Papers by Melberg
Elster Page
Ph.D work

About this web
Who am I?
Mail me
Search papers
List of titles only
Categorised titles

General Themes
Ph.D. in progress
Political Theory
Various papers

The Questions
Ph.D Work
Statistical Problems
Social Interaction
Centralization vs. Decentralization

Define economics!
Models, Formalism
Fluctuations, Crisis


Review of textbooks

Belief formation
Inifinite regress

Collapse of Communism
Political Culture

Political Science
State Intervention
Nationalism/Ethnic Violence

Yearly reviews



[Note for bibliographic reference: Melberg, Hans O. (1996) Dynamic counterfactual correlations : Some comments on B. Caplan's "The Mensheviks' Critique of Bolshevism and the Bolshevik State" ,]

Dynamic counterfactual correlations
Some comments on B. Caplan's "The Mensheviks' Critique of Bolshevism and the Bolshevik State"

1. Introduction
There are two ways to measure how important a difference is: static and dynamic. By a static difference I mean simply a description of how the position of the Bolsheviks and the Mensheviks differed at one point in time. A dynamic perspective implies an examination of the consequences of the static differences. The distinction is important because small initial static differences may develop into a large difference over time. As an illustration consider a marble on the top of a mountain. A small difference in the position of the marble on the mountain may imply very different end position over time because the difference determines which direction the marble will take down the mountain.

In his paper Brian D. Caplan is aware of this distinction, although he tend to focus somewhat more on a description of the static differences between the Bolsheviks and the Mensheviks. His basic argument is that the difference between the two factions was small compared to their overall agreement on Marxist doctrine. As for the dynamic analysis, Caplan argues that the small differences that existed would not lead to a radically different result than that achieved by the Bolsheviks. My basic position is that Caplan's argument is correct on both accounts. This does not mean that the paper is flawless. Although I agree with Caplan that the small difference between the Mensheviks and the Bolsheviks do not imply a large difference over time, I have less confidence in this conclusion than Caplan.

2. Would the history of the Soviet Union be different if the Mensheviks had gained power instead of the Bolsheviks?
Caplan answer is no and he presents two main arguments. First, that the Mensheviks basically had a dogmatic Marxist outlook and - based on their intellectual history - there was little indication that they were going to change their minds (develop in a more democratic direction), nor that they were particularly interested in issues of civil liberties. In short, the Mensheviks do not pass the acid test of allowing your enemies basic civil liberties such as freedom of speech. In fact, Caplan believes this follows directly from Marx's writings and is demonstrated by the fact that even the French and the Italian communists remained anti-democratic for a long time. This is supposed to show that anti-democratic tendencies are inherent to Marxism. Caplan's second argument is that it is possible that the Mensheviks could have resorted to violence on a massive sale just like the Bolsheviks did.

I tend to agree with Caplan's first argument, but I believe it is too weak to carry much confidence. It is weak because one cannot take the dogmatic intellectual history of a group out of power as a proof that they will remain dogmatic in power. As Caplan himself points out: People in power have to take tough choices which philosophers can avoid. This change in circumstance makes the inference from "out of power dogmatism" to "in power dogmatism" less than perfect.

I place even less confidence in Caplan's arguments relating to the French and Communist parties. True, it is easy to find deeply anti-democratic quotes in Marx's writings. It is also true, but less so, that the French and the Italian Communist parties remained anti-democratic (less so because the French Communist Party, if I remember correctly, changed its mind in 1974, and the Italian even earlier). Yet, the behaviour of the French and Italian communists when the Bolsheviks ruled in Moscow, cannot be taken as a sign of how they would have behaved if the Mensheviks had gained power. The inference is wrong because the behaviour of the French and Italian communists was determined precisely be the fact that it was the Bolsheviks who were in power in Moscow. With Mensheviks in power we do not know what would have happen and to infer that the French and Italian communists would have developed in the same direction as they did when the Bolshevik were in power, is false. (There may be an analogy here to the problem of the false counterfactual in game theory - see Jon Elster's article Some Unresolved Problems in the Theory of Rational Choice and his notes on counterfactuals in the book "Logic and Society")

Caplan's second argument is a bit curious. He admits that the Mensheviks were less likely than the Bolsheviks to resort to mass violence, but he qualifies this by the following argument:

"Still, how complimentary is it to say of a political party that, "They probably wouldn't have resorted to mass murder and totalitarianism"? The implication is that they possibly would have."

