Does “Social Choice” Theory Support Heightened Judicial Review?

William Riker and Barry Weingast argue in an article in the Virginia Law Review that social choice theory’s “chaos theorem” regarding majority decision-making augurs for heightened judicial review of legislative enactments. I’m skeptical. Not because majoritarian instability is a good thing. Rather, the categories from social choice theory of the “stability” or “instability” of majoritarian outcomes do not map consistently either onto legislation being “constitutional” or legislation being “unconstitutional.” So the concepts don’t help with thinking about judicial review.

“Social choice” relates to decision making within groups of people. Hence “social” because there’s more than one person making the decision. The “chaos” results of social choice theory relates to the fundamental insight that majoritarian decision making need not be rational. As I’ve discussed before (here and here, among other places) by arguing that majoritarian outcomes are “irrational” social choice theorists do not mean substantively foolish majoritarian outcomes. Rather, they mean majoritarian outcomes cannot be guaranteed to align with rational choice theory’s rationality postulates.

These are well known, but to avoid the need to click on the links above, “rationality” in social choice theory (which is a subset of rational choice theory) means that an actor [a] can rank all the choices the actor can make – A is preferred to B, B is preferred to A, or A is indifferent to B – and [b] the actor’s choices are transitive. So if one prefers A to B, and B to C, then one also prefers A to C.

The well-known problem with majoritarian decision making, however, is that it can be intransitive. The irony is this: a group of entirely rational individuals making decisions together by majority rule can make socially irrational choices. In speaking of “irrational” social choices of majority rule it’s important to keep in mind the narrow definition of what an irrational outcome is – basically that outcomes chosen by the group are intransitive.

Majoritarian “voting cycles” result from majoritarian intransitivity. Again, to avoid the need to click on a link, consider the simplest example. We have a three-person legislature voting over three policies. The legislators’ policy preferences are configured this way: Legislator 1 prefers policy A to policy B to policy C. Legislator 2 prefers policy B to policy C to Policy A. And legislator 3 prefers C to A to B.

In sincere, pairwise voting between the policies, it is straightforward to see that, depending on which policies get voted on first, policy A can be a majority-supported outcome. And policy B can be the majority-supported outcome. And policy C could be the majority-supported outcome. Any of the policies could be supported as majority outcome. Basically, then, whether policy A, B or C is actually adopted by the legislature is a result of factors other than majority preferences. What matters is who, or what, determines the sequence in which the policies are voted on.

It is this that worries Riker and Weingast. They point out that the person who determines the sequence of votes – for example, the majority leader of the legislative chamber – chooses the outcome. So, really, only that one person’s policy preferences determine the policy outcome, not majority preferences.

So, according to Riker and Weingast, actual legislative outcomes derive from non-majoritarian arbitrariness. This arbitrariness, for them, justifies heightened judicial review of legislation.

That said, one could spin it the other direction entirely: Majoritarian indeterminacy actually presents us with an embarrassment of riches in that whatever normative suasion majority rule provides can be satisfied consistently with any of the outcomes. It’s not as though a minority were imposing its will on the majority in any of the outcomes, after all.

Indeed, it is something of a misnomer to refer to “the” majority in this context. There are many majorities. Why would it be problematic that more than one policy crosses the majoritarian threshold?

Secondly, however, “stable” or “determinate” majority outcomes aren’t necessarily good things either. Consider, say, the Alabama legislature in the 1930s. Given the configuration of race-based preferences in the legislature, there was no question that racially-oppressive legislation was also “stable” and “determinate.” There was no possibility of cycling no matter who the agenda-setter was. Yet the substantive outcome would be unconstitutional.

One might be tempted to turn the tables and argue that majoritarian indeterminacy is a good thing in that bad legislation could be overturned by differently-configured majorities in the legislature. And that’s certainly true. But it is no more true than that good legislation could equally be overturned by differently-configured majorities in the legislature in favor of a bad alternative.

The standard of review with which judges review legislation has been a hot-button issue for over a century. It’s not a question of whether unconstitutional legislation should be permitted or not, it’s a question of who or what gets the benefit of the doubt when judging whether a piece of legislation is constitutional or not. Important results from social choice theory regarding majoritarian indeterminacy doesn’t really help with those determinations.

Reader Discussion

Law & Liberty welcomes civil and lively discussion of its articles. Abusive comments will not be tolerated. We reserve the right to delete comments - or ban users - without notification or explanation.

on April 08, 2019 at 10:17:17 am

I'm surprised Rogers didn't tie this piece in to his piece last week on the misnomer "political science," as "social choice" is the principal pseudo-science making up the subfield. They seem tailor-made for a tie-in.

