Hacker News Reader

Comment by mordymoop | Hacker News Reader

This is one of those instances where it would have been useful to:

- somehow not see that headline number of 1,500 first

- come up with a guess (or look up) how many animals are typically killed in medical testing procedures for drugs, surgeries, etc.

- guess how many animal deaths would be expected for a company doing what Neuralink is trying to do

- only then check the true number after having this estimate in mind

Since we see the 1,500 first we have some kind of immediate emotional reaction to it that is not calibrated to anything. It’s a textbook form of scope insensitivity. Would you have reacted any stronger if the number had been 150,000? Would you have reacted more mildly if the number was 150?

When a headline provides a number and then the article provides no real context for what one might reasonably have expected that number to be, it’s a sign that you are being preyed upon.

View on HN

From the article:

> Some Neuralink rivals are having more success. Synchron, which was launched in 2016 and is developing a different implant with less ambitious goals for medical advances, received FDA approval to start human trials in 2021. The company’s device has allowed paralyzed people to text and type by thinking alone. Synchron has also conducted tests on animals, but it has killed only about 80 sheep as part of its research, according to studies of the Synchron implant reviewed by Reuters.

Seems they are providing context, as long as you take the time to read through the article.

I would argue that "context" is intentional misdirection. Most people don't know that synchron is marketing an electrified stent that gives maybe one channel of data I/O whereas Neuralink has a robot that implants thousands of electrodes in multiple regions of the brain and spinal cord. They are different by orders of magnitude in complexity and potential utility.

There is nothing about what you said that suggests that this requires more animal testing in a shorter timeframe. If anything a single animal test should provide so much more data that it should take much longer to process and analyze the results, slowing down the speed at which animal tests occur.

I'm confused - nothing at Neuralink is off the shelf. How would you go from an empty room to an animal implanted functionally with lots of data?

What is there to be confused about?

On the one hand you seem to be arguing that neuralink is more complicated and has more unknowns, and that this explains why it needs to do more animal tests, but that depends on how much data one can capture in a single test. So it is not a given that the complexity-to-tests-needed graph is linear.

And assuming that we do actually capture a lot more data in a single test (I would hope so at least), then that also means more work needs to be done to interpret this data in a meaningful way and make improvements. Which would suggest that it should take more time to move on from one test to the next.

There is a difference between total animal tests and how quickly the tests are done.

One of the example applications proposes to stimulate patterns of neurons in visual cortex to produce vision. To even get to the point where you can train a monkey to detect phosphenes, Neuralink had to (a) develop a way of producing flexible electrodes that would measure activity from a single neuron, were robust enough to be implanted, would survive the punishing in-vivo environment, and retain their ability to measure and stimulate for long periods of time (none of those steps can be adequately done in in vitro models). (b) They then had to make sure that their robot could implant them to the proper depth (despite brain movement), and that during implantation they could maintain connection to the interface system. (c) They had to build a ASIC that would capture a 1000 of channels of neural data and wirelessly transmit them, while ensuring that the SNR was sufficient for use (again, no in vitro model). (d) They had to test the encapsulation of the ASIC to make sure that it, too, could survive in vivo.

They could certainly have done all these things serially, but it would have taken a lot longer. In the end, where they are is still far from a useful product - they have monkeys that can see phosphenes, they have electrodes that last 1 year, and they have a reasonable ASIC - but compared to the academic timescale, where we've been working on these individual pieces 1 PhD (6 years) at a time, they've made really remarkable progress.

It's amazing that we expect them to ever get that all right when they screw up what glue to use.

> I would argue that "context" is intentional misdirection. Most people don't know that synchron is marketing an electrified stent that gives maybe one channel of data I/O whereas Neuralink has a robot that implants thousands of electrodes in multiple regions of the brain and spinal cord. They are different by orders of magnitude in complexity and potential utility.

The literal quote you're responding to says that Synchron's implant is "less ambitious." It's worth noting that being overambitious is often the root cause of fuck ups.

>> fuck ups.

...and world changing success.

I'm cheering this on from the sidelines, but I'm not advocating. Keep your metal and silicon out of my body, thanks.

>> It's worth noting that being overambitious is often the root cause of fuck ups.

> ...and world changing success.

But mostly fuck ups.

But overambitious is only obvious in retrospect.

If you're successful, it was exactly the right amount of ambition.

> But overambitious is only obvious in retrospect.

I disagree. If you're trying to do, say, implement five revolutionary things no one's ever done all at once, so you can tell people you're going to go from zero to sci-fi in a couple years, there's an extremely high likelihood you're being overambitious. That's something that's obvious from the start.

Go tell that to Thomas Edison, I guess. Or Nikola Tesla.

> Go tell that to Thomas Edison, I guess. Or Nikola Tesla.

