A decline of violence? Week 1 (see below - second)
A decline of violence? Week 2 (see below - first)
A decline of violence? Week 2
Last week I posed the question, “Has there been a ‘decline of violence in history’?”. I asked whether events in Ukraine and the Middle East indicated that Steven Pinker’s claim of a decline was wrong. I compared Pinker’s graphs with those in this year’s PRIO report and noted declining battle deaths up to 2005 but increasing deaths since then. But did that mean that Pinker was wrong? What follows is a continuation of my investigations.
PRIO report, Conflict Trends: A Global Overview, 1946–2023. Figure 1 (section 2.1, page 9). https://www.prio.org/publications/14006.
The PRIO report distinguishes three categories of conflict: state-based conflict, non-state conflict and one-sided conflict (Sections 2, 3 and 4). In each case the number of conflicts and the number of deaths are given. In 2023:
State-based conflict 122,000 battle deaths
Non-state conflict 21,000 battle-related deaths
One-sided violence 11,900 fatalities
There is an important distinction between numbers and rates. If a population is rising then a constant number of deaths will give a declining death rate. Secondly, if the number of states (or in general the number of conflict actors) is rising then a constant number of conflicts will give a declining conflict rate.
State-based conflict (this was what was discussed in last week’s report). Over the period 1946 to 2023 there have been three peaks of around 250,000 deaths. Over the same period world population has more than tripled. So peak-year death rates have declined. In this sense Pinker’s thesis remains true up to the present time.
State-based conflict-affected countries by region (p. 21):
Africa 17
Asia 7
Middle East 6
Europe 3
Americas 1
Note: the population of Africa has increased six-fold in the period 1955 to 2024, from 0.253 billion to 1.515 billion.
https://www.worldometers.info/world-population/africa-population/
Non-state conflict (1989-2023, Figure 9, p.16). Three categories here are: formally organised groups; informally organised groups; and communal conflicts …
… Formally organised groups. Since 2005 there has been dramatic rise in number of conflicts and number of deaths involving formally organized groups. The increase has been about fivefold (much more than any population increase in that twenty-year period). There is marked variation between regions.
… Informally organised groups; and communal conflicts. There have been fluctuations but no overall trend over the whole period.*
One-sided conflict (1989-2023, Figure 11, p.18). There have been marked fluctuations but no overall trend over the whole period.*
*World population in 1990 was 5.3 billion and in 2020 7.9 billion. If the number of deaths was constant over that period then the death rate is now 67% of what it was (5.3/7.9=67%).
…
“Refuting the Pinker theory” – numbers and rates
The truth of a statement depends on what it means. The phrase ‘a decline of violence’ is not well-defined. It might refer to the number of deaths or it might refer to the death rate. If the phrase refers to the number of deaths then it is not always true to say that there has been a decline. This point is forcibly made by Dwyer and Micale (p. 10), quoting Malesevic who notes that the 120 million battle deaths in the twentieth century equal around two-thirds of all war deaths for the past five thousand years.
Malesevic is correct. But a primary explanation of the number of deaths is the size of the population. In looking for other explanatory variables it makes sense to look at death rates.
Dwyer, Philip & Mark Micale, eds (2021) The Darker Angels of Our Nature. Refuting the Pinker Theory of History & Violence. London: Bloomsbury Academic.
My review of Dwyer and Micale:
https://www.prio.org/journals/jpr/booknotes/283
…
The size distribution of conflicts – numbers
As is typical of size distributions there are a few large conflicts and many small conflicts. The report refers to large conflicts as ‘wars’.
[… ‘conflict’ to refer to both armed conflicts (with a minimum of 25 battle-related deaths per year) and wars (armed conflicts with more than 1,000 battle-related deaths per year). p. 23.]
“In 2022, Ethiopia (blue) and Ukraine (orange) were the two most dominant conflicts. In 2023, Ukraine continued to dominate and together with Israel/Palestine (yellow) escalated to be the two conflicts with the greatest number of battle-related deaths.”
Fig 6, p. 14.
“… in 2023, with nine wars. These nine wars included Burkina Faso, Ethiopia, Israel, Myanmar, Nigeria, Russia and Ukraine, Somalia, Sudan and Syria. Notably, for the first time since 2015, the conflict in Yemen was no longer categorized as a war.” p. 13.
So Africa had 5 wars; the Midde East had 2 wars; Europe had 1 war; and Asia had 1 war.
