Democide in 20th century

Professor Rummel has done a lot of research in democide. I picked up this table in his site about democide in 20th century.

One of his observation is that the more power you have, the more damage you can do. Not surprising, considering the biggest democides occurs in USSR and China, both are large countries.

But it might be more illuminating if we look at other measures as well. e.g.,

  • the notorious Khmer Rouge and Rwanda fell quite low in his list in such measures
  • the dictators will not succeed if there was not enough collaborators in their countries to carry out the orders or to act even without order (but with silent permission)
  • if you are put into the location in the time involved, what is the chance that you are killed, and what is the chance that you would have killed another human being?

So I decided to look at another measure, murder per capita. This is not to say the whole nation commit murder collectively, instead, the theoretical basis is that the % of collaborator within its population is roughly the same across population. It is not a well supported assumption, but it is indicative to certain extent (and it assumes no cultural or racial difference of % people turning wicked and crazy).

I then used Rummel's figures, and did a very rough estimate of the population of these countries during the relevant period, by averaging the beginning and end period populations of these countries.

CountryDemo -cide (k) Pop (M) YearMurder /cap
USSR61,911 141 193743.9%
Cambodia2,035 7 197527.9%
Germany20,946 87 193924.1%
Turkey1,883 17 1900-5011.1%
Japan5,964 63 1900-509.5%
Poland1,585 25 1900-506.3%
CCP38,702 650 1900-806.0%
Yugoslavia1,072 23 19874.7%
Vietnam1,670 43 1950-853.9%
Pakistan1,503 78 1950-801.9%
KMT10,075 530 1900-501.9%

A few notes

  • This is an incomplete list based only on Professor Rummel's table, he did not include the 1M or so killed in Rwanda. I am also curious as to why Korea War was labeled "suspect", perhaps because the numbers was too hard to verify
  • This is a very rough esstimate, considering error in both number killed and also in averaging the population base
  • I used "murder/cap" as a measure, because most of the cases occurred within a generation. The population base for USSR might be bigger if we count across generations, but I suppose most of the deaths are from 1924-1953, i.e., Stalin era.
  • The numbers do not include indirect kills, such as famine in Great Leap Forward, and the famine and wounded death of civilian under German and Japanese occupation, or German/Japanese soldiers/civilians who were killed by allies, but the ultimate responsibility should be shouldered by the war aggressors
    • Including 32M died in Great Leap forward, CCP's murder per head is 10.9%, immense for such a huge country
    • Indirect death (including famine death and those killed by CCP/KMT resistance) during the Japanese invasion in 1937-45 in China, according to Rummel was(line 143) 19.605M, 5 times the direct death of 3.949M inflicted by Japanese army; If we assume the same proportion for SE Asia, the total democide would be 5x5.964=29.82M; i.e. 47.3 death per 100, so Japan (and Germany) topped USSR if including indirect death
    • Indirect death inflicted by German (Russians killed) probably need to multiplied by a similar factor
    • Some argued that Nazi German is much worse than fascist Japan, we can quantify it now. Nazi is about twice as worse than fascist Japan, and fascist Japan probably beat fascist Italy by a much larger factor because Italy did not make the top 10 list by Rummel
  • Some of the death occurs during the process of self defense, e.g. KMT (China) conscripts died in the process of movement and desertion during Japanese invasion; and America/France were repsonsible for many of Vietnam's death. Some of the USSR kills should probably be counted on Hitler's bill by similar reasoning
  • Arguably CCP and KMT could be grouped together under China. I followed Rummel's classification because most of the KMT kills was before 1949 (though it continues in Taiwan, but the population base was much smaller), and CCP kills were after 1949. So the crimes were commited by different generations.

