Hacker News Reader

Comment by stfp | Hacker News Reader

Our whole understanding and approach with regards to risks from industrial chemicals is so completely off...

We keep churning out new compounds (~2000 every year??!) or new use cases for existing ones at insane scale or speed because "most of them are safe". Except the ones that aren't cause a wide range of damage, take 15 years to identify as causes, 15 more years to ban, 15 more to leave the ecosystem if they do at all, and are often replaced by minor variations that haven't been yet proven dangerous.

It seems to me these things should be considered unsafe by default, tested at small scale for years, proven to be bio-degradable and if used at scale introduced very slowly... which in many cases won't be possible, which will readjust our stance so we only use them when absolutely critical and necessary vs. eg. putting teflon or PFAS in literally everything because why not.

The other thing is individual accountability: chemical engineers who facilitate this insanity should be held responsible. Putting lead in gasoline and CFCs in the atmosphere used to get you medals, now similar "inventions" should get you a jail term proportionate to the damage caused.

View on HN

> It seems to me these things should be considered unsafe by default, tested at small scale for years, proven to be bio-degradable and if used at scale introduced very slowly...

It sounds like you are describing the Precautionary Principle, which is unfortunately made light of here due to its impact on GMO food adoption in the EU. I do agree with you, but this idea seems to go against the hyper-growth form of capitalism which is generally accepted as the only option in the USA.

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

https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=LEGISSUM...

>The other thing is individual accountability: chemical engineers who facilitate this insanity should be held responsible.

How would they be expected to know what the results of their chemical will be? Drug companies focus exclusively on biological effects and can't even get drugs to regularly pass phase 1 trials, and you're asking for a single chemist at Industrial Sealants Inc to do what entire pharma companies can't? Putting lead in gasoline sounds dumb, but so does putting mercury in vaccines. The only way to tell the difference is to do studies, which would take teams of people. It is absolutely not possible to expect individual chemists to accurately predict the biological consequences of their products.

Maybe we should have tiered rules for how much testing is done based on the number of tons that are going to be produced and how they are being disposed of.

(And if we already do, then the rules for the higher tiers need to be tightened.)

Given the extreme differences in badness per ton I'm not sure if that's a reasonable metric. Maybe instead we could mandate yucca mountain level storage for everything that hasn't been evaluated, and then let companies decide for themselves whether or not they will be making enough of the stuff to justify studying how to dispose of it more cheaply.

It might be possible to establish categories where if your molecule has XYZ properties it falls into this bucket. I think a category for really novel molecules that haven't been evaluated is a good idea with tight restrictions on how much can be produced.

There should also be a presumption that minor edits to known dangerous molecules should be presumed dangerous without testing.

Be careful about diluting responsibility---if everyone can point to someone else, no one will make a hard (and not immediately profitable) decision.

In all of those studies, etc., there is someone at the top who does, and should, make the ultimate decision, just like when building a bridge, there is an engineer who puts his career on the line.

If you're going to assign responsibility, you need to assign it to people who know the right decision to make. If you assign responsibility to a janitor or a chemist for any torts set against the company, then janitors and chemists will simply quit working there. The only person who could begin to approach responsibility would be an executive with the authority to organize and pay for clinical trials and toxicity studies.

> authority to organize and pay for clinical trials and toxicity studies.

But further to that, if said executive does follow regulatory guidance and accepted practice but the product has a problem, should they really be held individually responsible?

For the clinical trials example, if a trial fails to turn up a significant problem, is it the exec's fault or the regulators? The regulators or the legislative body?

Oftentimes, regulatory bodies make companies immune to lawsuits in exchange for compliance. That leads to there being no incentive against creating products that are compliant, yet dangerous. It might be a good idea to reform the system so that harm is the ultimate decider.

Putting mercury in vaccines was also dumb. It is impossible to quantify the damage just from the direct consequence of people beginning to avoid vaccines because of it. They don't put mercury in vaccines anymore.

> so does putting mercury in vaccines

Putting a known biologically tested and FDA cleared preservative into vaccines is not "dumb".

Having to remove said preservative because because people committed medical fraud (and later lost their licenses) to create fear about it and because a bunch of people picked up on it as a desperate, now-negatively validated reason for their child having autism?

That's dumb. But the manufacturers did it anyway because people are dumb.

You missed the phrasing slightly. I said they both sound dumb, and it's only testing that can distinguish them. In an alternate universe, it's leaded gasoline that is safe and mercury in vaccines that is poisoning people.

But in this universe, you can eat a can of tuna and have both!

>The other thing is individual accountability: chemical engineers who facilitate this insanity should be held responsible.

Hey, I wanted to point out for your future reference that you really should be blaming the chemists. They’re the ones who design the product and intentionally choose this. Chemical engineers are the ones who do the manufacturing of chemical products. I have CE friends and they’re the ones in the plants keeping processes running

Thanks, good clarification!

The precautionary principle, which is what you describe, is in greater use in Europe.

https://ec.europa.eu/environment/integration/research/newsal...

Underlying this all is a sort of scientism issue. On the one hand you have a culture that assumes the framework of a scientific framework is "truth" and that the lingua franca of risk should be demonstrable evidence. On the other is a culture that denies the realism of the scientific studies, and sees risk in misrepresentative evidence.

Generally speaking, the latter camp gets labeled as "antiscientific" by the former, and the former gets labeled as "elitist" or having conflicts of interest. It's part of a theme that seems to be popping up a lot in the last several years or so in numerous settings: which is worse, lack of evidence or misleading evidence?

This example with the salmon is interesting to me because (if I'm understanding it correctly), the compound in the tire isn't the problem, it's a byproduct of tire breakdown. So I could see a chemist in the lab sitting there saying "ok we ran our tests on X it's fine" without understanding that in the real world, X reacts and breaks down into other compounds that are problematic. It's kind of a prime example of how the scientific model isn't the correct one to be using in the first place. Within the bounds of the model, X is fine, but the model is wrong as a representation of reality, so the conclusions are wrong.

I think underlying a lot of this are differing ideas of what is "scientific", or what constitutes the universe of risks. I don't see this dynamic changing until certain cultures surrounding scientific authority change.

>It's kind of a prime example of how the scientific model isn't the correct one to be using in the first place.

What's your proposed alternative to science for determining the products of tire breakdown? Asking psychics? Polling facebook moms? Voting on it in Congress?

Unfortunately, alternatives to science are being pushed right now in the current culture war. Google “other ways of knowing”.

Worth pointing out just because your appeal to absurdism (which most of us would agree with) is losing steam in the marketplace of ideas.

Look up scientism.

The whole point is we need to accept that no single method will give us this kind of visibility and understanding, and given that fact we should be more careful about certain things, especially things we just seem to keep miscalculating and that have long-lasting, terrible effects.

That's actual science vs. commercial/industrial marketing masquerading as science.

The scientific method enables us to rigorously build models.

But it takes effort, so the model only expands in the directions where a person chooses to the do that work. Scientific models are a communal human artefact, flawed with the bias and inattention of those that built them.

The is a lot we know about the world, from history, from our experience, that we are not able to express with enough rigor to include them in a scientific model.

So we can be confident that our scientific understanding of any new compound, will not gives us a complete understanding of the impact of that compound in the world. From history we have a partial map of our ignorance. We know from past experience that there are realms of unknown - ecological effects, decomposition products, birth defects, health impacts - associated with new products. We know there is a very big difference between safe and no known adverse impacts

Rather than an alternative to science, we need a more vigorous meta model, better appreciation of where the model ends, and a better model for how we should act given our ignorance. I don't think we lack the ability to do this, it's a matter of will and power.

>The is a lot we know about the world, from history, from our experience, that we are not able to express with enough rigor to include them in a scientific model.

Of course, gut feeling and "pontification-based" methods still have to prove themselves effective. You would need to have one group pontificate about the future using your meta-level thinking, and another group just focus on the concrete-level, and demonstrate that the proposed meta-model would improve the situation. A system to figure out what scientific discoveries were going to be made before they are made would be nice, but you can't just pick an oracle and say you've found it.

Maybe what I wrote was misleading or vague. What I mean is that to do a study, you have some model. It's implicit in the way the data is collected, whether that be experiment or observational study. But it's a model with assumptions.

The problem is that all models are wrong to varying extents, but often in the scientific community the wrongness of models isn't recognized. You estimate your effect size or the p-value or posterior predictive p or whatever it is, but all those themselves are some estimates based on assumptions, and there's always some probability of being radically incorrect.

My concern isn't with science per se, it's that science as a community is often too arrogant and underestimates the extent to which it is likely to be incorrect. I just think critics need to be given more respect, especially when the argument against the critics is "we did a carefully controlled lab study under isolated conditions."

Maybe there needs to be a greater meta-scientific perspective with these things, like some policy that explicitly asks "how often when these types of evidence are submitted is something wrong?"

> It's kind of a prime example of how the scientific model isn't the correct one to be using in the first place

Do you mean "this particular scientific model" (which lacks the analysis of byproducts) or the scientific method? If the latter, what other method would you use instead?

The issue is that you can't prove a negative, not even the scientific method can achieve this. So you can prove harm, or benefits, but you can't prove an absence of harm.

