Comment by abeppu | Hacker News Reader

Comment by abeppu | Hacker News Reader

A bunch of comments here are roughly "there's not great evidence that these are harmful" -- but is that the right standard? Or should there be an obligation to provide evidence that something is "safe" in order to use it in some consumer applications (like food packaging)? It's tempting to say that we should conclusively establish that something isn't harmful before we use it absolutely everywhere ... but what standard of evidence would be both achievable and ethical?

Because generally people aren't getting acutely ill after eating food from a plastic package, we're left with the possibility that accumulative impacts over years might be harmful -- but it doesn't seem feasible to run long term studies where a treatment group is exposed to plastics for decades and a control group is not. It hardly seems achievable to do correlation studies, because you often don't know what's been in the packaging for all the food you've consumed, which may not even be in your control.

Your second paragraph is a very good argument against requiring that everything used in food packaging be shown to be safe before it can be used.

Can we reasonably run such a study to prove that wax paper is safe? What about plain paper? Do we just require all food producers to use no packaging at all until these controlled longitudinal studies are completed? If we allow them to use packaging, how do we define in a principled way what packaging is allowed while there are still unknowns?

One possibility would be to say that if something is currently widely used in some significant (think 5%) portion of the industry then we allow it, but that has two problems: First, it wouldn't exclude phthalates anyway, so it doesn't address the current concern. Second, it might exclude future packaging materials that we think might be safer than our current materials but which have yet to be tested.

> One possibility would be to say that if something is currently widely used in some significant (think 5%) portion of the industry then we allow it,

We could say "if it was used in the ancestral environment for similar purposes, and is not known to be harmful". Much of that stuff is bad, but our bodies have mitigations for it. (e.g. tannins leached from wooden bowls: harmful in principle, but we can metabolise them so it's fine)

Ideally rigorous testing should happen before something becomes common, but it isn’t the only option. Nothing stops you from a temporary ban say 20 years to allow for testing. Doing this for every chemical at the same time is impossible, but we can also cycle through existing chemicals.

Hypothetically, if phthalates exposure was responsible for 5% of cancer deaths that’s 30,000 dead Americans per year and we might not notice without a ban. It’s unlikely for this specific chemical to be that bad, but we don’t know.

Cellulose and algae are considered safe for human consumption and are also biodegradable; but is that an RCT study?

CO2 + Lignin is not edible but is biodegradable and could replace plastics. "CO2 and Lignin-Based Sustainable Polymers with Closed-Loop Chemical Recycling" https://onlinelibrary.wiley.com/doi/10.1002/adfm.202403035

What incentives would incentivize the market to change over to sustainable biodegradable food-safe packaging?

> Can we reasonably run such a study to prove that wax paper is safe?

We don't need to. We have a "Generally Recognized as Safe" standard that is well known and widely applied.

> widely used in some significant (think 5%) portion of the industry then we allow it

How about we just label things so consumers know whether or not the packaging contains phthalates? That way the market can decide if they want it in their package or not.

Labeling isn't bad, but it doesn't scale. It works for e.g. allergens because people who are allergic are aware of the rather rapid harm done if/when they consume allergens.

> but it doesn't scale.

Why not?

You don't even necessarily need to put all disclaimers directly on the package. Just put some kind of reference number on there, like an MSDS number, and let users go look up all the packaging information online if necessary.

The manufacturer has this information. They must in order to produce the product. The requirement would be they just have to publish it now when used to wrap food.

How does that not "scale?"

It doesn't scale in the sense of "How can I determine which of the 27 things on the label are likely to be bad for me?"

I'm not convinced labeling regulations work. Producers don't label accurately, and consumers aren't realistically in a position to make the choice being described. I live in California, where we have Prop 65 warnings on everything. The motivation there was quite similar -- people should be aware if a product or space is exposing them to substances which can cause serious harm. But I have never met a single person who refuses to enter a space with a Prop 65 warning because you basically couldn't function in society that way. When you buy a product with a Prop 65 warning, it's generally not actually feasible to know which chemical(s) (if any) prompted the warning. (Some companies may put the warning on all of their products even if only some of them actually contain such a substance.) Further, labeling for the nominal presence of some substance is not enough to make decisions -- you need to know what your actual exposure is likely to be. The system ends up being pretty useless.

In addition to what others have said, note that companies already voluntarily label their products as "phthalate free", so the market is already positioned to decide.

In the absence of concrete evidence of harm from phthalates I'm not convinced mandating yet another label would do more good than harm—it would potentially just further reduce the signal that labels provide by watering it down with yet another mandatory compliance label.

You can't prove things are safe, at best you can say harm hasn't been observed under these test conditions. In particular things with accumulative impacts may not show any signs of danger until millions of people have been using them for years. The only valid approach is to collect data as we go and make iterative changes as our understanding improves, we can not count on getting things right at the outset.

> You can't prove things are safe, at best you can say harm hasn't been observed under these test conditions.

Are we doing that?

Yes.

1. It's a lot harder to prove something is 100% assured safe then the other way around

2. Every substance/material/etc that is investigated and banned incurs overhead on taxpayers and industry directly.

What you're asking for would be extremely prohibitive to getting anything done at all.

