HN Books @HNBooksMonth

The two iconic texts are:

- "The Collapse of Complex Societies" by Joseph Tainter (Amazon link: https://www.amazon.com/Collapse-Complex-Societies-Studies-Ar...)

- "Normal Accidents" by Chuck Perrow (Amazon link: https://www.amazon.com/Normal-Accidents-Living-High-Risk-Tec...)

Tainter deals directly with the Roman Empire, but the nutshell is the cost of complexity begins to outweigh its returns, requiring more and more resources just to maintain the status quo, until the entire thing becomes weak and susceptible to failures large and small.

- "Panarchy: Understanding Human and Natural Systems" (Amazon link: https://www.amazon.com/Panarchy-Understanding-Transformation...) is also a fantastic book drawing from ecosystem science and proposes a general model for this. It's pretty well accepted in ecological circles but has been criticised for a lack of empirical data. The general model is the same as Tainter's though.

- "The Logic of Failure" by Dietrich Dorner is also a classic! (https://www.amazon.com/Logic-Failure-Recognizing-Avoiding-Si...)

Well worth reading all of the above.

There is some danger in publicly adopting a "science has become religion" position. Just as "trust the science" can be used as a rhetorical weapon, "science has become another religion" can be too .. and usually to more dangerous effect because a lot of what is essentially known bullshit (see https://www.callingbullshit.org) gets accommodated by the latter compared to the former.

There are reasonable accepted empirical approaches in accepted use today. Any team that follows these methods and reports honestly can be trusted more than a team that doesn't.

Also "trust" is a loaded word .. but mostly we just mean "predictability" - ex: to "trust someone" means to be sure that if you know what they've said, you can predict what they'll do. Much of empirical science is about offering up data and models that aid such predictability, so declaring "trust the science is extinct" is an outright rejection of these empirical methods (ex: randomised control trials) that have taken a long time to mature and take root as standard practice.

What is needed though is to be able to separate the science from policy making. As Dietrich Dörner has shown in "The Logic of Failure" [1], folks in the hard sciences don't fare very well when policy making in systems with complex causally interconnected parts is tasked upon them due to learnt heuristics that don't fare well in that world.

Let science do its thing - which is inform and educate. Let policies be made by those with more full understanding of the system into which changes need to be effected.

[1] https://www.amazon.com/Logic-Failure-Recognizing-Avoiding-Si...

⬐ paganel

At the beginning (early March 2020) we were told that the lockdowns would last for at most a month or two, in order to "flatten the curve", at least that's what "science" was telling us. I know I believed in that, even though I'm no scientist.

One year on and that "prophecy" (because that's what that was, just a prophecy) turned out to be pretty damn false. Nevertheless, we still let the people/scientists who deceived us back in March 2020 to take "science"-based decisions that will affect the lives of hundreds of millions going forward. Nobody will be held responsible, apart from a few politicians.

⬐ cjfd

This idea of 'flatten the curve' is pretty easy to prove false with a back-of-the-envelope calculation. Like: how many people that simultaneously have corona can the hospitals support? From that it follows how much time it takes for everyone to get corona. When I did that calculation back in March I found that it was a few years. So, the consequnce of 'flatten the curve' is a few years of lockdowns. A very bad conclusion....

⬐ ntsplnkv2

This doesn't work because this assumes no learning takes place.

As you can see we now have vaccines - it would not have been a few years of lockdowns.

⬐ trashtester

The story told to the general public was obviously not the story to be closest to scientific truth, but instead the story that was thought to lead to the least severe outcomes. Nobody really knew the real fatality rate last year, the numbers could range from more than 10% to about 0.1%. If they turned out to be about 10%, quarantene would be essential in stopping the spread of the infection. Also, higher fatality rate would mean that the infection rate would be relatively low (as the number of dead were known), meaning that it would be realistically possible to exterminate the virus. On the other edge of the spectrum, the fatality rate might be near 0.1%, but the infection rate would be much higher. If this held true, flatting the curve would actually be a good strategy. Finally, if the true fatality rate would be around 1% (roughly the real number), the one would "flatten the curve" until a vaccine could be provided, as if hospitals had been flooded way behond what has happened, the fatality rate could easily be 2-4x higher. In the case of the US, that could mean around 10 million dead by now, if no lockdown or social distancing had been attempted.

