Can protests lead to a better choice of government than elections with fraud?

In most countries the first-best mechanism of political choice, i.e. fair and competitive elections, is unfeasible. Thus, citizens are left to choose from either participating in rigged elections whose outcome is likely distorted in favor of incumbents, or resorting to protests to force a policy change or overthrow the government. I am wondering if there are circumstances in which mass action could lead to more desirable social outcomes than moderately unfair elections, and below I give a brief answer using main insights from mechanism design and social choice (all math omitted).

Choice functions

Most protests turn against presidents or governments headed by them. Thus, it is sufficient to analyze the properties of plurality voting, a social choice function used to select presidents. In plurality voting, every individual submits one vote for their (allegedly) most preferred candidate. The one who gets more votes than every other candidate, wins. A ‘corrupted’ version of this mechanism of interest that I shall call distorted plurality voting (DPV, for conciseness) assumes that the incumbent’s vote counts many times thereby replicating their ability to fix results.

Decision-making by protesting proceeds in several stages and takes a binary form. That is, at stage 1 individuals decide whether to protest or abstain, and if a sufficient number of people choose the former, the government falls, and people move to stage 2, where they decide whether to protest against its successor, and so on. Once there are not sufficiently many people to protest, the process stops and the procedure selects the current president/government. Therefore, protests can be modeled as a special version of successive voting where the incumbent comes first in the agenda while the remaining ordering is revealed by a stochastic process.

Since both DPV and protests can be modeled as social choice functions, we can compare them along several fundamental criteria: domain of voters, domain of candidates, truthfulness (‘strategy-proofness’), cycling and dictatorship. This comparison offers a formal argument for choosing a proper mechanism in the short-term, and this reasoning has to replace much romanticized pro-democratic or pro-revolutionary propaganda.

Domain of voters

Decision rules should not restrict the pool of preferences taken into account in collective decision-making.

DPV imposes quite weak restrictions related to age (at least 18), citizenship, sometimes other criteria as well. Additionally, costs of participation are low as polling stations have a pretty high density and casting a vote takes just a few minutes. In contrast, protests do not impose any formal restrictions as everyone can possibly take part, including minors and foreigners. However, costs of participation are not distributed equally across geographical and socio-economic parameters because uprisings typically occur in capitals and it takes weeks or months for them to succeed, so inhabitants of capitals have lower transportation costs and the unemployed bear lower opportunity costs. The bottom line is, DPV appears as a superior mechanism when it comes to the domain of voters as it equalizes opportunities for participation.

Domain of candidates

A good selection mechanism should not arbitrarily restrict the range of candidates for whom voters can submit their preferences.

DPV violates this requirement because the race is not fully competitive, by definition. Here, the incumbent has a say in who is allowed to run for presidency, but this influence can vary within a very wide range. Protests do not have this kind of incumbent-favouring bias, but they suffer from a natural limitation which arises from the fact that they do not usually last for more then 2-3 rounds. So, except for the incumbent, there are at most a couple of other candidates which individuals can consider for selection. In sum, under DPV voters choose from a wider set of candidates compared to protests, but these are likely to be the incumbent’s allies rather than actual competitors.

Strategy-proofness

A crucial requirement for a selection mechanism is truthfulness, i.e. the incentive for individuals to vote honestly.

Plurality voting is well known to be strategic, and this property spreads to DPV as well. Thus, some individuals could vote for candidates who are not their top choice if it leads to the winner they value more than the one they would get did they vote truthfully. Successive voting is truthful in a sense that individuals never vote for their least favourite candidate and always vote for their favourite one, but they still can support a candidate who is not their best choice at an earlier stage when they know that their top choice will not be selected later on. Although protesting is characterized by a probabilistic agenda and thus reasoning is essentially Bayesian, the room for strategic voting remains here, too, so both functions in question are not incentive-compatible.

Cycling

Here we just want that some candidate be actually chosen, so the process does not continue forever in cycles.

A good thing about plurality voting is that it can always tell the winner because there is always somebody with the largest number of votes. On the contrary, protests can go on for many periods and even run into cycles. Suppose that it is sufficient to have 100.000 people on the streets to overthrow the government, and there are 100.000 people who support the incumbent and 100.000 supporters of the opposition. Then, in period 1 the incumbent is ousted and the opposition reigns, in period 2 the opposition is ousted and the incumbent returns, in period 3 the incumbent is ousted again, and so on. To wrap up, DPV is superior to protesting due to its ability to generate a non-empty choice set, although the procedure itself is biased.

Non-Dictatorship

There should not be an individual whose most preferred candidate always wins the elections, regardless.

Neither standard plurality voting nor successive voting allow for dictators. However, DPV violates non-dictatorship when the incumbent is able to fix 50% of the votes, for obvious reasons. Is it realistic? Rather not, but DPV can still be thought of as a ‘somewhat dictatorial’ choice function. In contrast, dictatorship is certainly not true with regards to protests as they rely on spontaneous actions of thousands of people and the outcome can not be determined by the will of one person. This tells us something about the fact why it is so hard to predict if the protest succeeds compared to predicting the winner of the elections.

Conclusions

A sad fact about both mechanisms is their lack of ‘strategy proof’, which basically means that voters and protesters may intentionally misrepresent their preferences, so the property gives advantage to neither function. On the other hand, rigged elections always define some winner whereas protest could launch a cycle of coups where successors chaotically change one another. Also, unlike protests, presidential elections even with fraud equalize and minimize costs of participation in the selection mechanism. In addition, voters clearly know who are running for presidency whereas protesters cannot tell with certainty who comes to power once the government is ousted. The only good thing about protests seems to be their anti-dictatorial nature which perhaps makes it look such a nation-uniting enterprise to get involved in.

This analysis conveys a clear message for anyone caring about the social value of political choice. Although electoral fraud is an unfair and biased mechanism of elite selection, it still performs better than changing governments through protests in the short-term. Moreover, should we redirect our attention to monitoring electoral fraud instead of taking to the streets and chanting slogans, the society would be better-off in the long term as well because this could get us arbitrarily close to fair and competitive elections, a mechanism superior to both DPV and protests.