Hacker News new | past | comments | ask | show | jobs | submit login

> The rate of both mass killings and individual killings is WAY down from what it's been historically.

You're making a common statistical error, one that climate change deniers use daily -- if you perform a moving average using a long averaging interval, the climate is warming up. If you perform a moving average using a short averaging interval, the climate is cooling off.

The second measurement is an error because it can only predict the future if the future is limited to a year or two in duration. But the long averaging interval is a more honest and reliable predictor of the future. That predictor shows a warming of the climate, and it also shows that eventually, population growth will collide with mass death.




In the case of murder, the murder rate today is no higher than it was a century ago and much less than it was 30 years ago. So how big an interval do YOU think is appropriate to look at for the murder rate? Go back too far and you get into measurement problems. Merely counting incidents (as the wikipedia page does) is nonsense for several reasons (1) a vastly better news network is going to make more incidents easy to find today than in the past, (2) incidents that happened since the internet and in recent people's memories are going to be overrepresented, (3) population growth means we expect more incidents even if the base rate were flat or declining.

In the case of climate change, the trend for the last decade is flat/cooling - that's more than "a year or two". Which is not terribly consistent with the models - the models clearly have been running hotter than observations and are have been unreliable predictors.

http://www.woodfortrees.org/plot/wti/last:120/plot/wti/last:...


> In the case of murder, the murder rate today is no higher than it was a century ago and much less than it was 30 years ago.

To make that claim, you need to define "rate". Does "rate" mean the absolute number of murders per day or year, or the number of murders per day or year divided by the population?

The absolute rate is obviously higher -- the burden must be is yours to offer evidence that it's unchanged since 1800. But the rate per population might be unchanged (not likely), but if so, that supports the idea that increased population is a factor, which circles back to my original point.

Also, when you perform a moving average, you can "prove" anything by choosing the right interval. But, because the future is expected to be long, so should our averaging interval. To do otherwise is intellectually dishonest.

For example, climate change deniers use short averaging intervals to "prove" that the climate is actually cooling, where the same measure using a longer averaging interval shows this is false.

Oh, I see you do just that. Too bad -- choosing an inappropriate averaging interval is not science, it's politics.

> ... population growth means we expect more incidents even if the base rate were flat or declining.

Well ... now that you're arguing my points for me, I guess we're done.


> To make that claim, you need to define "rate".

"rate" means per capita per year. By "a century ago" I meant around 1913; 1800 would be TWO centuries ago. The murder rate in the US has been in the general ballpark of 5-10 per 100,000 annually since 1910. Currently we're near the low point of the last century. Um, here:

http://www.americanthinker.com/2012/12/listening_to_the_late...

Quote: "Today's murder rate is essentially at a low point of the past century. The murder rate in 2011 was lower than it was in 1911. And the trend is downward. Whatever we've been doing over the last 20-30 years seems to be working, more or less. The murder rate has been cut by more than half since 1980: from 10.7 to 4.7."

> But the rate per population might be unchanged (not likely), but if so, that supports the idea that increased population is a factor, which circles back to my original point.

You now appear to be stark raving bonkers. No, silly, if the rate is unchanged it does NOT support your point. Yes, as there are more people, there are likely to be more murders as an absolute number. There are also likely to be more hugs, more kisses, more birthday parties, and more people making silly arguments over the internet. NONE of those constitute a population-limiting crisis.

Well, except maybe that last one. :-)


> "rate" means per capita per year.

That's by no means the only meaning of "rate". The meaning of a rate depends on the two values involved -- the dividend and divisor. The identity of both defines a specific use of the term "rate".

"Murders divided by years" is one example of a rate.

"Murders divided by (years times population)" is another.

And "(mb-ma) divided by ((yb-ya) times (pb-pa))" starts us toward a moving average.

Rate doesn't have only one meaning -- to accept your definition would severely constrain mathematics.

> No, silly, if the rate is unchanged it does NOT support your point.

Of course it does -- given a rapidly increasing population and the same rate, it means the absolute number of incidents has increased. That was my point.

> There are also likely to be more hugs, more kisses, more birthday parties, and more people making silly arguments over the internet. NONE of those constitute a population-limiting crisis.