Of course, had no action been taken to slow it down, and millions had died last spring (in the US alone), the public opinion would strongly demand a lockdown, as well as become so scared that they voluntarily self-isolate.

⬐ Pyramus

I'm not sure I understand.

New Zealand did lockdowns for a month or two and became completely Covid free.

China did lockdowns for a month or two and became almost Covid free.

Most European countries did lockdowns for a month or two and were quite open over the summer.

So either it was a different prophecy in the rest of the world or the prophecy was correct after all.

⬐ paganel

Other 100+ countries did the lockdown as well and didn’t have the same results as NZ, science would thus tell us that what happened in NZ was the exception/statistical anomaly.

⬐ Pyramus

Okay, so you are saying that lockdowns do work, albeit in exceptional cases.

Then how did the lockdowns work in say New Zealand or Australia, or Singapore, or China? What is the mechanism of action?

⬐ cimi_

Through severe restrictions on travel, people coming in the country and invasion of privacy.

I know Australian people in Europe who haven't been able to visit home in a year. I know second-hand that in China if you had multi-stop flights you needed to quarantine at every stop. I don't know what test and trace measures the AU and NZ governments decided on, but I'm pretty sure that I wouldn't like them if I knew.

And it's easy to say 'well, good, people shouldn't travel now' - but this is not just about holiday. These restrictions keep families apart, make it more difficult for people to care for each other. Imagine someone close to you is in distress and you can't go help them because of this, how would you feel?

I don't know about you, but I would not want to live in China, even it gets to zero covid. And I wouldn't particularly like being isolated in AU or NZ either.

⬐ ntsplnkv2

Fine but this isn't evidence that lockdowns don't work, they clearly do work, it's just that in many places people put their freedom above other people's chance of getting sick and dying in Covid. If it was a higher risk of death I think that calculus would change.

⬐ Pyramus

Completely agree, would not want to live in China either.

My comment was a response to parent, who feels betrayed by science, because lockdowns were supposed to work.

What I will say though is that there is no win scenario. It's lose-lose, either way. A non-Covid world does not exist anymore. Lockdowns come with exactly the downsides you describe. No lockdowns come with an overload of the health care system and high death tolls. Light lockdowns are a combination of both.

Both proponents and opponents share the same motivations by the way - to be able to see relatives, meet other people etc.

⬐ collyw

> Lockdowns come with exactly the downsides you describe. No lockdowns come with an overload of the health care system and high death tolls.

Please stop pushing the unproven narrative that lockdowns stop the healthcare system being overloaded. It's morally wrong to push for such a destructive policy without good supporting evidence.

Sweden, Japan, Uraguay, South Korea. No lockdown. No problem for any of those healthcare systems. It can be easily shown that curves were flattening before lockdown had a chance to take effect.

This is about as close as we will get to a controlled trail on lockdowns.

https://www.medrxiv.org/content/10.1101/2020.12.28.20248936v...

⬐ Pyramus

You are absolutely right, lockdowns are just a proxy, the real factor is behaviour. If everybody behaves lockdown-like, lockdown can be avoided.

You don't need lockdowns to stop the healthcare system from being overloaded, if everybody wears masks, socially distances, agrees to track, trace and isolate programmes, there is mass testing, no incentives for people to come to work sick, etc. One year on we have a pretty good estimate how each behaviour/measure influences the reproduction rate of Covid-19.

> Please stop pushing the unproven narrative that lockdowns stop the healthcare system being overloaded.

I don't understand what exactly is unproven. Lockdowns correlate with behaviour. The Chinese healthcare system was completely overwhelmed, even with CCP spin on things. In Bergamo, Italy, the dead were carried away in army trucks. Look at fatality rates for both countries at the time.

⬐ collyw

> If everybody behaves lockdown-like, lockdown can be avoided.

Speculation without evidence.

Lockdowns are apolitical decision, not a scientific one. Lockdowns can be avoided if governments choose not to implement them. But most seem to be relishing in the emergency powers given to them.

⬐ collyw

>completely Covid free.

not any more https://time.com/5939502/new-zealand-covid-19-lockdown/

⬐ Pyramus

It's 3 cases and they are already implementing a city-wide lockdown.

