Thread by @ProfPaulPoast

1/ Covered the "Decline of War" debate in yesterday's "Quantitative Security" grad class.

👇is a summary

2/ We began by showing the data that @sapinker used to support the claim, namely that battlefield deaths/population are declining

3/ Of course, we also see a decline if we focus on just battle death figures since 1946 (using data from @UCDP & @PRIOUpdates)

4/ The claim of war's decline is actually a very old.

Indeed, one of the first pieces of quantitative security research, by Woods and Baltzly, was a test of the "War is Diminishing" claim.

books.google.com/books?id=eD4uA…

5/ And here is the graph underpinning their main claim -- that the percentage of years that European powers are at war has declined over the centuries.

6/ To Woods & Batzly's credit, they acknowledge that more data are needed: "But the chart, as it actually stands does not do more than throw a moderate amount of probability in favor of declining war years."

More data are needed = Enter Quincy Wright (covered last week)!

7/ Here is what Wright found in "Study of War".

This figure looks at causalities from battles. Given that Wright is writing in 1942, not surprising that he doesn't find a decline in war violence.

8/ Nevertheless, the empirical claim that war is diminishing has influenced discourse among security scholars.

One example is this @PoPpublicsphere symposium on the decline of war (featuring @PageFortna, John J Mearsheimer, & Jack Levy)

cambridge.org/core/journals/…

9/ The class then focused on pulling the "Decline of War" claim apart, starting with @tanishafazal's piece in @Journal_IS on improved battlefield medicine (and other critiques of Pinker's data)

mitpressjournals.org/doi/10.1162/IS…

10/ We didn't have time in class to discuss all of the various works in this debate, but the syllabus has them in the "Recommended Reading" section

11/ One of the best ways to evaluate this debate is to look at the data ourselves. So we did just that!

For the purposes of the in-class examples, we used the Correlates of War war data

correlatesofwar.org/data-sets/COW-…

12/ We created two graphs (DM me if you wish to see the in class @Stata do files for creating the graphs).

For the first graph, we looked at deaths from interstate wars from 1816-2005. The below graph shows the log(battledeaths) for each war, by the year that the war began.

13/ Students noted that the second half of the 19th century looked an awful lot like the second half of the 20th century. 🤔

Hence, they said there didn't seem to be much of a decline (or rise) in the trend!

14/ Why did we use log(battledeaths) instead of just battledeaths? Following Richardson, the data are highly skewed (due to WWI and WWII)

15/ For the second graph, we looked at the number of wars (inter-state, civil war, or extra-state) taking place per year. Here is the graph:

16/ What it shows is pretty clear: given COW's classification of "war" (which is probably overly conservative), there has not been a single year without a war since 1816, and the number of wars rose, not declined, in the late 20th century.

17/ Are these the only two indicators we could evaluate? No, there are many more!

But these data show that, at minimum, claims of "war's decline" are probably just a bit exaggerated.

Overall, a very productive class!

(end)