To be clear, I'm not asking for anything. I'm raising a question, because I'm genuinely conflicted. It seems intractable to test everything for long enough to know that it's safe. On the other hand, the 20th century had some good examples of cases where we understood that something harmful only very long after it was in the environment (e.g. leaded gasoline).

Also to be clear, at no point did I say 100% about anything. Please do not attempt to put words in my mouth, or to fight a straw man. We can all acknowledge that empirical studies don't give 100% certainty about anything.

But it does also seem fundamentally problematic for us to make substances or materials ubiquitous in our environments faster than we can study them. We then also arrive at a methodological problem, where, for example we cannot meaningfully study whether long-term low-level exposure to PFAS are causing impacts on fertility in part because there is not a population that hasn't already had that exposure.

It doesn't seem like there's really a right answer. There is such a wide spectrum between "yolo everything into the market and find out later whether it is dangerous" and "don't do anything unless it's proven safe." I personally feel we can afford to throw innovation under the bus a little more than we do today, in exchange for more safety assurances, but reasonable people can disagree about it.

I think a lot of people might look at this kind of regulatory intervention and think it might be overstepping because of a lack of evidence. Like, are they acting prematurely? The answer is a definitive, no. Feasibility is irrelevant. It’s about scale. Once a substance is handled and consumed by no less than 100% of the population, with no viable way to replace these substances without a ton of work, it’s way too late.

There are legitimate scientific concerns that we are currently damaging the human gene pool.

Good regulations work well. It took a global effort to ban the CFCs that were destroying the ozone layer, and it’s completely repaired now.

I don't know why this got down-voted so much. Maybe scaling up too fast is the biggest part of the problem.

Perhaps, putting it in a framing familiar to tech folk, if we have a new version of some important component of our application, "push it to 100% in prod before doing any data analysis because it hasn't been proven to be harmful to engagement" or whatever, probably wouldn't be acceptable to most of us. Maybe you do an A/B test and initially launch the new version to 5% and watch some stats, and if it does well, you scale up from there, incrementally expanding the audience and doing more data analysis. But that analysis depends on their being a population which you know hasn't received the new version.

The problem happens when a switch to selling stuff in plastic/plastic-lined packages is driven by parties who don't have the interest or the ability to release to only a slice of the population, collect the data or do this analysis (and to roll back when the data shows a negative impact).

In the US, at least, food regulations (and food packaging regulations) are relatively recent. And when they started, instead of requiring everybody prove that everything they’d been doing was safe, they came up with the idea of identifying substances or practices as Generally Recognized as Safe ( https://www.fda.gov/food/food-ingredients-packaging/generall... ), and then requiring safety analysis on new substances or practices. And things can lose their GRAS status if new evidence turns up.

How do you propose proving a negative?

With modus tollens

https://en.wikipedia.org/wiki/Modus_tollens

I think the civil burden of proof should apply here. Upon the preponderance of the evidence. Given so many studies that exhaustively show correlative effects of phthalates with metabolic and endocrine disruption, especially strongly during developmental stages, I think even that burden of proof is met.

Sure, you can't prove something doesn't exist, but you can make a reasonable determination that given exhaustive study, given correlative negative effects can be ascribed to no other known cause, and given that this is natal and pediatric health especially at stake, we should say until such time as a causative link is proven to lie elsewhere these chemicals should not be used in consumer goods out of an abundance of caution.

The redirection of that claim would be "these profits are more important than moving to protect bodily health especially for those unable to protect themselves, given what we currently know."

Honestly, no knowledge is perfect, but upon the balance we must rest our sacrosanct right to health.

>is that the right standard? Or should there be an obligation to provide evidence that something is "safe" in order to use it

I agree but the FDA might not have the legal authority to ban chemicals that are already in use that haven't been proven to be harmful. That is the case with the EPA IIRC. Although I think there actually is a lot of evidence pointing towards phthalates being harmful especially in utero.

For me, the standard is: if the material was used pre-industrial age, it's far more likely to be "safe".

Any person could overintellectualize this point and find exceptions. But it's just common sense.

I agree. I'm reminded of "the naturalistic fallacy" which asserts

> Things are not necessarily safe just because they're natural

Whatever "natural" means. It's not wrong exactly, but I think it distracts from some adjacent, non-fallacious inductive/anthropic reasoning:

> Things that have not harmed your ancestors are less likely to harm you than a novel thing which your ancestors did not come in contact with.

The former (safer) thing is more likely to end up with the label "natural" than the latter, newer thing. So "natural" ends up correlating with "safe" more often than chance would have it.

Why do we think our ancestors were not harmed? do we have any reason at all to think that?

Because we exist. Admittedly, "healthy enough to procreate" is not a high bar, but it's pretty good in comparison with no bar at all.

Yeah, there are cases where they were unaware of the harm--like lead. We should absolutely be finding better ways to identify and mitigate harms of that sort. But in the absence of reliable information from such efforts, natural==safe is a viable heuristic.

And I'd say that our scientific establishment is not sufficiently hardened against attempts by companies to tamper with scientific consensus in cases where that consensus might be bad for business--so an absence of reliable information is indeed our reality.