Here are a few obseravtions

  • USSR (under Stalin) tops the list, on average a 100 people killed 44, mostly their own people, followed by Khmer Rouge's Cambodia
  • If we do not limit the numbers to 20th century, Ghenghis Khan may top the list, because the Mongolian population is much smalled than those they killed
  • Turkish surprised me by making #4, I guess it is the Amernian genocide, civil war, plus hte first world war (for which the population base should probably be much larger).
  • The biggest muder on its own people are from dictatorships such as USSR, Cambodia and CCP
  • The biggest threat to neighbors are Germany (24/100 direct, perhaps 100 or 120 for indirect kills) and Japan (9.5/100 for direct and 47.3 for indirect kills)
    • If we accept that the German had sincerely reflected, it makes Japan the most dangerous aggressor and threat to its neighbors
    • On average, about one in every 2 Japanese is responsible for the death of one person. If we assume the women basically did not participate in the killing, then on average every one man in Japan living between 1937-1945 was responsible for a death, indirectly
    • But the above is only an indicator for you to see the scale of the horror, because many Japanese men (maybe 50-70%) did not participate in the war and those who did each scored an average of 2 or 3 death
    • This does not include the innocent death (and indirect death) in Hiroshima/ Nagasaki, or the Tokyo firebombs, because Tojo and his collaborators should ne held responsible, not America. It would top the 29.82M number up to around 30-31M
    • The millions of death in Russo-Japanese war in 1905 was also not counted, but that likely involves a different generation of criminals.

These are horrible things. We should never let them happen again.



Elam Bend said...

I think Rwanda is not there because it is not really quite an instance of government killing it's own people.

Sun Bin said...

it is probably not there because the timeline for that table is for 1900-1987 (e.g. Yugoslavia is for Tito, not the 1990s war)

Sun Bin said...

this is only one way to look at it. i am sure there are many.
i should say it is more accurate to say japan's (and germany) aggression caused the biggest damage to its neighbors.

you are right. rummel also used per year as one indicator (bottom of his tables). so i did not repeat what he already calculated.
rwanda and cambodia are pretty bad in such measure. so would great leap forward (although, that is 'indirect' kills)

Sun Bin said...

but it also make sense to use population as a basis, if one limit it on a generation (as i explained in the beginning of the post).

the number for USSR probably would drop a bit, also germany (the 2 WW would be counted separately), and so on. but my rough measure still tell something about the scale of crime.

Sun Bin said...

i saw that. it is how i discovered rummel's compilation.

IMO, rummel should not have waited for the Jung Chang books to do this. Indirectly responsibility has been very clear even 20 years ago. Jung Chang's book is full of rubbish and it hurts the credibility instead.

you should notice i had included indirect death so that the figure for china would be above 10% in a paragraph above.

Anonymous said...

Very, very interesting.

Include the deaths of the people in Hiroshima, Nagasaki, and Tokyo under Tojo...It's certainly what the US military would say, but does that mean that Hitler was primarily responsible for the people who died at Dresden? You'd need to accept all of these operations as militarily necessary to achieve victory. I'd probably agree with you, though. Usually the aggressor must accept the consequences when responsibility is in doubt.

Japan, even rearmed, is hardly the most dangerous aggressor. In the new age of warfare, perhaps we should find a way to measure kill efficiency, in which the United States is unparalleled. That is, when a target is chosen, what % of the time does the aggressor achieve success, and how long/how much does it cost to do so? The United States might be efficent, but spend massively to achieve its goals. Non-state actors might actually be more economical killers because of asymmetric warfare.

I wonder where Russia would be today if not for the tens of millions of deaths in the past century.

Sun Bin said...

btw, the kamikaze kids are definite under tojo's bill.

Anonymous said...

Sun Bin - Read "We wish to inform you that tomorrow we will be killed with our families" a journalist's investigation of Rwanda. It's a really moving book.

nice site by the way.

Anonymous said...

RE: Japan

Two questions and some observation

1) Whose figures are you using for those killed. China's estimates are known to be 'on the high side', others consider America's estimates to be too low (for example, Chinese figures for Nanking are 300,000, US figures are 140,000)?

