Huge math error corrected in black plastic study; authors say it doesn't matter
arstechnica.com
arstechnica.com
They’re just really really good at it.
That’s how ML is. “Prediction” is just another word for really really good guess.
If anything this seems like an ad for an AI. When I try these things they can even solve leetcode, for which they actually have the solution in the training set.
They encode their training data in their weights and do a kind of search on it given a prompt.
If leetcode problems are in their training data, then they’ll of course answer the question with ease.
(This is my understanding, I’m not ml engineer)
Clearly the authors had their conclusion even before doing the study, so of course the error doesn't affect it.
> "This calculation error does not affect the overall conclusion of the paper," the correction reads. The corrected study still ends by saying that the flame retardants "significantly contaminate" the plastic products, which have "high exposure potential."
Seems like the conclusion very obviously still holds, so I'd agree with the study authors here, that error doesn't alter the conclusion.
OK, but that's not what this study was about, right? The original claim by the authors was that these utensils brushed up against safe limits. When that turned out to be false, the claim has shifted to the safe limits are too high. But that's a different study. You need to do a safety analysis of exposures to the substance. That's not what their study did.
There's a lethal dose for alcohol, but having 1/10th of that dose can still strongly negatively impact me.
Based on what? The FDA's safe dose by definition means there are no known significant risks. They build in pretty hefty safety margins in these things. It definitely could still be too high! We get things wrong all the time. But you need a different kind of study to show that.
So, baseline, one day of using black cookware exposes you to a years worth of baseline exposure. Every day. That's why I consider it high. These are known bioaccumulating carcinogenic compounds. I can reduce my baseline by over 99% by spending $10 on a new spatula? Yeah, that's high.
Sorry but I don’t want any toxic material in the tools I use for food.
Therefore, simply averaging the flame retardant content of a random sample of spatulas could result in a bias.
It's precisely the ones that are the most toxic that one would expect to be the most edible.
Less than 10% of items studied contained the specific compound in the article.
FTA:
The study examined 203 black plastic household products, including 109 kitchen utensils, 36 toys, 30 hair accessories, and 28 food serviceware products. Of those 203 products, only 20 (10 percent) had any bromine-containing compounds at levels that might indicate contamination from bromine-based flame retardants, like BDE-209. Of the 109 kitchen utensils tested, only nine (8 percent) contained concerning bromine levels.
"[A] minority of black plastic products are contaminated at levels >50 ppm [bromine]," the study states.
But that's just bromine compounds. Overall, only 14 of the 203 products contained BDE-209 specifically.Probability is a bitch.
I would wager that you inhale more than zero molecules of this "inherently toxic" flame retardant when you breathe. So you are going to stop breathing?
But digging into the study, it becomes clear that the authors are really reaching in their methodology--subjecting utensils to rather implausible scenarios (when was the last time you microwaved your sushi tray?)--to try to get some samples that would exceed the limit. The hyperbolic media coverage of "ditch all your black plastics" isn't really sustained when the research finds that just a few specimens, when subjected to unusually high heat stresses, managed to just barely cross the incorrectly-calculated FDA threshold (itself likely set at a factor of a tenth to a hundredth of the lowest threshold observed to cause issues in previous studies!). That the authors aren't willing to disavow that media coverage is more telling then their unwillingness to adjust their conclusions in response to their math error.
That's not implausible. Sushi actually tastes better a little warm than chilled.
Also I think it would make sense to test all food trays the same way, as some of those do get heated quite a bit (e.g. I've microwaved take-out noodles in plastic trays very similar to a take-out sushi trays). It's not like it's plausible that there's some special sushi-tray supply chain where they've carefully determined they can safely get away with using e-waste plastic. It's almost certain that industry consists of many takeout tray companies that each make a whole line of takeout trays for different uses using the same raw materials (e.g. the sushi cart at my employer sold sushi and hot noodle soup in black plastic trays, which plausibly could come form the same manufacturer).
So if HN doesn't even care...
Interesting. My understanding was that HN yanks a thread off the front page as soon as it has more comments than upvotes, because that's their indicator of being "controversial".
Which never made sense to me, because I leave hundreds of times more comments than article upvotes.
