On the statistical properties and tail risk of violent conflicts

post by morganism · 2017-02-23T03:46:22.671Z · LW · GW · Legacy · 5 comments

This is a link post for https://arxiv.org/abs/1505.04722

5 comments

Comments sorted by top scores.

comment by gwern · 2017-02-23T16:45:50.779Z · LW(p) · GW(p)

I recall reading this when it was released, and my problem with it then was simple: a long peace must be evidence for a decrease in war risk, because a long period of many wars/casualties would be evidence for an increase in war risk; all they do is demonstrate that long peaces are somewhat consistent with power laws, when what they should be doing is estimating the decrease and discussing how much evidence the long peace provides for a decrease in risk.

Replies from: bsima
↑ comment by bsima · 2017-02-23T18:05:04.659Z · LW(p) · GW(p)

"a long peace must be evidence for a decrease in war risk, because a long period of many wars/casualties would be evidence for an increase in war risk"

I do not think this is true. Perhaps you are confusing the incidence of war and the risk of war. A long period of many wars would be evidence for an increase of war, not an increase of war risk.

At about 1:00 in this video NNT explains how a small amount of benefit also exposes you to an increased risk; the corollary would be: an increase of peacetime exposes you to an increase risk of war. (The video is meant to provide the intuition, not a proof.)

Replies from: gwern
↑ comment by gwern · 2017-02-24T01:19:32.268Z · LW(p) · GW(p)

Perhaps you are confusing the incidence of war and the risk of war. A long period of many wars would be evidence for an increase of war, not an increase of war risk.

Since it was using casualties I'm not sure what you think the difference is.

At about 1:00 in this video NNT explains how a small amount of benefit also exposes you to an increased risk

If we want to assume far-fetched models with no evidence. Note that if you are trying to use this in conjunction with OP, you are guilty of fully general counterarguments: 'long period of wars? proof of risk! long period of peace? actually, it just makes the risk even worse!' No matter what happens, Taleb will claim he's right, like he usually does.

comment by morganism · 2017-02-23T03:47:30.439Z · LW(p) · GW(p)

"The dual method allows us to calculate the real mean of war casualties, which proves to be considerably larger than the sample mean, meaning severe underestimation of the tail risks of conflicts from naive observation. We analyze the robustness of our results to errors in historical reports, taking into account the unreliability of accounts by historians and absence of critical data. We study inter-arrival times between tail events and find that no particular trend can be asserted. All the statistical pictures obtained are at variance with the prevailing claims about "long peace", namely that violence has been declining over time. "