my point. It's quite obvious to me that running tasks in parallel produces both more waste and faster results.

A good comparison would be Musk's other company SpaceX. You can surely put more rigor into r&d to reduce waste and failures (i.e like NASA) but you will progress slower.

Chances are that your company/project will get bankrupt(like so many biotech companies) if you don't move fast enough.

Do you think we're making monkey burgers?

Reply on news.ycombinator.com