I have read that a lot of times but this doesn't seem to be the case, I can remember multiple times when I clicked on something on the front page and it had more comments than upvotes.
This is currently on the front page with 245 points and 274 comments:
If you had told me that Cocoa Krispies were 50% rat feces, but then came back and said you'd made a mistake and was off by an order of magnitude, it's only 5%, I think it would be pretty reasonable to not update your conclusion that Cocoa Krispies were significantly contaminated with rat feces.
I love how threat inflation turns chemicals from innocuous to deadly in the absence of any strong evidence.
No, we cannot say this chemical is a carcinogen. Read the wikipedia page.
https://en.wikipedia.org/wiki/Decabromodiphenyl_ether
> In 2004, ATSDR wrote "Nothing definite is known about the health effects of PBDEs in people. Practically all of the available information is from studies of laboratory animals. Animal studies indicate that commercial decaBDE mixtures are generally much less toxic than the products containing lower brominated PBDEs. DecaBDE is expected to have relatively little effect on the health of humans."
If you adopted a life strategy of "only avoid things after scientists have conclusive evidence it's dangerous for you" then you'd have unnecessarily ingested hundreds of various poisons over the last 100 years.
2024 https://www.alzheimersresearchuk.org/news/aluminium-and-alzh...
> This research has shown higher aluminium levels in the brain and the fluids that bathe the brain in people with Alzheimer’s when compared to healthy individuals. But just spotting higher levels doesn’t mean they are a cause of the disease. > [...] most scientists think that aluminium build-up in the brain is more likely to be a consequence of Alzheimer’s disease rather than a cause.
Ah yes, the "most scientists" card. "more likely" is also hilarious! Most doctors are more likely to smoke camel.
2010 https://pubmed.ncbi.nlm.nih.gov/21157018/
> The hypothesis that Al significantly contributes to AD is built upon very solid experimental evidence and should not be dismissed. Immediate steps should be taken to lessen human exposure to Al, which may be the single most aggravating and avoidable factor related to AD.
The science folk need an entirely different level of evidence than I do.
No, the wise man doesn't have strong evidence to the contrary. And no, he's not interested in finding out either. The consequences of ingesting a novel non-Lindy chemical is an unknown unknown that the wise man is not interested in discovering.
Of course, if the potential upsides are great enough, the risk of downsides might be worthwhile.
In the case of the chemical spatula, the downsides are uncertain, but there's the possibility of cancer. The upsides are...greater corporate profits?
The wise man is going to have to pass on that one.
Moderns would do well to try and be more like the wise man. Scientific studies are not the holy grail of knowledge. New studies are coming out all the time, both negating and reaffirming old conclusions. Is this schizophrenic flip-flopping not enough to convince the modern that Scientism isn't the end-all-be-all?
Unknown unknowns emerge at the tails of novel changes introduced to complex systems. Scientific studies are unable to account for these long-tail events. When it comes to your environment and your body, be more Lindy, and stop deferring to the myopicity of Scientism to guide you.
>Eat no fruits from the past one thousand years; drink nothing from the past four thousand years.
The longer something has been around, the longer it will continue to be around. The longevity of something also validates its efficacy and resiliency.
If people have been eating figs and drinking wine for thousands of years, then it's probably good and safe for you to do as well.
If traditional salad recipes avoid the use of conium maculatum, then you should probably avoid it too.
If your ancient ancestors didn't cook with margarine, then you probably shouldn't cook with it either.
This is ignores the amounts consumed. Just because a thing is safe at N mg/day doesn't mean it is safe at all doses. The change circumstances of human existence make attempts to come up with simple, eternal rubrics at best a bit chancy, and at worst completely misleading.
Just enjoy your figs, Paul.
But it does not pass the smell test.
There is a plethora of substances and practices that are quite harmful, but have been used for millenia. This is because your suggested methodology fails to detect really anything that does not cause traceable and observable harm before the next generation is raised. And that's a lot of things.
"Science" adds value compared to pure traditionalism because it analyzes precisely how things are harmful, and helps discover mitigations and strategies that pure outcome-driven traditionalism would never have explored.
Examples:
- Lead pipes (used successfully for over two millenia-- harmless? no.)
