Hacker News Reader

Comment by chpmrc | Hacker News Reader

I wasn't sure about publishing this but I took some time to go over Worldometers' data and associate a lockdown type to each country in the EU (+ a few more countries). There doesn't seem to be a strong correlation between how strict the lockdown is and the total cases per million people: https://www.notion.so/Lockdown-data-361fe56821d046a787d76535.... So, yeah, I completely agree with you. We should definitely try to find out what's going on.

I'm writing this because I think HN is one of the very few communities able to properly discuss ideas with hard science and facts. So any observation, correction, criticism etc. is more than welcome!

View on HN

That’s not a useful comparison though. You want to compare cases now to cases when lockdown or similar measures were imposed. Otherwise a country that wastes 2 months doing nothing and then takes action far too late just gets put in the “lockdown” bin.

While I mostly agree with you, I wouldn't call the comparison useless. I want to add a few additional things as well that add to the difficulty of these comparisons (and why these simple comparisons people are making are potentially wildly inaccurate)

1) Testing is all over the place and mostly done only on people that are hospitalized. This creates a huge selection bias in official numbers, especially when we're talking about a disease where we're seeing claims from 25-80% of carriers being asymptomatic. There's plenty of evidence to show that many people that suspect themselves (i.e. they show symptoms) are not getting tested because they aren't "sick enough". So all we can do is guess that official numbers are on the low side. This is part of why we see wildly different CFR numbers (as opposed to IFR) in different countries. (Obviously there are other factors as well, but this one captures effects like living conditions)

2) Deaths are being attributed in different ways. In some areas any pneumonia death is being counted while in others the person must have tested positive for covid. In America the attribution even varies by state. What makes this more complicated is that when the metric changes these tend to be deaths added on a single day (creating spikes). This makes it difficult to do 1 to 1 comparisons.

3) The disease's fatality rate drastically changes with the victim's age. There's strong evidence to show that if you're healthy and young that you have little to worry. But if you're relatively healthy and old you should be concerned. There are drastic differences in country's age, and thus this makes a big difference in comparisons.

4) The number of cases isn't exactly a bad thing. Let's take Sweden for example. If their method for containment (develop herd immunity in the general population, isolate the elderly and immunocompromised, and social distance strictly to not overwhelm hospitals) is effective you'd expect a faster rate of new cases AND deaths. But if their method is the better solution they will have less total deaths by the end. But are they right? No one knows yet because we need time. Time is a difficult factor to work into these comparisons, which are often single snapshots rather than growth rates and inflections.

> I'm writing this because I think HN is one of the very few communities able to properly discuss ideas with hard science and facts.

I propose that if you pay very close attention to the nature of discussions, over a long period of time, across a broad spectrum of differing topics, you may notice that the ability to "discuss ideas with hard science and facts" varies greatly according to the topic.

Or to be more specific...when the topic involves issues that are "political or values oriented", as this one does, the discussion and thinking styles exhibited in the comments much more resembles that which you would find within less "logically capable" communities.

One part of being a community is slowly building values in new members. Many communities, many subreddits as an example, form monocultures and ruthlessly stamp out opinions not aligned to the deigned philosophy (and then being monocultures, they die a slow death).

HN as a community places value on interesting, well informed debates, but that value isn't universally shared outside of HN. It takes regular exposure to build that value in new members but over time they'll either embrace it or wash out of the community as long as we do our best to consistently embrace it positively.

Keep debating and pushing forward on interesting conversations, keep doing it politely and in the interest of meaningful debate.

> One part of being a community is slowly building values in new members. Many communities, many subreddits as an example, form monocultures and ruthlessly stamp out opinions not aligned to the deigned philosophy (and then being monocultures, they die a slow death).

Yes indeed. Reddit is famous for this, and getting worse by the day. But how might we know, for sure, the degree to which that also happens here? And I don't mean that rhetorically. We can all instantaneously generate "an" answer to that question, based on heuristics, but how accurate is that answer? To gain some certainty, it would require looking at actual data, which is firstly dependent on a willingness to look at the data, which is firstly dependent on a willingness to consider the idea that there may be a problem. This last point seems like a rather touchy subject.

