Hacker News Reader

Comment by TeMPOraL | Hacker News Reader

Yes, seriously. Number one failure in sensationalized reporting is base rate fallacy: they point out a high absolute number, implying something strange is going on, whereas usually the denominator - which they neglect to mention - has rose proportionally, meaning there's nothing newsworthy there.

I don't mean this is the case here (I'm yet to read the article itself) - but this failure mode in reporting is so common that it's a good habit to immediately look for / question lack of base rate data.

View on HN

I get that at some point murder becomes statistics, but I still feel the need to re-emphasize that this is murder. Not more folks getting the flu because there are more folks.

The larger point is that some (very small) percentage of people will be murdered every year. If the number of climate activists increases, the chance of one of them being murdered increases with it - possibly just by being involved in a random shooting or family drama, completely unrelated to their activism.

I agree that murder is horrible, but the article seems to try to sell that these murders are related to their activism, which is not something you can say just based on the absolute number. You need to have the ratio to compare to the last year and to the base rate in the general population to make a better statement. Without it, we can't be sure of the real cause and might try to fix this at the wrong place, leading to us to not prevent murders effectively - so yes, you need to get your statistics right, even when talking about horrible incidents.

i disagree with the implication that murder somehow shouldn't be looked at in statistical terms. I am not saying it should only be looked at in statistical terms, but it is one important way of looking at it.

Fair enough, and I agree. People are being murdered, and that's the more important topic.

>whereas usually the denominator - which they neglect to mention - has rose proportionally, meaning there's nothing newsworthy there.

there is not necessarily a reason to assume that the percentage of murder of activists is going to stay the same if you increase the number of activists, although it may be the case.

The point of having the denominator is so that you don't need to assume anything, you can just know which way the rate is going.

Perhaps you're right that in this case, the rate isn't going to stay constant. But in general, with news reporting, I found it safe to assume that if the rate was rising, they would've reported on the denominator too - the reason to omit it is usually because it would kill the story.

EDIT:

To expand on this further - and again, I'm defending the original question as a perfectly legit one, not trying to diminish the fact that we're talking about murders here - you always want to see the whole fraction spelled out, numerator and denominator, because the other common trick to deceive with numbers is by providing only the ratio itself. E.g. "over past year, incidents of $crime rose by 80%", or "$substance increases your cancer risk by 300%", where the unmentioned denominator is "3" and "0.00001", respectively.

The way I see it: even if we're talking about clear tragedies or acts of evil, it's always better to have an accurate picture of what is being talked about.

> there is not necessarily a reason to assume that the percentage of murder of activists is going to stay the same if you increase the number of activists

If activists are being murdered for activism, the rate among activists will go up. If a greater share of the population is becoming activist, and are being randomly murdered for non-activist reasons, that’s a different story.

if the number of police rise does the ratio of police being killed on the job stay the same?

what if 5% of the workers at big corporations are willing to kill activists for activism, those 5% are busy killing activists, if the activists increase unless the corporations increase the number of people they have willing to kill activists it follows that there are not going to be a significant increase in the number of murdered activists.

In short there can be all sorts of reasons why the percentage of activists killed for activism would not remain constant.

All you wrote is exactly why you want to know the base rate (or several, against different possible hypothesis). Otherwise, it's not possible to even speculate about these things productively.

as you'll notice in my first reply to you I never said I did not want to know the base rate - what I said was - first quoting you

>whereas usually the denominator - which they neglect to mention - has rose proportionally, meaning there's nothing newsworthy there.

then in my reply to you

>there is not necessarily a reason to assume that the percentage of murder of activists is going to stay the same if you increase the number of activists, although it may be the case.

again, nothing about not needing to know the base rate, just noting that there might reasons for the rate of murders to not rise proportionally.

After writing something so straightforward that it might as well be observing that water is wet I've gotten a number of responses from people whom, just as straightforward, seem to want to inform me that the sun is out today and are intent on hammering it into my thick skull despite my never having said otherwise.

This assumes that there is an infinite capability and willingness to kill activists.

If you have 100 ppl demonstrating against a hydro dam and 1 gets murdered later that night it does not follow that there likely would have been 2 killed if 200 ppl demonstrated.

These are murders of people. The pertinent comparison (for those interested in the whole "murders" thing) is with how many murders there were in previous years, whether it's gone up, etc, information which is covered in the article. To say there's nothing newsworthy about these murders, just because you were personally interested in a number you couldn't find in the article, is crass.

No. These are murders of "people who do XYZ". If we looked at murders of people who play Fortnite from 5 years ago until now, we'd see a clear trend, for no reason other than the fact that the number of people who play Fortnite has ebbed and flowed.

I don't know anything about this particular situation, but your callous dismissal of this indisputably correct warning is wrong.

You are right, I missed that point.

I didn't say there's nothing newsworthy about these murders. I said, in my paragraph, that sensationalized articles in general (and note that I didn't say this one definitely is in this category) often don't report the base rate, because it would reveal they're reporting a non-story. And I said that to counter cenophor's objection to ArchieMaclean's question, by pointing out it's pretty much the first question one should be asking when discussing an article/report that doesn't provide that number.

> The pertinent comparison (for those interested in the whole "murders" thing) is with how many murders there were in previous years, whether it's gone up, etc, information which is covered in the article.

It wasn't, at least not directly. The numbers I saw are:

- 227 activists killed in 2020, the titular record number

- 4 activits / week on average being killed since the signing of the Paris Agreement in 2015[0]

You can tease out some insight from deconstructing the average (e.g. the 2020 number is above the average), but not much really. Not enough to tell you whether something has changed about this situation.

Also, please don't assume people asking for more clarity and better data are ignoring the human life context of the story.

There seems to be an error in the reported context of that figure - Paris Agreement was adopted in 2015, but signed in April 2016 and became effective from November 2016 - per - https://en.wikipedia.org/wiki/Paris_Agreement.

I was wrong, you were making a very important point that just went over my head (it was early in the morning for me). Sorry.

Other than the increase in Murder of activists

Reply on news.ycombinator.com