⬐ falcrist

"Flattening the curve" definitely worked, and you can see that in the data. However, what you're not accounting for is what happened AFTER the lockdowns were effective.

Experts worked on detailed plans for staged re-opening after the lockdown. I don't know about Europe, but the US did not follow such a plan. Different states reacted differently, and politics had a bigger say in how reopening was done than any epidemiologist.

⬐ collyw

I think you are very mistaken if you think that. Lockodwn appeared to work, but given that it takes at least 20 days from catching covid to dying, you can see in many countries that the curves for deaths were flattening off well before lockdown could have been the cause. Uruguay, Japan, Sweden, Belarus didn't lockdown, and don't stand out in any way for their numbers.

⬐ ntsplnkv2

Japan has a far different culture than the US, people wear masks in normal times there.

And Sweden wasn't exactly a success, and has done far worse than their nordic neighbors, say, Norway. The numbers do not lie here.

Belarus had a very effective plan - they tested far earlier and more often, and have far more hospital beds per person.

Straight up country comparison is just a fool's game. We will have to learn from this as best we can - what worked, what didn't work, where, and why. If everyone wore masks properly, was lockdown necessary? We need time to study and find answers.

⬐ collyw

Sweden was very average for Europe without any tyranny from government. I would say that is success. Here in Spain we had many more of our rights taken away and still had more deaths with corona.

> Sweden wasn't exactly a success, and has done far worse than their nordic neighbors, say, Norway

> Straight up country comparison is just a fool's game

⬐ notsureaboutpg

>Let science do its thing - which is inform and educate. Let policies be made by those with more full understanding of the system into which changes need to be effected.

But what about when scientists collaborate with the policy-makers to shape public perception about how to inform and educate people?

One prime example is last years long and detailed study on how there was no "gay gene". In effect, homosexual activity and inclinations could not be shown to be the result of any specific gene expression after many trials and searches for such.

The scientists, worried about how this result would look to the public, worked with LGBTQ advocacy groups to shape how the paper would be released to the media and explained to the public[1].

Now, a sizable portion in HN's demographic will feel this is all well and good, but how far should this go? The uncomfortable truth here is that people are not "born gay" as we have been told so often. And that has been a very important talking point in the past two decades when debating about gay rights. Many people were convinced to support gay rights for this reason alone (however we might feel about that, it's true). They very well may feel duped by scientists and no longer "trust the science". How far should scientists go to shield the public from uncomfortable truths? And how much will the public continue to trust them when such veiling of the truth is made known to them?

[1] https://www.nature.com/articles/d41586-019-02585-6

⬐ epgui

While you appear to suggest that reference [1] is a bad thing, or even a slippery slope, there's not really any indication that these consultations had an adverse impact on the quality of the results.

Communication is an incredibly tricky thing. Even when the science is sound, shortsighted approaches at communication can have serious repercussions, even potentially causing deaths (think public health guidance being misinterpreted, misunderstood, or decontextualized, for example).

Being conscious of this doesn't dilute the truth of the message, and being respectful/mindful of others doesn't mean we need to compromise on the science either.

For clarity: this is not a response to any part of your comment other than the reference to the nature article.

⬐ notsureaboutpg

My point was never about the quality of the results.

In this discussion about science as a religion, I made it clear that working with activists to communicate science means people need to believe the non-scientific claims of those activists as much as they believe the results of the experiment described in a paper.

And in many situations even HN readers will believe it's necessary to massage truths to the public in order to avoid science being read one way or another. This is what turns science into a religion. One must believe the publicly sanctioned, regionally accepted version of science communication if they are to be said to "trust the science".

Had this same study been done in, say, Iran, the science communication would be vastly different, the results the same, and one's reaction to the vastly different communication would still be gauged as a commitment to "science" and how much one believes in it.

⬐ sriku

> But what about when scientists collaborate with the policy-makers to shape public perception about how to inform and educate people?

That isn't a problem with science. It is politics .. as @AnthonyMouse noted.

⬐ Ambolia

It's generally cheaper to buy a scientist than to buy a politician.

⬐ myrion

The problem I have with your comment is likely the reason the scientists were worried about how best to communicate their results:

"sexual orientation is not the expression of any one gene" is not the same as "people are not born gay".