In the first 2/3 of this essay Rogers uses the words "rational" and "irrational" and only near the end does he use the words "good" and "bad." Since political thinkers from Socrates onward have thought of politics as a means to the good life and not to the rational life, it would seem that shifting the study of politics from the good to the rational was done solely for the purpose of mathematizing politics in order that persons of analytical bent but otherwise not gifted thinkers could participate in the career opportunities made available by this practical doubling in size of the Dept. of Economics.

As Rogers demonstrates, any of the 3 policy choices may win a majority vote and thus each is normatively justifiable on that basis. Yet it may not satisfy either the narrow poli-sci definition of rationality or the philosophical criteria for the good. So now we have 3 normative standards: the majority, the rational and the good. And in a sort of meta-chaos theorem, we can imagine 3 political science professors, each of whom ranks these three preferences differently, so that any of the three standards is the "right" standard. A circle spinning inside a spinning circle. A/k/a vertigo.

Rogers implicitly subordinates the rational to the good, making it unclear why anyone would devote a professional lifetime to analyses of the rational in politics, other than for an occupation.

read full comment
Image of QET
on April 08, 2019 at 10:59:50 am

Left unsaid is the assumption that the 3 legislators represent 3 equal segments of the polity.

How do "majoritarian" views get enacted if #1 represents 60% of the polity, #2-35% and #3 - 5% , and #2 and #3 constantly vote down #1?

In senate arrangements, this is a valid outcome...

read full comment
Image of OH Anarcho-Capitalist
OH Anarcho-Capitalist
on April 08, 2019 at 12:47:30 pm

Yep! Absotively!

Here we observe the perennial attempt to obscure (diminish?) concepts of good / evil with a purportedly neutral framework that is most significantly (putatively) reducible two mathematics; or SCIENCE by another name.

Has Roger's memory begun to fail as his last essay specifically questioned the basis for such legerdemain.

Even the, the terms as employed by the Political *SCIENTISTS* do not meet the basic requirement of effective speech. In these rareified venues, "rational" is conflated with a mathematical equation and may very well be completely "ir-rational" in the common understanding.

Now it may be done, as QET asserts, for "occupational" reasons; BUT I suspect it simply [provides evidence of that special form of human hubris which arrogates to itself the ability to a) reduce the complex / irresolvable to a mathematical understanding AND b) thus allowing the hubristic among us to SOLVE such long standing intractable problems.

Bully for them. You may have your math.
I remain content with my Oakeshottian *collisions*.

read full comment
Image of gabe
on April 08, 2019 at 15:28:12 pm

The returns to a literature that was rather extensive in the 1990s, and to which I personally contributed. Along with Professor Rogers, I argued that social choice does not provide an independent justification for heightened judicial view. My reasoning was a bit different. I explained that one must identify the differing social choice profiles of the relevant institutions, here legislatures and courts, both of which manifest dynamics associated with cycling member preferences, and with Arrow's Theorem, but that they do so in different ways. In general, legislatures tend to relax what Arrow identified as Independence of Irrelevant Alternatives (IIA), yet adhere to Range, with the result that legislators may register intensities of preference (or to commodify) rather than being limited to strict ordinal rankings, and that legislatures may remain inert in the event that for any given option, another is preferred by simple majority rule, sometimes defaulting to the status quo. By contrast, appellate courts tend to relax Range, yet adhere to IIA, thereby encouraging judicial outcomes even when there is an embedded judicial cycle, as sometimes occurs, and courts further rest that outcome on the position of the median jurist under the narrowest grounds doctrine absent both a majority opinion and a judicial cycle. Professor Rogers is correct that majoritarian preferences, like Condorcet preferences (change Rogers's Legislator 3's preferences from CAB to CBA, and B, the Condorcet winner, will then defeat A and C), bears no correlation to normative merit of the policy under review, although it's worth mentioning that many southern racist policies were sustained well past the 1930. In the end, when assessing the social choice implications for legal policy, it is important to bear in mind that all institutions have social choice dynamics, but that those dynamics manifest themselves differently depending on how the relevant institutions are structured. Understanding these differences is important in assessing social choice based policy proposals such as the one advanced in the VLR article.

read full comment
Image of Max Stearns
Max Stearns
on April 08, 2019 at 16:54:06 pm

It is also possible that majorities change when people see the effects of the laws they put into place.
Before prohibition, there really was a majority that supported prohibition. Half-way through prohibition, there really was a majority that supported getting rid of prohibition.
The same goes for sex-control laws and gun-control laws that fail to have any effect or become impossible to enforce without violating the fourth amendment.
Or speech-control laws like the fairness doctrine--which both sound good and are possible to enforce--yet turn the american people into mindless sheep. Who'd've guessed that simply not having a Rush Limbaugh on the radio would've made the sixties and seventies so much worse than today?

read full comment
Image of Glock Cerac
Glock Cerac

Law & Liberty welcomes civil and lively discussion of its articles. Abusive comments will not be tolerated. We reserve the right to delete comments - or ban users - without notification or explanation.