So while the scientific method is fine, applying it without an abundance of caution commensurate to the potential negative impacts leads to poor outcomes in the long run.

You can't truly "prove" harm or benefits either they just happen to be things you can very easily make the error bounds minuscule on. Likewise the problem with "absence of harm" is making the error bounds small is extremely slow and costly so we avoid it. The scientific method (or any method to this point) never ends up "proving" anything, you always have to set what error bounds are acceptable.

Which leads to the real overall point I suppose - is there something better we could be using as "just never make anything new" isn't free of problems either so it's a question of "what's the best method" not "what method can guarantee something" (none so far can).

The solution to poor results from scientific work is better scientific work, not throwing out the scientific model.

The issue described is akin to testing some new code with unit, integration, and end-to-end tests, and running it for some time in a dev environment, but not running it in a QA environment that mirrors production, and not monitoring it for issues when it does get deployed to production.

The response there wouldn't be "the test/devops model isn't the correct one to use in the first place", it's that the testing and QA was inadequate because it only looked at an idealized model of reality and didn't consider "where the rubber meets the road" (literally). Testing something in isolation is insufficient, since you need to test the downstream effects and potential complex interactions with everything else in the environment.

There is no testism problem; only a poor testing problem. Unless, as you say in your last sentence, one is just arguing about definition semantics - e.g. if "QA" falls under "testing", or, here, if "real-world observation" falls under "scientific testing". I'd argue both do. I think your concerns might be better phrased as "lab model" and "lab-ism".

> the scientific model

What you described isn't "the scientific model" but merely one flawed study. Science is the process of learning about the world. Scientific paradigms change. Right now we seem to have a statistical frequentist paradigm in most fields, but it's shifting towards Bayesian. The business science field still seems stuck in the case study / anecdote paradigm, though it's also shifting.

If the science doesn't seek to learn about "the real world" then it's bad science.

> It's kind of a prime example of how the scientific model isn't the correct one to be using in the first place.

This seems to be a category error - after all it was science that found the problem also.

I think you are poking around the edges of something real here, but suspect the thinking is muddy still and falling into a false dichotomy. There is a lot of confusion about the role of science and what it can and can't do, but I don't think it maps at all purely onto the "culture wars" as you are seeming to suggest.

Probably the best known example of this (who I assume you are referring to) is incidentally responsible for both leaded fuel and CFCs:

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

there was a fun dollop episode about him if you want a little levity while listening to an absolute ecological monster... https://soundcloud.com/the-dollop/393-thomas-midgley

He also appears on this amazing list: https://en.wikipedia.org/wiki/List_of_inventors_killed_by_th...

Agreed. The same thing happened with BPA — banned in toys and bottles after decades of harm to be replaced by at least 40 other chemicals that haven’t proven to be unsafe — yet.

>It seems to me these things should be considered unsafe by default, tested at small scale for years, proven to be bio-degradable and if used at scale introduced very slowly

That's no way to create shareholder value.

Socialist countries (without capitalism) have a worse track record.

Oh? I'm sure you have evidence for that, I'd like to see it.

Google Volga pollution.

Russia hasn't been socialist for 30 years. If we're dredging up the past, look up "Cuyahoga River pollution". Spoiler alert: it was addressed through government regulation, not the free market.

> eg. putting teflon or PFAS in literally everything because why not.

But it wasn't because "why not" it was because there were very concrete benefits. This is why the fundamental issue is policy, not science - the science can only help you understand a situation better, it can't answer "what is the right amount of risk to take".

The precautionary approach you suggest isn't a new idea, but it has very real negative economic effect at minimum, and a complex relationship with other issues. Collectively, somehow (e.g. legislative bodies) we have to attempt the impossible task of finding the "right" position on the scale between don't-trust-anything-new to anything goes. Clearly both these extremes are problematic, but balancing them is really hard.

> chemical engineers who facilitate this insanity should be held responsible.

It took decades for a lot of people working on this problem to identify the cause of the salmon deaths.

Do not expect the individual chemical engineer to be able to come close to knowing how safe a chemical is.

This has lots of practicality problems, obviously.

The individual accountability is certainly tough, particularly since we learn about impacts only over time. And discoveries like these are studied, applied and distributed by ultimately hundreds to thousands of people.

I guess my proposal is that new chemicals need to be tested at a 100x dose on 15 different species, up to primate. On lesser species, things such as number of offspring need to be measured (reproductive health seems to be the first to go a lot of times). On primates, we'd look for sperm count relative to a control group (again the first thing to go).

If any effect is seen at a 100x dose (of the amount expected to be ingested) the chemical would be dropped.

IDK, I think disposal and recycling is really where we need to focus.