"building values in new members" also seems like a well-accepted but heuristically-based idea. Is it, in fact, only new members who lack these values? Luckily, HN profiles contain an "accounted_created" attribute...with a little effort writing some scraping code (something I don't presently know how to do), I suspect some more precise, evidence-based answers would quickly reveal themselves.

What would be even more revealing would be if one had access to voting history. This way, we could get a decent feel for how certain individuals feel about this general topic. But alas, you and I don't have access to that data. But someone does (see: willingness to look at the data). Shall we truly pursue righteousness, or should we just declare ourselves to be, by fiat, as seems to be the most popular approach in broad society?

> HN as a community places value on interesting, well informed debates...

In the abstract, principled[1] sense, sure. But, how well do we "walk the talk"? Once again: what does the data say?

> as long as we do our best to consistently embrace it positively

And consistently. But now we're back to where we started.

> Keep debating and pushing forward on interesting conversations, keep doing it politely and in the interest of meaningful debate.

Despite the obvious unpopularity, and repeatedly inconsistent (again, based on topic) evaluation of "meaningful debate", as well as the HN Guidelines in general, I seem unable to stop lol

[1] "Lord, I promise not to sin!" Inconsistent adherence to an ideology, however strongly held in theory or principle, is not a problem only within the domain of religion, I suspect. Rather, it seems more like something inherent in human beings. Failure rates may vary fairly significantly across domains (or may not!), but this is a difference in degree, not a difference in kind.

You're right that at least large parts of this can be tested and answered with data. Obviously short of a magical classifier for questions like "is this a productive debate" and "is this ideological fighting or a search for truth?" then we're going to need someone to hand classify a bunch of things which is expensive, slow and only as accurate as the person making the judgement call. In other words, we'd just be making ad hoc judgements again, but calling them something else because we made those ad hoc judgements in a csv instead of a comment.

The other issue is that communities are dynamic, and behaviors trend. You're probably not going to see a flat correlation, you're going to see something that resembles a wave function.

> then we're going to need someone to hand classify a bunch of things which is expensive, slow

True. But not impossible.

And I would recommend comparing effort to potential reward. Imagine, just for fun, the possibility that HN could find a way to discuss climate change[1] in a disciplined, truthful, and non-heuristic manner. What might the value of that discovery be? And if we could find a way do it, perhaps others could as well?

> and only as accurate as the person making the judgement call

How deeply have you thought about that? Like, say it was an engineering problem you've been asked to solve - have you put as much effort into thinking about it as you would if it was a paid assignment?

https://en.wikipedia.org/wiki/Clarke%27s_three_laws

> In other words, we'd just be making ad hoc judgements again, but calling them something else because we made those ad hoc judgements in a csv instead of a comment.

Do you realize you are:

a) predicting the future

b) making several assumptions

c) utilizing loaded terms (which is harmful not only to others, but sometimes also to oneself)

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

> The other issue is that communities are dynamic, and behaviors trend.

This is true.

> You're probably not going to see a flat correlation, you're going to see something that resembles a wave function.

Your prediction may be correct. But it also may be incorrect.

[1] A topic we seem to loooooove to talk about as if it only involves climate science, and has no psychological aspect whatsoever.

You should double check your definition for loaded language as I agree with the article you linked but not your usage of it. To quote from your source:

> Loaded words and phrases have significant emotional implications and involve strongly positive or negative reactions beyond their literal meaning.

I hope 'ad hoc', 'judgements', 'csv' and 'comment' don't carry any strongly positive or negative reactions, nor stereotypes or implications. (Though I have known a few people to have negative reactions to the idea of csv files...)

> I hope 'ad hoc', 'judgements', 'csv' and 'comment' don't carry any strongly positive or negative reactions, nor stereotypes or implications.

I would very much like for us to have a written language that has the same bandwidth as a long form, in-person discussion. But we have to play the hand we've been dealt, not the hand we wish we'd been dealt.

And to be clear, I don't object to your complaint, or consider it unreasonable - quite the contrary: this is the type of conversation I think we need more of. My comment above could have been improved by adding some disclaimers to make it explicit that I am not intending to speak in an "accusatory" manner, but I reckon the vast majority of people would have read it that way, and for good reason. So, apologies for that.

