Other things not being equal

In my previous essay Proven politics I explained how political science (at least one version of it) takes the path of least resistance in working out supposedly scientific information on political systems. The assumption that political systems can be considered so independent of society that they are amenable to mathematical analysis is so egregious, and the quantified variables so artificial, that no conclusions reached by these methods can be convincing. The philosophical justification for this form of political science is very weak and its aspirations for exactness and objectivity reveal a seriously misguided view of social phenomena.

However, we can’t entirely discount the possibility that ceteris paribus (all other things being equal) assumptions could be formulated for political science with a greater measure of philosophical cogency and plausibility. After all, it’s a question of degrees. A very strong assumption, like the one I discussed in Proven politics, legitimates even quantitative analysis in principle. But the results of that analysis are only as plausible as the assumption itself is.

Weaker assumptions which recognize that each political system is to some extent idiosyncratic do not legitimate any quantitative analysis but may be fruitful for qualitative comparisons. The analysis is more plausible although its conclusions have to be more modest. In this essay I will discuss how weaker but philosophically justifiable all-other-things-equal assumptions can yield tentative conclusions about the nature of political systems.

One general theme from Proven politics will be central for this essay as well: comparison. When we try to understand complex social phenomena like political systems, there’s no hope of understanding it in isolation, by itself. We have a better chance of understanding it by studying how it changes or comparing it to other political systems.

My suggestion is that political science has much to learn from historiography, that is, the art of writing history. Historiography also relies to a great extent on comparison, differences and similarities, changes and invariances. Historians must take the limits of the available sources seriously and profess ignorance on any subject which exceeds these limits. It seems to me that such external checks are sorely needed in political science where the pseudo-science of Proven politics still prevails.

Systemized politics

It will be useful to start with some general considerations on complex systems. As I already noted in Proven politics, the lack of experimental control in social science shifts the control burden to the investigator. It is up to researchers to formulate conditions which show when all other things can be considered equal and to critically judge the results that their analyses produce. It will serve us well to consider in a bit more detail why this is a difficult task.

In a book called System Effects Robert Jervis discusses the theory of systems as it applies to social phenomena. Although his principal emphasis is on the study of international relations he also makes some general points which apply to other areas of politics. He summarizes these in the following way.

“Much of our world is unintelligible without attending to three kinds of interactions which form and generate a system. Two or more elements produce results that cannot be understood by examining each alone; the fate of an actor’s policy or strategy depends on those that are adopted by others; behaviour alters the environment in ways that affect the trajectory of actors, outcomes and environments.” (SE p.60)

Let’s look a bit closer at these interactions, one at a time. The first one is about combined effects. When elements A and B interact, they may produce a result which is very different from what elements A and B separately would have led us to expect. For instance, in an election a populist politician may surprisingly strike a chord with discontented parts of the electorate which usually have not voted at all. Populism by itself would not be effective if there wasn’t much discontent, and discontented people would not vote if the available alternatives were just the usual ones . But when these two elements interact, the result can be stronger than anyone could have expected.

A second example can be given from party politics. A change in party leadership, for instance, may have a larger influence on the political agenda than most people expected if the new leader feels that there was too much compromise and backpedaling under the previous leader. Again, that there have been compromises before does not by itself necessarily lead to a more aggressive agenda, neither does a change in leadership, but the two elements may together produce a surprisingly strong effect if the new leader feels a need to be more assertive.

As a third example, a chosen policy may have surprising effects when all repercussions have not been (or could not be) considered. Adopting a tougher stance against drugs may increase drug-related violence if rival gangs squeezed by the police have to fight each other for their share of a reduced market. It’s also difficult to predict whether drug legalization would be bad for society because more people would use drugs, or good because the harmful effects of illegal trafficking would go away. The answer depends on so many variables that only trying it in real life can give an answer to the question.

This lack of predictability is typical for all system effects. Let’s proceed to the second interaction noted by Jervis, that the fate of an actor’s strategy depends on those adopted by others. One example of this interaction is the prisoner’s dilemma, but we can find many practical examples as well. It is obvious that the fate of a voter’s decision to vote for a particular candidate depends on the strategies adopted by others. Most of the time the decision is inconsequential – the candidate would or would not be elected regardless of this one vote – but sometimes the last vote may be crucial.