It may be a complex genetic interaction or it may be some other effect at play, but whatever it is, it doesn't contradict that people don't choose their sexual orientation.

Since people have justified all sorts of atrocities based on the misconception that being queer is a choice for people that they could just not make, I can understand scientists who want to make sure their research doesn't fuel that again.

⬐ notsureaboutpg

I don't wish to argue about the actual results of the paper since that was not the point. The point was that if you work with activist groups to inform the public of your findings you are going to make it hard for people to "trust the science" since they need to also trust the activists you've teamed up with.

But the paper, on the surface, does contradict the idea that people's sexual orientation is innate because it rules out many of the most likely complex gene expressions that could have explained this behavior.

I understand the sociopolitical dilemma involved with this unfortunate truth but that's exactly why "trust the science" is complicated and it's exactly why science ends up needing to be like a religion, either you believe it or you don't. Scientists cannot afford (and have chosen not to afford) for people to take the results and interpret the sociopolitical consequences of those results themselves.

⬐ AnthonyMouse

> There are reasonable accepted empirical approaches in accepted use today. Any team that follows these methods and reports honestly can be trusted more than a team that doesn't.

The underlying problem isn't actually a problem with science, it's a problem with politics.

You need to hire some scientists to do your experiments. If they follow the scientific method, you get results using the best system we know how.

But if they follow the scientific method then the Republicans work to defund climate science research and the Democrats work to cancel any scientist whose results challenge the orthodoxy on race or gender.

And if politics interferes then it's not science anymore. If research leading to politically inconvenient results has its funding withdrawn or scientists are intimidated into self-censorship then it's all just selection bias and fudging the numbers.

Maybe what we need is to do something like the court system, where you have multiple research funding "judges" which have lifetime appointments and each get a given amount of research money to dole out every year, and their job is to decide how to spend it. Then they can spend it however they like and can only be removed for e.g. corruption, not for funding research somebody doesn't like.

⬐ dataduck

This is the fundamental problem with "evidence led policy". As soon as science can reliably affect policy outcomes, people who want to change policy outcomes will corrupt the science.

⬐ ntsplnkv2

It won't matter.

Eventually the governments who support technology will win - either via war or economic advantage.

⬐ AnthonyMouse

How is that not mattering? So if China uses science and the US descends further into hyper-partisan culture war, that doesn't matter?

It matters kind of a lot whether the winner is a liberal democracy with free speech and civil liberties or a totalitarian communist dictatorship.

⬐ ntsplnkv2

My point was that eventually the anti-science policymakers will eventually lose out. Societies that value technology and science will win in the end - they will have economic advantages and advantages in war.

> It matters kind of a lot whether the winner is a liberal democracy with free speech and civil liberties or a totalitarian communist dictatorship.

This is irrelevant to the conversation, but yes that is why we spend trillions on defense technology, so that if it comes to blows we don't lose out.

⬐ ufmace

Yup this. In much the same way as things like judging the productivity of software developers by the number of lines of code they write. It sounds like a good idea to the inexperienced, and it may even seem sort of true if you look at it in the right way. But we all know, as soon as you try to make consequential decisions based on it, it will immediately be gamed and not true anymore.

⬐ sriku

You raise an interesting hypothesis - that all interactions between science and policy making is subject to Goodhart's law (https://en.wikipedia.org/wiki/Goodhart%27s_law).

In as much as science deals with things that cannot be affected by policy (ex: stellar physics or nuclear fusion), the "law" doesn't apply I think, except maybe in perception (people may believe that a particular nuclear fusion technique doesn't work even though it's been demonstrated but policy forbade it to be implemented).

For some cases, we would want Goodhart's law to apply - ex: climate change. We'd want actions to be taken to reduce the impact on the planet as we gather and report climate parameters accurately. Similarly, we'd want accurate reporting of gender discrimination in society to result in the reduction of such discrimination. The question really is the way the change is effected. If "salary difference" is an indicator of gender discrimination for example, law makers can pass laws that fix it, thereby removing its effectiveness as an indicator .. but we're happy for even that at some level. That would be similar to a government sending teams to re-freeze melting glaciers because climate scientists are using it as an indicator of global warming.