- Basically every carcinogen ever (e.g. Radon: people in affected regions did simply not know about keeping it out of cellars/dwellings, and just died of lung cancer sometimes)
- Salmonella, syphilis, cholera and other pathogens-- they are non-issues with proper prevention and/or countermeasures-- without those, people just suffer and/or die.
- Alcohol consumption during pregnancy
edit: I'm not saying that "sticking with what worked in the past" is wrong, or useless information, but its just that-- a statistical prior for harm. It won't reliably tell you neither which things are harmless nor which are harmful, it just gives a rough indication of which it might be.
I can also point out--we've been drinking out of lead pipes for thousands of years, so they're obviously safe, right? ... right?
>>> Eat no fruits from the past one thousand years; drink nothing from the past four thousand years.
> I was with you until the quote at the end. What does that even mean?
I think it means "don't consume anything foodstuff that doesn't have a long record of safety (v.s. trusting the guy in the white lab coat that says the new thing is safe, since in 20 years some other guy in a white lab coat may find it's actually very unsafe in some previously unknown way).
It appears to be a quote from Nassim Nicholas Taleb, here's a fuller version: https://manassaloi.com/booksummaries/2016/01/21/bed-procrust...:
> Read nothing from the past one hundred years; eat no fruits from the past one thousand years; drink nothing from the past four thousand years (just wine and water); but talk to no ordinary man over forty. A man without a heroic bent starts dying at the age of thirty.
And doesn't this just privilege tradition? Chemicals will be grandfathered in for spurious reasons, not because they are actually any safer. Famously, chemicals in plants very often light up the Ames test for carcinogenicity and would be ruled out by your argument if they weren't "natural".
Thank you for a most excellent example illustrating my point and demolishing yours.
Alcohol is estimated to cause 4.5% of all cancers in Europe. It's a Group 1 carcinogen, yet it is privileged because of tradition.
Also, FWIW, cow's milk has objectively changed in the last few decades.
Do you not think that it is rational to choose option 1, given our understanding of the Lindy effect?
Isn't this why we (used to?) feed things to mice to learn about them? To collect additional evidence of safety after basic things like knowing what general kinds of things interact with people's biology say it's probably safe?
> The wise man is going to have to pass on that one.
We should be all happy to deliver increase value to the shareholders, in whatever way we can. After all, they are the most important people in the world. The wise man is not too wise if he doesn't believe that.
Really, you bought Tupperware? Even if you don't ask yourself whether any contaminants were released by it in your food, maybe the fact that it gets stained by pasta sauce to a point where it becomes an Internet meme makes it shitty kitchenware?
Even without exposure to food, plastic deteriorates with sunlight exposure or even just age... Why spend money on stuff like this? Buy a porcelain salad bowl instead, get a wooden spatula, put some oil on it so it endures washings.
For example:
- If you're a smoker, your exposure to the hazardous substances in cigarette smoke is sky-high. The risk to your health of being a smoker is so high and so clear that societies spend millions of dollars to try to convince people of the risk.
- If you're living with a smoker, but not a smoker yourself, then your exposure is lower, but still high enough to be actionable. So we see indoor smoking bans and 2nd hand smoke information also included in anti-smoking campaigns.
- If you're walking down the sidewalk and pass by a smoker for three seconds, you're still encountering the exact same hazard, but your exposure is so low that the risk is not really actionable. It's not like we see big public campaigns about crossing the street to avoid the risk of 3 seconds of exposure to second hand smoke. That's because the risk isn't there.
- Someone smoking in their house 50 miles from you is obviously zero risk to you at all, even though the hazard still exists.
You can see the same kind of pattern everywhere: car exhaust is the same hazard everywhere, but whether it's a risk worth actioning depends on whether you live in a dense urban center, or next to a busy freeway, or out in the exurbs.
So back to the plastic stuff. The study authors were claiming that the level of risk was high enough to justify everyone in the world throwing their utensils away and buying brand new ones. When it came out that the exposure factor of their "risk = hazard * exposure" formula was actually off by a factor of 10, they... stuck with the exact same story? I don't know, man. The original claim was below the acceptable risk threshold, and now it's even ten times lower than that. Surely that has some impact on the level of risk? Does this drop the risk from "smoker" to "passing by a smoker for 3 seconds"?