I think natural==safe is a terrible heuristic for the exact reason you said.

Healthy enough to procreate is a low bar, and many natural things provide a benefit in terms of procreation but detrimental to health. Almost everything in nature is a trade-off between how much it'll harm you and how much it'll help you procreate.

This reminds me of how whales get the Bends from diving despite 50 million years of evolution. People question whether it causes them pain, and it stands to reason that it would cause them pain to prevent them from diving more than necessary. Nature isn't inherently kind, it is results oriented.

You can make a valid argument that avoiding something entirely is usually safer than something, as few things provide a protective effect. However, when you talk about substitutions, all bets are off.

> many natural things provide a benefit in terms of procreation but detrimental to health

Maybe... but many more either kill you or they don't. Lets take your whales-getting-the-bends example. That gives us a dimension: pressure. Suppose you're whale.

- Let x be a randomly chosen pressure

- Let y be a pressure that your parents lived through

You have to spend a few minutes at x or y, which pressure do you chose? y would be a safer choice.

The same goes for potentially hazardous chemicals: the safer bet is to pick the one that comes with evidence that things like you can survive contact with it, and that'll be the one that was in your environment when you were born. Otherwise you wouldn't have been born there.

If somebody wants to sell us something that our parents never came in contact with, the evidence that it is safe should be stronger than the effect described above. I'm not a statistician, so I can't quantify that effect, but it exists, and the naturalistic fallacy distracts from it.

Sure, there is a precautionary principle. Change nothing and you won't end up worse than before.

It depends on how risky you think the baseline factors are, and how risky you think the average new factor is. It also depends on how likely catastrophic risks are.

We know people don't drop dead from most things, so risk is generally at the fringe. Long lead times and low impact.

You then have to consider how many safe products you are missing out on to avoid the bad ones.

I think the the current paradigm is pretty reasonable. Screen for major known risks and then monitor and study

I think that breaks down when you compare Neolithic or pre-industrial man to now.

Just because I can: see pre-industrial Romans and their lead acetate. Or Egyptians with their kohl eyeliner (also lead). I prefer strong regulatory bodies that methodically analyze products available to general consumers.

I think you can have both.

I also want strong regulatory bodies doing those analyses, but I also would like to see them using reasonable Bayesian priors on safety. I think assuming the Bayesian prior that: compounds that occur naturally in our environment at concentrations and particle sizes similar to what we'd be exposed to in a proposed new product are likely safe to use. The opposite just is not a good Bayesian prior, and such compounds should require greater assurance of safety before being allowed to be used in novel ways.

The above is particularly true for organic compounds.

I'd also like to see regulation of analogs like they do with drugs. Swapping out BPA for other analog chemicals with very similar shape/composition is something most people with a decent background in biochemsitry would be extremely skeptical of. Let's say they banned BPA entirely. I'd like to then see regulators step in and ban analogs by default for the same use until proven otherwise.

It doesn't require overintellectualizing, the exceptions are really obvious, like lead pipes. Generally speaking public health knowledge was extremely low in the pre-industrial era, compared to what we have now, so I don't know why we would use this as a standard. I wouldn't use pre-industrial standards for dealing with sewage or prepping for surgery, either.

Except that we've had time to observe the effects of stuff between then and now. Which is why the I don't think the parent is lobbying for a return to lead pipes.

Overintellectualizing

Don't try to pre-dodge by accusing anyone who replies and disagrees with you of overintellectualizing lol

As a proxy for "safety", look at average life expectancy. Even excluding infant mortality, life expectancy has increased by like 20-25% since the year 1900 (for people who reached age 20). We're doing something right in the modern age, and suggesting that we revert to how things were in the 1800s is not common sense, it's being a luddite (imo).

https://lh5.googleusercontent.com/qtYQp1x-ZF9iXc-zVh7Kg2xJBX...

Something made in a factory is less likely to be good for you than something made by a tree and stone tools.

The industrial age has brought many good things and many bad things. To acknowledge the bad things is not to discredit the good things. It's quite simple.

Higher life expectancy is a good thing. Infertility, cancer, and obesity are bad things.

More of each was brought by the industrial age. I'm not saying the industrial age as a whole was bad.

Again, all of this is just overintellecualizing it.

Lead glazed pottery?

I think you need to flesh that standard out a bit. Maybe something more like this:

> If the material has been in continual use since before the industrial age and we have yet to find any evidence of harm, it is far more likely to be safe than newer materials with which we have fewer centuries of experience.

Our ancestors used all kinds of things that were terrible for them and it doesn't require overintellectualizing to find exceptions, they're everywhere. But this late in the development of medicine it is fairly reasonable to suggest that if there's a long-standing tradition of using a material and we have yet to find that it causes harm, it is more likely to be safe than something new.

So why was the expected age of death lower in the pre-industrial age than it is now if the materials used then are safer than modern ones?

> if the material was used pre-industrial age, it's far more likely to be "safe"

So much lead and mercury was used in the pre-industrial age.

Why do you think it is common sense? I'm genuinely curious why you think the pre-industrial environment was safe or safer.

Reply on news.ycombinator.com