If we think of politics more generally, the process of government formation in multi-party democracies is a good example of a situation where choices made by others determine the outcome of a party’s strategy. Since there are usually no detailed rules for how the government should be formed, the process is one of complex bargaining and compromise. Any situation where multiple players bargain with each other is a situation of uncertainty. One can have a clear strategy in mind going in, but one must be ready to change this strategy on the fly in response to other’s actions.

And finally, participation in many forms of international cooperation will be good for a country only if the venture is successful, but it will be successful only if enough countries sign up. Participants in a military coalition, for instance, risk fighting their war from a position of weakness if they commit to it prematurely without anticipating the intentions of other potential participants.

The third and final interaction which Jervis notes is that the behaviour of an actor alters the environment. A well-known example is the tragedy of the commons, where the combined effect of self-interested actions produces an unnecessary negative effect. But if we look for more political examples we might specifically note actions that take place in the gray area between legitimate and illegitimate campaign financing and political lobbying. Financing and lobbying are natural parts of democratic politics, but if the system is abused in a manner which hints towards corruption, the environment may be changed by stricter legal regulation.

Specific policy choices may also alter the environment in unpredictable ways. This applies particularly to new forms of taxation. Policymakers necessarily have very little empirical information about the preferences and expectations of people or companies. They cannot even guess what effects the aggregate reaction to a tax “adjustment” will have on society. This is why changes in the tax system are always small. Nobody ever presents a proposal for reforming the tax system from top to bottom, because the effects of large changes would be completely unpredictable. Keeping changes small is playing it safe.

All these examples illustrate once again why the “scientific” approach of Proven politics produces meaningless results. But the more important implications of systemic interactions in politics is that

When we are dealing with systems, things cannot change one at a time – everything else cannot be held constant. (…) When the outcome emerges from the interaction of several elements, it may not be possible to determine the contribution of each taken separately. Neither can we readily determine the effects of a policy by comparing a case in which an actor followed one policy with another case in which it adopted a different one: it is likely that the actor chose to behave differently because it saw the two situations as being different. Furthermore, because actors are part of the system, it may not be possible to judge progress or the success of a policy by using a fixed yardstick. We can, however, try to trace the ramifications of an action and detect the operation of the system’s dynamics. (SE p.73-74)

The last sentence is of particular importance because it hints at this fundamental fact: if we want to speak intelligently about politics, we must speak historically.

Political analysis as history

With the help of Jervis’s systemic analysis, we can see why strict all-other-things-being-equal assumptions rarely yield interesting information about political systems. They are unrealistic. Yet political systems do have some similarities and we should not completely give up the idea of comparing them.

It is therefore instructive to see why the study of politics resembles the study of history more than other fields of social study do. Jervis’s three characteristics of systems – combined effects, the strategies of others and the changing environment – apply to social systems in general, past and present. But there’s a specific congruence between the study of political and historical systems which I will trace below.

I will make use of Paul Veyne’s excellent book Writing History, where he writes:

First, every history is in some way a comparative history. For the features considered relevant, in relation to which an individual fact is described, are universal in that way. (…) Next, every “fact” is surrounded with an implicit margin of the non-eventworthy, and it is that margin that leaves room to arrange it otherwise than has been done traditionally. (WH p.46)

This applies well to all social systems. We cannot describe them without a vocabulary which assumes that the relevant features are at least to some extent universal. So in a way we make a linguistic all-other-things-equal assumption even when no explicit comparative analysis is involved. But the second point is even more important: the margin between relevant features and non-eventworthy features is only implicit. It is not fixed as in Proven politics. It leaves room for a multitude of interpretations and perspectives.

The thing which links political analysis particularly closely to historical study is the relationship to source material. Veyne writes about historical sources as follows:

Let us begin with a very simple historical proposition: “Louis XIV became unpopular because the taxes were too heavy”. (…) a phrase of this kind may have been written with two very different meanings. (…) Written in its first meaning, the sentence means that the historian knows, from documents, that the taxes were indeed the cause of the king’s unpopularity (…). In the second meaning the historian knows only that the taxes were heavy and that, in other respects, the king had become unpopular at the end of his reign. (…) In the first case he relates a plot to us that he has read in documents, in the second he makes a retrodiction, working back from the unpopularity to a presumed cause, to an explanatory hypothesis. (WH p.145)

Political analysis must also consist of similar retrodictions from written and spoken evidence to explanations. In retrodiction as Veyne understands it each explanation is particular, not general. There is no unique and final explanation to be found. Each investigator relates a different plot and different explanatory narrative.