But in truth, all of them do. It's a classic case of an increasing population held within a finite land mass -- eventually the system collapses.

If the population is increasing without constraint, it doesn't matter what those people are doing apart from increasing their numbers -- the fact that they're increasing their numbers eventually becomes the only issue worth addressing.


>> "rate" means per capita per year.

>That's by no means the only meaning of "rate".

It is the standard definition when talking about a CRIME rate. Googling "define crime rate" returns things like this:

"the ratio of crimes in an area to the population of that area; expressed per 1000 population per year" http://wordnetweb.princeton.edu/perl/webwn?s=crime%20rate

A "murder rate", being a fair bit smaller, is almost always expressed per 100,000 people per year.

> If the population is increasing without constraint

It isn't. Population growth rates have rapidly declined all over the world. The US population would already be shrinking were it not for immigration. Even Bangladesh and India are currently reproducing at right around replacement levels.


> It is the standard definition when talking about a CRIME rate.

That wasn't your claim, and it's false in any case. There's no standard definition of "rate" -- you need to define your terms, and "rate" means a measure of change, not any specific measure without additional information. In any case, the exchange below tells me that you really don't understand what "rate" means.

>> If the population is increasing without constraint

> It isn't. Population growth rates [emphasis added] have rapidly declined all over the world.

Honestly. I say population is increasing. You reply by saying that the "growth rate", i.e. the first derivative of population, has declined. Both are true. To sort this out, take Calculus. Until you understand the difference between a function and its first derivative, we won't be able to discuss this issue.

http://www.arachnoid.com/calculus/

Also, for the actual worldwide population growth rate:

http://en.wikipedia.org/wiki/Population_growth

"In 2009, the estimated annual growth rate was 1.1%."

That rate, if sustained, means the world's population will double in 63 years.


> Honestly. I say population is increasing.

No. What you said was that population was increasing without constraint. It's that last bit I was responding to.

The fact that the rate of growth is declining does suggest the presence of one or more constraints, even if the overall growth rate hasn't quite reversed just yet. The primary form of the constraint appears to be that as childhood survival rates and living standards improve, families voluntarily choose to have fewer kids. This is happening all over the world, there's no reason to think it will stop any time soon, and it argues against your implication that "mass death" is the only likely way out.

(For what it's worth, I have taken calculus. And if you're the guy who wrote AppleWriter: Nice job! I liked that program. I used it in high school.)


> What you said was that population was increasing without constraint.

And that's correct -- any population increase above zero is without constraint. All one need do is calculate the doubling time:

http://arachnoid.com/lutusp/populati.html

> The fact that the rate of growth is declining does suggest the presence of one or more constraints ...

Yes, or more likely, it's a random fluctuation indicative of nothing in particular. The latter assumption, which may seem overly skeptical, is how a scientist is expected to look at an unexplained change in the rate, based on a precept called the null hypothesis -- a presumption that (in a manner of speaking) it's all random and meaningless.

> ... and it argues against your implication that "mass death" is the only likely way out.

Not really. Remember that nature efficiently picks out those with the highest birthrates and makes them the entire future population. My point is that population growth is naturally unstable, depending only on food sources and a willingness to push other species out of the way.

1.1% per annum may not seem like a very large rate of increase, but 63 years doesn't seem like a very long time to double the world's population either.

Exponential increases of all kinds are rather scary to model. They're scary enough when one looks at compound interest and how that naturally creates a chasm between rich and poor over time, but the "big show" for exponential increases is population, where much more is at stake.

> For what it's worth, I have taken calculus.

Okay, glad to hear it. Given that, I think you will appreciate what I thought when I heard you object that population couldn't be increasing because the rate of increase was decreasing.

> And if you're the guy who wrote AppleWriter: Nice job! I liked that program. I used it in high school.

Yes, that's me. Thanks.


>Yes, or more likely, it's a random fluctuation indicative of nothing in particular. The latter assumption, which may seem overly skeptical, is how a scientist is expected to look at an unexplained change in the rate, based on a precept called the null hypothesis

It's NOT UNEXPLAINED and it's NOT RANDOM.