The vast majority of people, including engineers and business leaders, are not Edisons or Telsas. Sometimes people get lucky or they have unusually well-resourced and competent teams to tackle an unusual amount of stuff at once. That's why I qualified my statement with "there's an extremely high likelihood..." Given Neuralink is facing "a federal probe and employee backlash" over shoddy practices, I highly doubt they have the unusually well-resourced and competent teams to not be overambitious.

What's an example of Edison or Tesla doing multiple "revolutionary things no one's ever done all at once"? I'm genuinely interested, but details matter a lot. For example, the Edison company's contribution to electric light was mainly extending the life and reducing the amperage of light bulbs. Other people had already demonstrated the potential of electric light.

It's normally a dozen or two animals per product. It's usually 2-4 years to develop a device and get it cleared, an animal or two a month. Not anywhere close to 1500, this is criminal.

I'm not sure what tech space your experience comes from, but this is ridiculous. Thes numbers are true to demonstrate*biocompatibility* but are 4 or 5 orders of magnitude smaller than what is required to show *functionality*, let alone to develop a new technology.

I'm in devices. In pharma you need way more. In devices we do a lot of In vitro work, cadaver work, etc. so animals is just the final stage of that. For drugs you need for more admittedly, but it's usually rabbits and mice you need in high quantities, not monkeys. Usually you do a bunch or rabbits, and then just one or two monkeys to verify. I've never heard of anything close to 1500 monkeys for a product that isn't even on the market.

Read the article. Of the 1,500, 280 are "sheep, pigs and monkeys." The rest are presumably mice. The only specific figure for monkeys is the claim that "two monkeys" were involved in experiments marred by human errors.

This is still way way more than is typical. For context I've worked on artificial hearts where they did a few dozen sheep and pigs before going to human implants, some other devices where you do 5 or 10. Never hundreds just to not even get there yet.

The first artificial hearts were implanted in the 1960's. I think the point I'm trying to make is that Neuralink is doing a huge amount basic science / R&D in addition to trying to get a product into FDA. In their final FDA application, it sounds like (from the article) that they'll have data from ~100 animals. That doesn't mean that the other animals were wasted - it's more that the 100 animals for the FDA are wasted in the sense that they're tested on the pristine final product rather than all the steps before.

They're not.

Typically work like this gets published, shared, talked about publicly exactly so our collective understanding is improved. Presented on at clinical and scientific conferences, etc. What Neuralink is doing looks completely different than this. It's clearly being run like many of his other endeavors. With a significant disrespect for regulators, for the health of the general public, etc.

How does anything Neuralink has done demonstrate disrespect for the health of the general public?

I was saying that in reference to his other projects (Tesla FSD)

AFAIK nobody has been injured due to Tesla FSD, so not exactly sure how that qualifies either.

Actual self-driving cars will save thousands of lives, so (to me) Tesla being the only company that is pursuing it in a way that might actually get there before the end of the decade says they care more about the health of the general public, not less.

The last time I saw the statistics, Tesla autopilot was about average for US drivers, but it's actual usage was biased towards the roads least likely to cause fatalities.

I'm all in favour of the vision behind self-driving cars, but the reality isn't quite as impressive as I'd like, especially given the AI has inherent advantages such as "no blind spot" and "can pay full attention to everything all at the same time" and "not distracted by adverts, hunger, and phone" and "doesn't get tired or drunk".

And — at risk of displaying the limits of my domain knowledge — I think the AI can be tested without being in charge, by asking the AI to create a model of the world as it drives, and/or by testing its predictions of what it will observe in the next few seconds.

I killed a dozen rats for my undergrad thesis.

How many monkeys did you torture though?

The article specifically mentions the monkeys are treated abnormally well.

There are other articles that disagree:

"In two separate incidents, experimenters used an unapproved adhesive called BioGlue to fill holes in the animals’ skulls, which seeped through to the monkeys’ brains. In one monkey, the use of BioGlue caused bleeding in her brain, and she vomited so much from the resulting side effects that she developed open sores in her esophagus"

https://www.businesswire.com/news/home/20220926005606/en/Phy...

Seems independent of if the animals were treated well

Not at all. These animals seems to have been treated very poorly.

Why do you say that? For me, to say they were treated poorly means that they were subjected to unnecessary harm or suffering. Animals can still be harmed as long as it is necessary and appropriate care is taken to minimize the harm.

The quote simply demonstrated that the animal was harmed. It says nothing about the necessity of that harm or care that was taken to minimize it.

Nearly all Labs go to great lengths to minimize harm and ensure the animals are comfortable.

Our current standard of care in surgury is better than this. They didn't need to use bioglue in that way, we have better technology than this, they used it because they were lazy.

What works better?

Looking at the literature, it appears that bioglue is or at least has been used in human Neurosurgical procedures to seal CSF leaks. Without evidence otherwise, it seems like you are just making things up

Primates being treated "abnormally well" usually means they are left to do what they want without interference in their lives. This is a false descriptor of what they are being put through.