The ideas I had in mind are somewhat (again, limited by the fundamental shortcomings in our written language) captured in here:

> Loaded language (also known as loaded terms, emotive language, high-inference language and language-persuasive techniques) is rhetoric used to influence an audience by using words and phrases with strong connotations associated with them in order to invoke an emotional response and/or exploit stereotypes.[1][2][3] Loaded words and phrases have significant emotional implications and involve strongly positive or negative reactions beyond their literal meaning.

(The intent, emotional, and exploit parts I do not consider at all relevant to this discussion).

The larger, more abstract point that I'd like to make is: reality is infinitely complex, and we as individuals reason upon it with far from infinitely powerful minds. Now, start layering on all the other complexities, like collective reasoning (hampered by language and a variety of other things), necessities of being forced to make incredibly imprecise/coarse group decisions on a strict timeline (our political systems), complex and imperfect incentive systems, asymmetry and inaccuracy of information, advances in some domains (science, technology, industrialization) without corresponding advances in complementary domains (psychology, consciousness, etc), and so forth...and before too long you've got a bit of a mess on your hands.

I couldn’t agree more.

The COVID discussions have been the single most frustrating experience for me on HN since I joined. Ever single comment I’ve made downvoted, even accused of contributing to millions of deaths (an account which was thankfully banned for making the remark).

It’s only in the last week or two that it’s started to become remotely possible to have intellectual discussions on COVID, but it’s still overwhelmingly politicized and unscientific.

agreed, it's been hard having discerning discussions about this disease. the tide has been turning though in the last few weeks as the hysteria wanes.

The problem is that you have every Jane and Joe who's ever put a line through a set of points making graphs and arguments that conflict with the consensus of epidemiologists/social statisticians. It's fine (and good!) to investigate that sort of thing and share your results, but then you should be sharing "I did this sort of analysis, it seems to conflict with all these other analyses, does anyone know why? Maybe X?". Instead you have "OMG when you do this it looks like Y, the powers that be are lying to us!".

For a non-critical issue this would lower the quality of discourse slightly, but it's not an exceptional problem. In the case of critical major issues involving active disinformation this is highly irresponsible behavior.

> For a non-critical issue this would lower the quality of discourse slightly, but it's not an exceptional problem. In the case of critical major issues involving active disinformation this is highly irresponsible behavior.

Assuming your description of "the" problem is accurate (I suspect it isn't - and it's "a" problem, not "the" problem, pardon the nitpicking), what do you propose should be done about it? On an individual basis, a community basis, a governmental basis, etc - what should be do? Because as far as I can tell, we (the citizens of planet Earth) seem to be stuck in this recursive loop of anger, finger pointing, and general failure. And it doesn't seem like anyone else notices.

Well, mostly no one.

https://i.imgur.com/QA6vSzC.png

Not the OP, but one thing that is vital is that those in power are inscrutably honest and value scientific method, IMO.

If a politician stands between the scientific advisors and the people then when they make pronouncements saying "this is for medical reasons" when in fact the reasons are primarily financial, or tempered by the interests of other parties (their personal backers), that needs to be made clear.

When the UK said "we're using different models" they should have been publishing those models, among with the facts they were relying on, published a list of all people who had input to that decision and the context.

A rolling diary of all people any senior politician had contact with; a list of every shareholding by any member of a politician's family; and such information need to be openly accessible in real time.

Basically we need to treat them as an obsessive helicopter-parent would treat their teenage children, they need to have tracking devices, they need to have "friends" (employees of opposition parties) that attend every meeting they go to, they've down they simply can't be trusted.

But of course you can't do any of those things because corrupt politicians won't vote for it. We're stuck at a local minima of democraacy, IMO.

How to progress? I don't see a way, politicians have enough of the population in their thrall (eg using media manipulation) to sustain the current system. Two things that may help are to have bodies parallel to government adopted ones, like UK's SAGE but excluding the politicians, and have media that will report direct from those bodies to the people rather than via a filter of politicisation. Maybe then enough people can get onboard to bit-by-bit elect trustworthy officials interested in government for the good of the people.

> When the UK said "we're using different models" they should have been publishing those models, among with the facts they were relying on, published a list of all people who had input to that decision and the context.

Everything should be published (it isn't), and nothing should be censored, without good reason (not the case today). That so few people seem to realize this is mind boggling to me. But then, it's human nature, so it shouldn't be surprising.

> How to progress? I don't see a way, politicians have enough of the population in their thrall (eg using media manipulation) to sustain the current system. Two things that may help are to have bodies parallel to government adopted ones, like UK's SAGE but excluding the politicians, and have media that will report direct from those bodies to the people rather than via a filter of politicisation. Maybe then enough people can get onboard to bit-by-bit elect trustworthy officials interested in government for the good of the people.

This idea, and others like it, seem like a very plausible way out of this mess. I really see no other way.

But how much effort, in man-hours, is being expended in this direction, as compared to, oh I don't know, people having the same arguments year after year, in thread after thread?

I'd say that behavior is similar to if an individual in, say, a school or assembly, shouted "FIRE" upon witnessing a chemistry demonstration involving a controlled flame. If that somehow caused a stampede resulting in loss of life they might be held accountable by the courts. These acts on social media can be quite similar philosophically, so one would expect that the legal treatment should be similar (albeit perhaps adjusted proportionately for impact and reasonableness of the held belief).

I don't necessarily think that's what should be done in practice, but it seems consistent with how western society responds to this behavior in other settings.

I don't necessarily think there is a good solution, either - I just mean to say that people who do that, even if they are well-meaning, are behaving unethically. That's because they believe in their own perspective with 100% probability, but that's obviously unreasonable from a Bayesian perspective. If they accounted for even a small probability that they are wrong then their behavior would change dramatically.

> I don't necessarily think there is a good solution, either - I just mean to say that people who do that, even if they are well-meaning, are behaving unethically. That's because they believe in their own perspective with 100% probability, but that's obviously unreasonable from a Bayesian perspective. If they accounted for even a small probability that they are wrong then their behavior would change dramatically.

Completely agree, but I believe this behaviour applies to 90%+ of those who express confident opinions on the matters of the day, and downvote anyone who dares question those opinions when they can't put together a convincing argument for their strong assertions.

Some evidence for the frequency of this behaviour exists, we could have insight into it, but I suspect we shall choose to not discuss such things. Only downvote them.

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

https://behavioralscientist.org/epistemic-humility-coronavir...

https://plato.stanford.edu/entries/wisdom/

(These links are not directed at you personally, just to be clear).

Looks like either every single person I reply to disapproves of my words, or we have yet another busy little beaver, eh dang?

Let's see how this one fares.

All sorts of confounding variables. Hard to test just one variable at a time.

I think cases/1M population is a good variable though. Deaths depend on the quality of the healthcare system, number of infected people (per 1M) only by how quickly the virus spreads + number of tested people (assuming the quality of the tests is the same across all EU countries). Would you please elaborate on that?

Cases are a difficult metric because there's a selection bias. Most places, including the US, does not have adequate testing. Especially for a virus we suspect a significant amount of people to show no symptoms. We're mostly testing those that are hospitalized.

Deaths, on the other hand, are easier (but not perfectly) to count because it is much more clear that someone has died. The symptoms are trivially obvious. The difficulty here is that there is no universal attribution (even within the US) as to what constitutes a covid death. Let's show a tricky example. Let's say an elderly patient had a stroke and are in the hospital. You THEN get infected and die. Did you die from the stroke or covid? Is this as deadly as a disease that kills a healthy 60 year old? How do we distinguish this?

I'm just trying to say that without context these simple comparisons can miss a lot of nuance of the reality of the situation. The thing here is that there are a significant number of variables at play and without discussing those variables you can get inaccurate results.

I agree, that's why I published the table and anyone can comment (in Notion itself), therefore contribute. I think such a comparison can at the very least spark the discussion you mention.

Isn't this what Issac Ben-Israel claims in his article[1]? Basically there seems to be a common pattern in all countries, independent from the type of lockdown applied. After 40 days the peak is reached, after 70 days there are almost no new infections.

[1] https://www.timesofisrael.com/the-end-of-exponential-growth-...

Reply on news.ycombinator.com