I believe this is false reasoning in the sense that Caplan cannot draw the conclusions he want to draw from this argument. The question is whether it is likely that history would have been different with the Mensheviks in power, not whether the Menshevik position on violence is complementary or not. By his own admission the answer to the first is that it is likely that the Mensheviks would have acted differently. The qualification does not alter this fact, it simply brings up a new and irrelevant (in the present discussion) issue.

In addition to these objections I want to add a third: Counterfactual statements about the development of whole societies are inherently very uncertain. In short, we need a general theory about causal connections in society in order to make these statements - a theory which we do not have (and, quite possibly, never will have).

It is important to note that I do not really disagree with Caplan's substantial position. I agree that there are strong anti-democratic tendencies in Marx and that the Mensheviks were no real democrats. The point is, however, that I am less confident than Caplan about counterfactual assertions about how history would have developed if the Mensheviks had been in power.

2. Laws, Correlations and Theory?
Caplan writes that:

"Even if this trend [that non-democratic communist parties tend to develop into more democratic social-democratic parties] in the history of socialism correlated perfectly with the facts, it would remain a mere historical truth (such as the "law" that American presidents elected in years ending in a "0" invariably die in office - incidentally refuted by the case of Ronald Reagan) until justified by a cogent theory."

The distinction between the reliability of a belief based on correlation, as opposed to the reliability of a belief based on a theory of causal connections, is both valid and important. For example, assume that three people dies after a dinner with ten participants. You then hear that these three had a dish which none of the other seven had. Some would argue that it is reasonably safe to say that the three did from food poisoning. Others would argue that the correlation is not enough, that we have to examine the three medically to see whether it actually was the food that caused them - who knows, maybe they all got a heart attack at the same time? (For more on this see the debate between G. A. Cohen and Jon Elster in Theory and Society no. 11 and 12, 1982 and further exchanges in Inquiry). However, I have two objections to Caplan's use of this distinction.

First, I do not think one can give exclusive weight to one of the above methods for proving a belief. Both correlation and a theory of causal connections are important. For example, we do not know the causal link between cancer and smoking, but the statistical evidence that there is a relation is strong. Hence, it is false to dismiss the belief (that non-democratic Menshevik type parties develop into democratic parties) only because it is based on correlation and not a theory. It is false because it amounts to giving exclusive weight to theory in belief formation - ignoring information from correlations. (For more on this see my article "Against Correlation" - at - in which I argue, as Caplan does, against correlation as evidence. However, I do not go as far as dismissing correlations altogether.)

Second, the comparison with the "law" of dying American presidents and the law of democratic development of parties is invalid. In short, it seems possible to a give plausible causal story as to why radical parties tend to moderate their positions, while it is not possible to develop a plausible connection between the year in which a president is elected and if he dies in office or not. For example, psychological theories may suggest that new converts have radical beliefs, while real life experience over time tend to moderate these beliefs. Caplan may argue that he does not explicitly draw this comparison, in which case I would simply argue that his dismissal of the "law" of moderation is only half the story since he does not even mention that it is possible to develop plausible stories of this kind.

4. A short critique of one argument
Caplan ends his paper with the assertion that the Mensheviks ignored the central question, which in his opinion is the following: 'What reason do I have to believe that the economic condition of workers under socialism will be better than under capitalism?" I agree with Caplan that this is indeed a central question, but I do not think this is ignored in Marxist thinking. Indeed, their central argument is that capitalism develops alienation which is unjust. Socialism on the other hand, they (wrongly) believed would not create alienation and therefore socialism was better for the workers than capitalism. Another argument is to be found in the theory of central planning which they believed was more effective (again, wrongly so) than capitalism since they could avoid some of the waste inherent in capitalism (for example, unemployment). Hence, I agree that this is a crucial question, but I disagree with Caplan that the Mensheviks ignored it. (For more on the centrality of this issue see my paper "Logical Logic" - a review of M. Malia at

5. In conclusion
A review is by its nature critical and I have focused on what I believe were the weak points of Caplan's arguments. In conclusion I would like to balance my arguments by emphasising that I am fundamentally in agreement with the tendency of Caplan's paper. My main point was, however, that I am less certain that Caplan appears to be since I have little faith in counterfactual comparisons.

[Note for bibliographic reference: Melberg, Hans O. (1996) Dynamic counterfactual correlations : Some comments on B. Caplan's "The Mensheviks' Critique of Bolshevism and the Bolshevik State" ,]