Abnormally well in conjunction with the medical use for which they were bred and raised.

My point is that if 1400 of those animals are rats, using the stat "1500 killed" upfront is a bit misleading to laypeople in my non-professional opinion. The article doesn't say, but from other articles it appears that no monkeys were killed.

There seems to be a lot of goalpost moving in this thread.

> X animals were killed!

>> Yes but most of those weren't animals the vast majority of people consider particularly intelligent / morally questionable to kill for medical research purposes

>>> But they tortured monkeys!

> Not anywhere close to 1500, this is criminal.

10 billion animals are killed in the USA each year, but 1500 is criminal?

Yes, because of the details of how and why. It's not the number. Yes, thousands of people die every day, but they doesn't make the needless murder of one person any less of an issue.

Gotcha.

Intentionally killing pigs for food when there are many alternatives: morally fine.

Unintentionally killing pigs for potentially life-saving / life-changing medical research: morally bad.

No, you missed it by a mile.

Killing pigs in a respectful way for food or medical research - ok.

Killing pigs for medical science in a unnecessary, disrespectful way that causes unnecessary tortuous pain to them while doing so- bad.

Not that it contradicts your core argument, but I'd be surprised if the number of animals slaughtered in a way that fails the official legal standard for ethical, but nobody noticed for one reason or another, is less than 0.1% of the total.

How do you kill something respectfully? Bow before you shoot it in the head? And why would that matter?

"Unnecessary" is a value judgement made by you, not an objective property of their death. They died for the purpose of furthering science and medical research – many people would consider that necessary even if fewer mistakes could've been made.

This varies wildly depending on the class of device and technology. There is no "normal".

There is a normal for this class and type of device. This is way outside normal for this class and type of device- source, I've designed them for 20 years.

I'm skeptical. I've worked on devices with no electronics and no moving parts that used as many or more large animals.

It really depends on why you are doing animal studies in the first place.

If you have a working product and are just confirming functionality before going to human trials, sure you don't need that much animal work.

If you are doing active research, I'm trying to understand,fix, or improve a design, you can run through a lot of animals.

Usually it's best to try to do that stuff in vitro first, figure out some analog for animals that you can use on the bench.

Surely you understand that some systems don't have a in vitro model. Particularly , the biological response and interaction is the unknown. I'm working on such a project now. It'll likely require many dozens of large animals if not hundreds to collect some semblance of data to support human use. I would consider anything less unethical and irresponsible development

Suggesting we calibrate our reaction based off other examples of animal suffering isn’t an objective take. Some people don’t want any animals to suffer by human hands. Some people are opposed to animal trials in their entirety. Some people might oppose their use in this particular instance. Some people might just have a hard cut off at three dead mice. Whether or not this number is industry standard can be beside the point for someone else, and it doesn’t mean they’re being “preyed upon” or manipulated for reading and responding to a fact using a different moral framework.

According to the most recent data from the USDA, approximately 915,000 animals were used in research in 2019, which is the latest year for which data is available.

Exact, 1500 is not a such big amount as people thinks. If we use correctly the term animals, expanding it to flies, worms and zebra fishes instead to use it incorrectly as a synonym for "birds and mammals", the real number should be much higher.

To kill 1500 vinegar flies in a week for research is not uncommon or a strange thing.

We need to remind that the faster you do research, the more people will be saved by research. The problem here is not the amount of lab animals killed.

If it's within the normal course of this type of business, why would the feds be investigating it?

Neuralink is way outside the rails. Yes, it is normal to do animal studies with a few dozen carefully planned, documented and cared for animal cases. It is not standard practice to do poorly documented, half baked studies where 1500 monkeys are used and tortured. This is way off the ranks of "normal"

The article mentions 2 monkeys, not 1500. Where did you get that impression and why did you decide to state it as fact?

It was an honest misunderstanding

You are within the edit window as of time of writing and should update your comment, then, rather than leave misinformation standing. Or delete it.

I can no longer edit it.

I would have edited it to say 1500 animals, striking "monkeys".

As someone that works with devices, you should know that 1500 monkeys is a wildly absurd claim.

As someone who also works with devices, I would be extremely skeptical and wouldn't even believe 1500 monkeys if the article did state it, unless proof was included.

Really good point. 1500 now removed from the title

For context Peta says 110m animals killed for science each year https://www.peta.org/issues/animals-used-for-experimentation...

Douglas Houfstedder I am a strange loop. Great book on consciousness. He says monkeys have a higher level of consciousness than say mosquitoes. Maybe similar to a 3 year old human. That's why we feel it is worse morally

to put the number in context in usa we kill 9 billion chickens per year.

The headline is manipulated to imply it's worse than it is. A small animal bio research lab kills thousands of animals a year. They're trying to take Musk down, and unsurprisingly, some on Hacker News are taking the bait.

Reply on news.ycombinator.com