Populations in millions:
Burkina Faso 23.5
Ethiopia 132.1
Israel 9.4
Myanmar 54.5
Nigeria 232.7
Palestine State 5.5
Russia 144.8
Ukraine 37.9
Somalia 19.0
Sudan 50.4
Syria 24.7
https://www.worldometers.info/geography/alphabetical-list-of-countries/#google_vignette
…
The variables … numbers and rates (deaths)
A variety of variables are invoked in discussions of violence. Last week I referred to the number of battle deaths (the PRIO report) and the rate of battle deaths (Pinker). The distinction matters.
Result 1
If the population is increasing and the number of deaths increases or stays the same, then the rate of deaths decreases.
Number of deaths. In the period 1946 to 2023 there have been four major peaks: the first peak with over 500,000 battle-related deaths in 1949*; and then three peaks with between 200,000 deaths and 300,000 deaths, in 1971*, in 1983* and in 2022*. (*the years are noted from Figure 1).
Population. In the same period the world population has more than tripled. It was
2.2 billion in 1945; 3.7 billion in 1970; 4.9 billion in 1985; and 7.9 billion in 2020.
https://www.worldometers.info/world-population/
https://www.worldometers.info/world-population/world-population-by-year/
Rate of deaths. Suppose that the battle deaths were the same each year at 250,000. Then the death rates would decrease as the population increased: 1.15 in 1945; 0.68 in 1970; 0.51 in 1985; and 0.32 in 2020. The death rates here are measured in a unit of 100,000 deaths per billion population.
Despite large fluctuations in the number of deaths, the three major peaks have indeed been much the same in 1971, 1983 and 2022. So the death rates in these three years have shown a large decline. In this sense, Pinker is right: there has been a decline of violence.
...
Postscript
“Consider the following equation, defining death rate:
death rate = number of deaths / number in population [1]
The equation can be re-expressed as:
number of deaths = number in population x death rate [2]
One interpretation of this equation is that the number of deaths is caused by a combination of the two variables, the number in the population and the death rate.
In my argument I consider the situation where the number of deaths is constant. This gives equation [3] below:
death rate = constant / number in population [3]
This indicates an inverse mathematical relationship between the two explanatory variables. However there may or may not be a causal relationship. The two may be independent: the factors driving population increase may be quite different from the factors driving reduction in the death rate.
[In a related medical example, as population has increased so has neonatal death rate decreased. But this latter change has been due to improved medical care, not due to population increase.]”
To be continued
THE END
A decline of violence? Week 1
Eleven years ago, I presented Steven Pinker and Joshua Goldstein with the Conflict Research Society’s Book of the Year Award. Both books advanced much the same thesis which was expressed in Pinker’s subtitle as “The Decline of Violence in History …”.
Since then, Russia has invaded first Crimea in 2014 then Ukraine in 2022; and, amidst continuing tension in the West Bank and on the Lebanese border, and since the Hamas attack on Israel on 7th October 2023, there has been mass urban devastation in Gaza and now the recent onslaught on Hezbollah in Lebanon. President Zelensky of Ukraine visited Washington and Prime Minister Netanyahu of Israel addressed the UN on 27th September 2024. The same day, Israeli strikes on Beirut killed Hezbollah leader Hassan Nasrallah.
BBC, Nasrallah killed: https://www.bbc.co.uk/news/live/c981g8mrl8lt;
Israel and Palestine: self and other, positive and negative; 2023;
Ukraine ... and World Society 2023.
At last year’s CRS conference I presented a paper entitled;
“A history of conflict; a history of the Conflict Research Society, 1963-2023”
One section of my paper asked “Were Pinker and Goldstein wrong?”
Part 1 My talk:
Part 2: Background material for my talk
https://sites.google.com/view/values-world-society-modelling/part-2-background-material-for ;
Part of my talk looked at the PRIO report, Conflict Trends: A Global Overview, 1946–2022. This year’s report was published in June 2024. Figure 1 (section 2.1, page 9) provides a helpful overview. It presents the number of battle deaths in state-based armed conflicts 1946-2023. There are dramatic fluctuations with peaks in 1949, 1970, 1983 and 1998, with the height of the peaks themselves declining markedly over the period. The number of deaths reaches an all-time low in the year 2005.
https://www.prio.org/publications/14006.
This decline is consistent with Figure 6.1 in Pinker’s book which gives the rate of battle deaths in state-based conflicts, 1900-2005. The rate of deaths is almost zero in 2005. So Pinker was right.
But the PRIO graph changes in 2005, the last two decades showing a rising trend (with major fluctuations) and with two peaks in 2014 and 2022. Somewhat similarly, the number of conflicts of different types - civil, internationalised civil and interstate conflicts have all increased.
Does that mean Pinker is now wrong? For my answer see the aforementioned section of my paper ask entitled “Were Pinker and Goldstein wrong?”
Part 1 My talk:
…
To be continued.
THE END