2) Are you only counting people who were killed directly and indirectly by Japanese troops (starvation, exposure, gun shots etc) or are you also counting Chinese who were killed by Chinese (KMT V communist) during the occupation? and what about Chinese and whites killed by accident by the US etc (EG Korean slave laborers killed when the US bombed Japan). Shouldn't these be on the US tab and off of the Japanese tab as Japan didn't intend for them to die?

By my count, indirect deaths should not be counted equally to direct deaths because there was no intent behind them.

Many Millions of Chinese died as a result of starvation and exposure. Some of this was from the Japanese disruption of food production and mobility, but some of it was the result of Chinese being marched to death by other Chinese during evacuations. They might have otherwise have survived had they stayed put. I think that these should be split between the Chinese and Japanese tabs (ratio to be decided upon later).

Also, what about Chinese-born Japanese?

You probably do not know this, but there was a substantial Chinese-born Japanese population in Manchuria. Are they Chinese or are they Japanese? and should their deaths be counted on the Chinese or Japanese tabs?

I ask this because many did not consider them selves to be Japanese (and many, particularly those of mixed parantage, were not considered to be Japanese by Japan either) and did not support the war. Were they Chinese of Japanese origin whose deaths were the fault of Japan? Were they pure Japanese whose deaths were the fault of Japan (and if so should they cancel out themselves or be counted on the Japanese tab)?

Where they Japanese who were killed by Chinese and so should be counted on the Chinese tab? Where they Chinese of Japanese origin who were killed by Chinese/Russians who should be counted on the Chinese and Russian Tabs?


Anonymous said...

When calculating the Chinese population, from what point are you calculating it, and over what teritory?

Tibet, Chinese-Taiwan and east turkistan have a substantial combined population, but at certain points in history since the start of WWII, all three have left offiical Chinese control for a time (Taiwan through Japanese occupation and its current state of autonomy, Tibet and East Turkistan through declarations of independence). If you remove them or add them, they will throw out your figures by several tens of millions.


Sun Bin said...

I used Rummel's figure (in the links). He has 3 estimates, high, medium, low. And he (and I) used the medium numbers.

(I guess high would correspond to 300k, low to 100k. his estimate for DIRECT death by Japanese is 1.578, 3.949, 4.547. and I used 3.949. Total death range is 10.628, 19.605, 37.042. I used 19.605)
see this table

Sun Bin said...

to your second question.
This is how I interpret Rummel's table.

Directly responsible # (3.949M) does not include any of those ctegories you mentioned.

FYI, according to Rummel's classification, KMT "killed" more people than Japanese 'directly', which includes killing CCP, deserters and blowing up the dam in Yangtze.

But I have to give this to Japan in the "Indirect #". I also included death by famine. As I said, the invasion is the ultimate cause of these deaths, even though CKS is no less guilty in blowing up the dam.
However, I presented both numbers so that one can see the difference between the two classifications.

Sun Bin said...

"Japanese borned in Manchuria/etc" I think you have to dig into Rummel's source. (But as you can see, the range of estimate is so large that it probably change the results very little)

But AFAIK, the death includes everybody. i.e. KMT killls includes Japanese soldiers killed in action. civilian death should include all races. But I am not 100% certain.

again, if these people are counted. i believe they are not in Direct #, but in Indirect #.
i.e. Tojo should be held responsible for their death as well, though only in the indirectly/unintentionally category.

Sun Bin said...


to your last few questions.
IMO everybody should be counted under Tojo. My estimate did not include Japanese who are (indirectly) killed in the pacific theater, the kamikaze kids, the japanese innocnt civilians who died in Hiroshima and Tokyo.
If Tojo did not start the war, they did not have to die.

But I would be careful to put them into the 'indirect' category.

Sun Bin said...


I believe Rummel considers Tibet and Xinjiang as Chinese territories. and therefore, yes, the numbers included them.

the total population of Tibet was 2-3M, Xinjiang 4-6M, Taiwan 10-12M in 1940s. what do you mean by 'tens of millions'?
if you are talking about WWII, there was basically no war in these areas.