Does your browser support Flash? If I point you at a chart composed using gapminder.org could you please please LOOK AT IT? (Or if you can't, could you let me know what the constraints are on what your browser can see, so I can find another way to get you the information?) The link will follow this paragraph. Press the "play" button below this chart to see an animated plot of fertility versus child mortality. Several specific countries are hilighted - you see a trail of their progress over time - but the ENTIRE MASS of countries follows much the same pattern - it moves down and to the left. Here's the link:

http://www.bit.ly/18hUMjA

It truly boggles my mind that you could think even for a second that the slowdown in population growth is "likely a random fluctuation indicative of nothing in particular". It is a TREND. The phenomenon is quite consistent. Across the ENTIRE PLANET, countries are increasing their GDP, decreasing child mortality and decreasing the rate at which they have kids, with all those changes roughly in tandem. And it's not a random walk where one needs to cherry-pick any specific interval to see these trends - if you just look at the entirety of ALL the data we have available and plot it, you can see a consistent movement.

Please look at the animation. Play around with it - use the checkbox list on the right to highlight other specific countries. Get a sense of what the data is actually DOING before you claim it's a random fluctuation.

A random walk WOULD be susceptible to cherry-picking - the trend would only show up if you pick certain time ranges or certain sets of countries.

This is not that.

(Given current trends, the world population would not double again - it would stop growing short of that.)


> You're making a common statistical error

You are the one that said things are worse compared to 1932. You don't get to have your cake and eat it.


> You are the one that said things are worse compared to 1932.

And they are -- much worse. More incidents in an absolute sense, and more incidents per capita.


>> You are the one that said things are worse compared to 1932.

> And they are -- much worse. More incidents in an absolute sense, and more incidents per capita

Except that there are almost certainly fewer incidents per capita compared to 1932. 1932 was near a local maximum in terms of violence rates in the US; currently we are near a local minimum. (The current homicide rate is less than half what it was in 1932.)


> Except that there are almost certainly fewer incidents per capita compared to 1932.

No, not if a reasonable moving average is performed. And per capita modifies the logic in a way that reduces the importance of population, when my point is the size of that population.

Here's a small interval graph that shows an increase per capita since 1960:

http://en.wikipedia.org/wiki/Crime_in_the_United_States#Crim...

Another graph that shows the same trend:

http://www.utahfoundation.org/img/briefs/2008_08_crime_fig4....

The above graph shows the risk of using too short a sampling interval. If you sample on the interval 1990 to the present, crime appears to have gone down. If you sample from 1960 or any longer intervals to the present, per capita crime has certainly increased.


> If you sample on the interval 1990 to the present, crime appears to have gone down. If you sample from 1960 or any longer intervals to the present, per capita crime has certainly increased.

How about instead of SAMPLING, and without performing any sort of "moving average", you just SHOW the homicide rate, and do so over a range wide enough to INCLUDE 1932. Can you do that?

If you do that, I think you will find that you are wrong about this - the homicide rate today is lower than it was in 1932. Because there was a huge peak around the 1930s PRIOR TO the relatively low point in the 1950s. I don't see much point in applying "a reasonable moving average" since the underlying trend is pretty smooth, but if you did, say, a 5-year moving average it would STILL be lower today than in 1932.


> How about instead of SAMPLING, and without performing any sort of "moving average", you just SHOW the homicide rate ...

The answer is simple -- it serves no purpose. If the point is to establish historical trends, and in particular if it's desirable to create a future trend line, the first thing to do is avoid the classic mistake of examining the raw data. (This is true in climate studies also, for the same reasons.)

In this case, a moving average meant to confirm or refute the thesis that crime rates are higher now than in the past, is best conducted with a long interval. Such an interval shows that crime per capita is much, much higher now than it was in the past, along with population (which also fluctuates over time).

The 1930s were a historical anomaly -- the Great Depression (an increased level of social upheaval and desperation), and prohibition coming to an end (which meant that gangsters had to think of a new way to make a living), arriving at once, make it unrepresentative. Any statistician hoping to establish social trends would treat that period as an obstacle to producing a meaningful trend line.

> I don't see much point in applying "a reasonable moving average"

Yes, I got that. And I won't be likely to successfully explain to you why it's needed.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: