Hacker News Reader

Comment by freddie_mercury | Hacker News Reader

Ah yes, the good old "my anecdote trumps science" post on Hacker News. If this happens on Hacker News, no wonder the rest of the world is even worse.

"Two systematic reviews and meta-analyses of the latest evidence were conducted for this study. Data from 37 studies on ischemic heart disease covering more than 768 000 participants and 22 studies on stroke covering more than 839 000 participants were synthesized. The study covered global, regional and national levels, and was based on data from more than 2300 surveys collected in 154 countries from 1970-2018."

The study has nothing to do with sitting in an office, despite what you claim.

View on HN

I think you’re confusing the map for the territory. The study pooled all workers together, but that doesn’t invalidate the hypothesis that certain types of workers were more affected. Unless the study explicitly broke out the purported subdivision, we can’t say for sure.

For example imagine 50% of workers are office workers. And say heart attack risk is twice as high for overworked office workers but 10% less for non office workers. The study would find a broad-based 40% increase in risk for all workers, despite OPs original hypothesis being exactly correct.

The parent comment shoots from the hip, though. He has one anecdotal case of a more active career living to 92, then he postulates his inactive career will lead him to an extremely early death (60 is young in modern economies). That's just not the depth of discussion HN is designed for, does this ignite curiosity or push the discussion further?

Yes, it does. This is exactly how you pick holes at research. What factors didn't it look at? That helps us understand the limits of what we can deduce from the data. Can we think of an intuitive story about why those factors might have a causal relationship to the thing being measured? If so, we have especially good reason to be skeptical of any conclusions from the study that don't address that possible explanation.

Somewhere in the last couple of decades, the SCIENCE WORKS, BITCHES people seem to have forgotten how real science is actually done. Intuition, anecdote, common sense, hunches are very important parts of the process.

Science works that way in the minds of scientists, whose vast experience gives them dozens of actual data studies they've read to speculate from. Researchers have read about thousands of cases, they're the ones with the refined intuition, not forum readers.

Specifically: Is it productive for HN users to have anecdotal evidence rise to the top, or actual peer-reviewed evidence from studies with N=100 or N=10,000? I'd argue if layman HN users have time to read only one comment, we should upvote the N>1 comments, not N=1.

From the guidelines: "Comments should get more thoughtful and substantive, not less, as a topic gets more divisive."

We went from an international article (N=745k) to anecdata, which is a downgrade of substance in my view.

I'm not sure what's the point of coming up with farfetched theoreticals disconnected from the actual study when it is transparently obvious the OP was just reacting to a headline with a preconceived notion.

They found the biggest effects in Southeast Asia. I live in Southeast Asia. If you think office workers make up a significant part of workers in Southeast Asia....That's just not a "let's assume" I'm going to join you with.

If you think office workers are the cause then do the hard work of actually looking at the data and use that to make your case.

Still the point above is valid. A good 'job' is a source of well being. A bad job will kill you faster. The former might be very rare though.

I wonder if that's true. A good job is still a job. If you enjoy it and that leads to you spending more time working because it's "fun", it's not hard to imagine that good job could actually lead you to die younger than a job you dislike and only work the minimum amount possible at. That's not to mention the fact that you might also sacrifice more time commuting, lower salary, and less benefits for the sake of a "good" job doing what you enjoy.

There's plenty of reasons to imagine a job you enjoy could lead to a shorter life.

Maybe it doesn't though. And even if it does, is prolonging your life for as long as possible actually a noble goal in the first place?

Then it's unrelated to the nature of the task. I thought it was meant to attest of working + time (which I don't believe in[0]). You will probably die faster if you're a passionate climber than if you're a library clerk even if you despise books.

Commute time is also part of the blend btw, it's a balance.

[0] for instance I have money and am jobless but my actual context is unhealthy as hell (social, emotional needs unmet to the fullest)

Job or not, humans spend all of their waken hours doing something.

It may be that the name we give to those things is enough to make them harmful, but I do find that very unlikely. If it's not calling it a "job" that makes it stressful, then it must be some properties, and it's pretty much possible that some jobs lack those properties.

I also wonder a lot how daily life (aka what 'modern' society call work) felt like in a group. It seems that in smaller places (villages, tribes) you're linked to everybody, your efforts are giving visible benefits to people you know, it's to make the group thrive. Maybe more motivating than an abstract duty for some big corp in a large field.

Statistics by definition lose information, and part of understanding if you’ve lost the “right” information is to look at the individual cases that are excluded by the model. So yes, in those cases the anecdotes do trump the “science”, if by “science” you mean a lossy, probably-biased model of the real world.

What you can’t do with anecdotes is generalize them to the entire population, but taking an honest look at them is an excellent way to find methodological gaps.

None of this stuff is definitive. A lot of self-directed craft work like trades and farming doesn’t have an easy “work” vs “not-work” mode.

They also are roles with a lot of agency and variation in schedule. A carpenter may be working 60 hour weeks in June, but is usually partially idle in January.

Did you read the study? I did. It aggregates scores of other studies with orthogonal datasets, widely disparate abstracts, and what appear to be dramatically different outcomes, then applies what appear to be arbitrary and AFAIK undefined models to achieve its probably unfalsifiable outcomes.

Not saying the conclusions they came to are wrong. I’m saying you can’t possibly determine whether it’s right from the data they present. I’m not sure they could replicate their own conclusion if you sat down at a desk with them.

> If this happens on Hacker News, no wonder the rest of the world is even worse.

Is there any reason to assume HN should be better then anyone else?