The distinction matters when deciding how much effort mitigating this risk requires: should I throw everything out right now, or just buy something else when my current ones wear out, or is the risk actually so low that it really does not matter? If I replace them, what is the level of risk of the things I replace them with? Should society put in the effort to ban these chemicals? What are the pros/cons of that? Are the risks mitigated worth the costs?
The authors' refusal to acknowledge this really makes me doubt that their conclusion is not affected by some kind of bias. The jury's still out, sure, but I'm not yet sold on their risk claims.
And it's trivial to replace with materials that don't have that risk. For a couple bucks you can remove fire retardants from your food. Why wouldn't you?
These have already been banned in CA from furniture because it creates household exposure. Why would it be ok in cooking utensils?
BTW the chemicals don’t actually prevent house fires it’s basically an industry scam to put them into products.
Flame retardants were introduced into everything because of CA and Federal laws requiring items to smolder and not catch fire. While I don't believe flame retardants should be as prevalent as they are today, I also think its unfair to say they don't prevent house fires. They absolutely have done some saving but I don't think across the entire population its a net positive. These rules were originally put into place because of a number of high profile cases where kids died, the biggest vector were beds, people smoked and dropped the butt on their mattress and poof.
That rule in CA is in the right direction, glad they are helping right some of the wrong they did but it is still in everything. I think its more helpful to paint the accurate historical picture as opposed to yours which is using hyperbole to generate a reaction.
https://www.sfenvironment.org/how-can-i-avoid-flame-retardan...
I am not here defending flame retardants, I absolutely believe they don't provide a net positive to the population. Your lack of information and hyperbole is what I am after. These were not "scams". Your SF website does a good job of leaving out California’s TB117 which is one of the pivotal laws that created widespread adoption of retardants in items. There were also some prior Federal laws but TB117 is seen as one of pivotal ones.
Now, I don't know the history behind the chemical manufacturers and if they were behind the fear mongering but there absolutely were tragic cases that moved the nation to implement these laws. It was not just a "scam" that gets added to everything.
They should make flame-retardent bedware and non-flame-retwrdent bedware, and should be legally obligated to disclose every flame-retarded chemical and daily expected daily exposure levels on the tags and box.
when I have done readings before I honestly could not find note of regulatory capture but sometimes these things get muddied with history. Saying its a scam is just hyperbole. Most of the laws on the book are tied in time to some fairly large (100+ person) fire death events. There is a reason those laws were created and we were still in a period of time where chemicals could solve all problems.
Problem is, in capitalism's endless march towards ... well, who knows what, precisely ... companies began to make bedware out of synthetic fabric because it was (a) cheaper to make (b) allowing lower retail prices and potentially (c) higher profit margins. There's also some sense in which synthetic fabrics can be longer lived than non-synthetics.
Once this stuff was out in people's lives, we realized that there was (at least) one downside: these fabrics also ignite much more easily than non-synthetics, and when they do, they generate flame which spreads a fire even more rapidly.
One option would have been to just ban any fabrics that ignite more easily than (say) cotton. That would have been cast by some as a move against the interests of lower income people (not necessarily incorrectly).
Another option would have been to just leave things alone, and let the people who choose to buy synthetic bedware sans flame retardants deal with the consequences themselves. Alas, that's not actually how our society works. When your neighbor's house goes up in flames because of their bedding choices, you still want your fire department to show up and get things under control, lest you lose your home too.
So .. we set standards for how much and what types of flame retardants were acceptable (standards that are subject to and have been changed over time), and let people continue to buy synthetic bedware (and furniture and clothes and ....) all of which contribute to the fuel load should a fire break out.
I am a firefighter (II), and the increase in the speed with which homes can now be fully engulfed because of the decline in the low of low-flammable materials and the rise of synthetics is utterly terrifying.
Sadly we both live in the real world. "A small fraction of the recommended limit" is perfectly acceptable.
This assertion cannot possibly be correct, regardless of the study
I shouldn't be too cynical, but it's a reminder to be skeptical, always.
https://arstechnica.com/health/2024/12/journal-that-publishe...