True policy thinkers, though, would want to ask the "five whys" and work to fix the underlying problem.

You may enjoy this book: https://www.amazon.com/Logic-Failure-Recognizing-Avoiding-Si...

⬐ swolpers

Thanks for pointing at it!

⬐ xtiansimon

I second Dorner’s book. Very readable and equally insightful.

I've been noticing a trend recently when reading about large scale failures of any system: it's never just one thing.

AWS EBS outage, Fukushima, Chernobyl, even the great Chicago Fire (forgive me for comparing AWS to those events).

Sure there's always a "root" cause, but more importantly, it's the related events that keep adding up to make the failure even worse. I can only imagine how many minor failures happen world wide on a daily basis where there's only a root cause and no further chain of events.

Once a system is sufficiently complex, I'm not sure it's possible to make it completely fault-tolerant. I'm starting to believe that there's always some chain of events which would lead to a massive failure. And the more complex a system is, the more "chains of failure" exist. It would also become increasingly difficult to plan around failures.

edit: The Logic of Failure is recommended to anyone wanted to know more about this subject: http://www.amazon.com/Logic-Failure-Recognizing-Avoiding-Sit...

⬐ siculars

This is an interesting point you hit on and something that is stressed in scuba diving. Basically whenever you go out on a dive you only want to change one thing at a time. Only one thing can be "new" or "untested" or "new to you". Otherwise you run the risk of being task overloaded which leads to cascading failure - potentially catastrophic and/or nonrecoverable.

⬐ holri

That's why Sergei Korolev the chief engineer behind the successfull russion space program that led to Gargarins victory said:

„The genius of a construction lies in its simplicity. Everybody can build complicated things."

⬐ rapind

I've been an AWS (S3, EC2, SQS) user for over 3 years now and this article detailing their systems at a mid-level is kind of scaring me off of their platform. It just sounds so complicated and I'm not sure I want to rely on it for anything critical until I can really understand it myself.

Also, a couple other complex systems for your trend are; financial markets and commercial jets.

⬐ olefoo

A related book that examines some of the same themes is Charles Perrow's Normal Accidents http://press.princeton.edu/titles/6596.html

The examples he draws from are nuclear power plant failures (TMI in particular), civil aviation and oil transport. But the basics will be recognizable to anyone who has dealt with large computing installations; interactive complexity, tight coupling and cascading failures.

It is not a reassuring book, you won't be able to look at any complex system without asking yourself what sequence of simple, predictable failures of widely separated parts could tip it into a catastrophic failure mode.

⬐ bd_at_rivenhill

Here's another good resource for understanding these types of problems:

http://www.amazon.com/Normal-Accidents-Living-High-Risk-Tech...

⬐ akuchling

A similar point is made in Gene Weingarten's "Fatal Distraction" (http://www.pulitzer.org/works/2010-Feature-Writing), which was about parents who forget a child in the car. Excerpt: "[British psychologist James Reason] likens the layers to slices of Swiss cheese, piled upon each other, five or six deep. The holes represent small, potentially insignificant weaknesses. Things will totally collapse only rarely, he says, but when they do, it is by coincidence -- when all the holes happen to align so that there is a breach through the entire system."

I believe the scope of than answer is greater than a HN thread, but I might just be wussing out. Hopefully others will engage you. If not, happy to take it offline.

I will say this: be careful of selection bias! Looking back, sure, if I show you a thousand examples that ended poorly your response will be something like "But they weren't really smart. Look how poorly it all turned out!" This is, at best, circular reasoning. The important thing is that, at the time, these folks were the best and brightest and put in charge for that very reason.

Good starting point: http://www.amazon.com/Logic-Failure-Recognizing-Avoiding-Sit...

⬐ tocomment

I'd like to know also. Can you start new topic on HN perhaps?

⬐ DanielBMarkham

Will do. Perhaps I can set this up as a blog entry and then cross-post.

The problem, I think, is to keep this non-political. Somebody brought up the Rumsfield example, which I think is a good one. I also think McNamarra is another good example. The problem is that decisions that are very visible and make a big impact can also politically-charged. I think using a couple of SecDefs is okay, but I'm not sure.

⬐ None

None