Generally, HN people are more educated and better trained in logical thinking (most being engineers of some sort or in other professions that require a bit more thinking). Hence, the GP's expectation, I assume.

However, this is just my impression from reading comments here for a few years. I may as well be wrong. Smart people do have their own blindspots, too.

Edit: I noticed you were downvoted, and I disagree with that. Your question is reasonable.

Whether logical thinking gets you to correct result strongly depends on assumptions you make and model you apply the logical thinking on. And my impression is that HN is not really good at those. It also starts to fail when the assumptions are only sorta kinda true, with many asterixis, because logical implications then treat them as true.

Very very often, people start with very wild assumptions and then make wild implications on them, even the ones that are directly contradicted by article they comment on or simple search.

That's how you get conversations, and eventually progress. To me, that's the attraction of public forums like HN: people (including myself) make assumptions, and conversation often (or sometimes?) leads to corrections. If we'd all be working with perfectly correct assumptions, we would not need forums anymore. We'd just be fed information, process it in exactly the same way and reach the exact same conclusions. In my experience, this is not really possible with humans.

What it primary does instead is to disseminate feel good assumptions. Because the people don't state them as assumptions, they claim them to be facts. The conversation does not lead to corrections often enough in programming and business unrelated topics.

That is fairly normal, group of people talking about something they dont know much about is not supposed to randomly pivot to correct conclusions. But, HN is uniquely convinced HN is superior in this regard.

I take it you are also uniquely convinced that you are superior in this regard, then. :)

But here's an honest question: why are you on HN, then? Sure, it's a decent source of tech&related news. But there's no reason for you to spend time reading and writing comments, if you truly believe it's as bad as you say.

It is light tech adjacent news with a taste of drama and arguing, but not too much drama. Because I wait for tests to run and want to idle kill time. Cause I am procastinating.

I use it same way as reddit, basically. Except this don't sends notifications so it is less pushy.

> Generally, HN people are more educated and better trained in logical thinking

The main effect this has is that HN people have an overinflated sense of their own competence, and are more blind to their own ignorance.

I do not disagree, and I hoped it was already visible from my initial comment. But, on average, you probably still get better results from HN crowd than from other crowds, in matters that require knowledge and logical thinking. Do you disagree?

Absolutely disagree. I think HN is massively blind to its own biases, and makes the wrong calls very often as a result.

Ok, I can accept that as a possibility.

But here's a curiosity: why are you here, then? What do you derive from reading all the comments from people that are by default wrong? Isn't it a massive waste of your time to read large number of comments, just to occasionally find one that you agree with as being correct?

I am here to get thoughts on tech, tech news, and tech related things. I am outside of that world, and want to get an insider's perspective on my own views.

When it comes to education (my field), HN is generally very, very similar to any other large online group such as facebook or reddit - the biases and areas of ignorance are the same.

So, in other words, I come here to read and learn about the things this site should know about - technology and related fields. Whereas comment sections such as this one, I consume much the same as 'opinion articles'.

Not who you were originally replying to, but I find myself in a similar boat. I don't necessary add any more weight to a comment simply because its on HN.

But I do find HN useful in terms of randomly coming across topics that may be of tangential use in my work. Also, it tends to have more discussions that I'm interested in if I just want to blow off some steam on a forum and tends to have less vitriol than, say, a sub-Reddit

I am here for similar reasons, and I don't spend much time anywhere else, again for reasons similar to yours. On average, I found HN to be good, and that I believe is due to the people who contribute here. And I am rather surprised to see that some people have such a bad impression of the overall quality of the contributors.

Maybe it's due to my approach to HN comments: if I find something interesting, I read, upvote and dive further into the thread; sometimes I comment. Otherwise, I move on.

The problem is that “other crowds” is an amorphous and meaningless concept. In one context “other crowds” means the US/EU population as a whole, in which case sure, HN readers are on average probably more educated and better-informed. But in another context “other crowds” might mean “experts in the humanities and social sciences,” compared to which the average HN reader is hopelessly ignorant unless the question is specifically about programming. And indeed it is the humanities and social sciences where HN’s blindness and ignorance is most obvious, and its arrogance most undeserved.

For esoteric topics, HN can’t be beat. For specific topics, like medical issues, Reddit or Facebook are often amazing.

Have condition X. Doctors say Y, which doesn’t actually help.

Support groups will have a long list of things to try. Questions to ask doctors lifestyle changes. Up to date info on diagnosis/ treatment.

What tests should be run, how to interpret tests.

Medications side effects doctors don’t know about.

I always assumed a Facebook medical support group would be all nonsense. Instead it’s got tons of people who have spent years researching and experimenting, and have amazing advise to give.

Simple example. Was going to the bathroom 10+ times a night. Years of doctors were zero help.

Someone suggested aloe Vera freeze dried pills. 3 days later I was sleeping through the night.

Medical community is both amazing and full of major gaps.

I found http://n-gate.com/ helpful.

Well yes. They most certainly have a higher rate of tertiary education (and thus exposure to scientific method) than the general population.

Higher education does not make you less prone to cognitive bias. And education in one field does not make you accurate in another field.

I don’t believe you. Can you back that up?

Daniel Khaneman has done extensive work in cognitive biases and still admits it doesn't improve his ability to avoid them in himself. If someone well-educated in that specific field can't improve much, I doubt there can be claims that education in general is much help.

Maybe it's you who doesn't understand the epistemology of science. See https://news.ycombinator.com/item?id=27185221

Reply on news.ycombinator.com