Some chemcials are damn useful while being toxic. I don't think the right thing to do is banning all toxic chemcials because they might cause environmental damage. Instead, let's focus on figuring out safe handling upfront.

A good example of this is batteries. Super toxic, super useful.

Damn useful but at what cost? Unfortunately we often don’t know the cost until 10 to 50 years on. It’s impossible to close Pandora’s box once it’s open. Look at DDT https://www.latimes.com/projects/la-coast-ddt-dumping-ground... and mercury in the food chain and and and and

I think GPs point is to keep them contained while in use, with a plan to keep them out of the environment once you're done with them.

Batteries are a great example. We really need good battery recycling. We also really need good batteries, since it looks like they'll play a major role in preventing catastrophic global warming.

Strong acute exposure and slow chronic exposure will show different effects overall. But yours sounds like a decent heuristic as a first pass.

No.

First, what is a 'dose' for a new chemical?

Second, there are all sorts of critical things that if you took 100x the recommended dose you would see a significant negative physiological effect. Like salt. Or water.

This is a far more nuanced problem that "let's just shoot up some monkeys with a gallon of the stuff and see if their nuts shrivel up". In fact, that sort of reductionist approach is why we have this problem.

> First, what is a 'dose' for a new chemical?

The dose is the amount you expect an animal to ingest in its lifetime.

> Second, there are all sorts of critical things that if you took 100x the recommended dose you would see a significant negative physiological effect. Like salt. Or water.

Uh, lol, no. 100x and 1000x multiples are standard EPA testing procedure, so you're not disagreeing with me, you're disagreeing with the EPA, and I'm gonna stick with the EPA on this one. [1]

> This is a far more nuanced problem that "let's just shoot up some monkeys with a gallon of the stuff and see if their nuts shrivel up". In fact, that sort of reductionist approach is why we have this problem.

That's a disingenuous reading. A 1000x dose of these chemicals would be less than a miligram, which is plenty to do harm. Even lead can do brain damage at the ppm level (some chemicals are even ppb). Don't be too embarrassed by your comment though, we all start somewhere.

1. https://en.wikipedia.org/wiki/Reference_dose

Having a 100x multiples for clearance leaves a large safety margin but again take water, 0.5L per day is good, 1L won't kill you but above 8L people have known to die. That's only 16x. People also get get cyanide poisoning from raw almond smoothies, alcohol poisoning from binge drinking and caffeine poisoning at and also less than 100x regular portion.

Radiation too, 2-30 is pretty typical and no statistically significant results but bring that to 300 and you get measurable increase in cancer rates.

Definition of dose (Websters)

(Entry 1 of 2) 1a : the measured quantity of a therapeutic agent to be taken at one time b : the quantity of radiation administered or absorbed 2 : a portion of a substance added during a process 3 : an amount of something likened to a prescribed or measured quantity of medicine a daily dose of hard work a dose of scandal 4 : a gonorrheal infection

So the common definition isn't the one you are using. But if you pick your own definition, you get to be right.

Being able to write with accuracy and precision is hard. Don't be too embarrassed by your comment though, we all start somewhere.

There's millions, maybe billions of different chemicals created by biological processes too.

But I get your point and I think there's something to be done advising caution with certain classes of chemicals. For example, ones containing Bromine and Fluorine have a much higher chance of being dangerous because their bonds are so strong

In the EU new substances seem to need registration at least. It's called REACH regulation:

https://echa.europa.eu/regulations/reach/understanding-reach

The problem with the Precautionary Principal (that you seem to describe) is that it doesn't account for opportunity cost. New compounds are developed for a reason and to contribute some value to society. That could be a very small convenience optimization, or could be for a life-saving purpose.

While we're waiting for some compound to be deemed 'safe' for some particular use, society is missing out on the benefits of that use.

The COVID vaccine is a very timely example of this. It/they haven't gone through nearly the amount of testing and trials that are normally required. But we're going ahead with it anyway, since waiting would cost more lives. This is a judgement call that some set of fallible humans has made. Easy in this huge, public, major case. But for the thousands of compounds developed yearly, who decides this, and at what cost to society in lives, quality of life and resources?

And this doesn't count the resources that society has to spend to make the determination of safety, which have their own opportunity cost as well.

Taking a step back: here, incidentally, they were trying to basically facilitate driving, which is not really in the interest of society / life on the planet ¯\_(ツ)_/¯

Maybe the vaccine is actually a bad idea? I don't think so.

But even if it turned out to have terrible side effects, that would only affect people (the people alive now and taking the vaccine) and not the environment (all the humans and other animals, present and future).

How does one individual remedy this for themselves?

Exactly why the covid vaccine is so risky... we won't know long term effects for years.

Reply on news.ycombinator.com