History does not explain, in the sense that it cannot deduce and foresee (…); its explanations are not the referral to a principle that would make the event intelligible, but are the meaning that the historian gives to the account. (WH p.90)

But why does political analysis hinge on documents more than other fields of social study? Because of the essentially concealed character of politics. Much like history, it is available to us only indirectly. When the people who are involved in a political system make their opinions known directly, they have a specific political agenda in mind. Polling or interviewing them, to the extent that it’s possible, necessarily produces biased answers. That’s why critical interpretation of a range of documentary evidence is crucial if political analysis is to achieve any measure of objectivity. In this respect political analysis differs a lot from sociological fields where no equally powerful interests are at stake.

Much like the historian, a political researcher must impose their own interpretation on the evidence, keeping in mind

(…) that the documents, which provide us with the answers, also dictate the questions to us; in that way they not only leave us ignorant of many things, but they also leave us ignorant of the fact that we are ignorant. (WH p.13)

Just like in history, each event will be one of a kind because the system effects which contributed to bringing it about will not occur in the same constellation a second time. Other social phenomena are just as complex and unique as political ones, but system effects can be dealt with more easily when polls and interviews can be reliably applied.

So what does all of this mean in terms of all-other-things equal assumptions and comparative analysis? First of all that there will be many assumptions and that they will not be absolute or exclusive. They will be more like a flexible hierarchy of assumptions. The political analyst will begin with a small number of key features and assume that these are unambiguous general aspects of the political system. But in keeping the margin toward non-eventworthy features implicit, as Veyne puts it, the analyst leaves open the possibility of changing prioritizations.

The value of comparative analysis is that it gives the political analyst pointers towards the key features on which the study can be focused. The historian needs a theoretical framework for conceptualizing society. Without that framework it is impossible to describe how society changes. Historians get this framework for comparative analysis, which means from their working knowledge of other time periods and societies and especially from the generally accepted historical vocabulary. Similarly, political analysts need to conceptualize the political system to separate their most interesting features from idiosyncratic ones and to formulate a hierarchy of all-other-things-equal assumptions. Their knowledge of foreign political systems is important in this formulation process.

Ignorance in politics

To summarize, I began this essay by explaining in a bit more detail why aping the methods of natural science is so unproductive in political research. In social systems things hardly ever change one at a time. Instead different elements interact to produce results which could not have been predicted from separate actions, the choices of other actors determine the success of one’s own plans and each action may also change the surrounding environment. That’s why we have good reason to be sceptical of supposedly exact “explanations” for social phenomena.

I also emphasized that systemic interactions are by no means an exclusively property of political systems. All forms of social interaction produce system effects. But the thing that separates political systems from many other social systems is that it is exceedingly hard to look beneath their surface. The workings of the political system will in fact be unknown even to those on the inside, because there’s little reason to think that opposing players in the political game would reveal their motivations and strategies to that politician with any greater sincerity than they reveal them to the outside world.

That’s why political research, whether conducted by academics, lobbyists, think tanks or political parties, becomes informative only if it works by similar methods as historical research. It must be recognized that the available documentary evidence leaves the researcher ignorant of their own ignorance, as Veyne so poignantly writes in the quotation above. This is the big difference between the methods I described in Proven politics and the ones I’ve described here.

This is one of the reasons (there certainly are many more) why politics cannot be conducted by scientific methods. Any one person, whether they be outside or inside of the political system, will be profoundly ignorant of the actual workings of the system and even ignorant of their own ignorance. Consequently, no matter how well-informed the person is about society, they will have great difficulty in assessing the eventual outcomes and repercussions of their political actions to change it.

So due to system effects political actions never exhibit entirely predictable cause-effect relationships, but this does not rule out the possibility that political actors can acquire some form of “working experience” with the system. This practical experience is strongly subjective and thus different from the kind of political knowledge I’ve described in this essay, a kind which aims at some degree of intersubjectivity. But practical knowledge will be a topic for future essays.


(SE) Jervis, System Effects, Princeton University Press 1997.

(WH) Veyne, Writing History, Wesleyan University Press 1984.