clay shentrup
4 min readNov 26, 2022


an ideal policy-making body has two fundamental characteristics.

  • aim
  • alignment

aim is like an archer’s ability to accurately relate the position, angle, and tension of his bow to the arrow’s point of impact. in policy terms, it is the ability to accurately assess the outcome of a particular policy.

alignment is the intent to hit the right target. in policy terms, it is the policy-making body’s intent to pursue policy that is in the best interest of society.

aim can be subdivided into knowledge and analytical aptitude.

  • knowledge entails access to information.
  • analytical aptitude entails the cognitive ability to rationally process information.

knowledge matters because a genius can’t craft good policy with inaccurate and/or incomplete information. and analytical aptitude matters because an idiot can’t craft good policy even with the best information. (just consider that washington voters voted down a carbon tax via ballot initiative twice in the past decade, and you can see that voters often don’t know what’s in their own best interest.) the devil’s advocate might propose that a mediocre intellect can simply consult with experts, but it takes intelligence even to accurately judge the competence of disagreeing experts.

alignment can be subdivided into representativeness and benevolence. a body that is statistically representative of the overall populace will tend to choose the best policy. and even an unrepresentative (“biased”) body will choose the best policy if it is altruistic (“benevolent”). what doesn’t work is an unrepresentative body that is also selfish.

a statistically demographically representative body will be aligned (choose what’s in its own best interest), regardless of whether it’s selfish or altruistic. whereas a biased body will only be aligned if it is altruistic. a misaligned and selfish body will go against the public good by definition.

so, putting this all together, we have:

  • aim
    - knowledge
    - analytical aptitude
  • alignment
    - representativeness
    - altruism

the dominant paradigm of electoral democracy is decent at selecting representatives who have aim, but not so great with alignment. most politicians who are savvy enough to get themselves elected are capable of assessing the outcomes of policy. while charismatic lunatics are not unheard of, the vast majority of the USA’s congress people are highly educated elites. but these politicians can be misaligned for a variety of reasons. this was summarized as follows in a recent article.

Elections…reward well-positioned insiders who have the connections and war chest to wage a campaign. They also attract ambitious social climbers. Today even the most virtuous candidates have to solicit truckloads of money — anywhere from $500,000 to $2 million for a credible run. Once in office, winners spend much of their time raising revenue for reelection. The amount of actual legislating — investigations of issues, research into policy, seeking the common good — is small, and legions of lobbyists exercise influence to the tune of $3.5 billion a year.

the underlying causes of this misalignment also extend to suboptimal voting methods (e.g. plurality voting as opposed to score voting or approval voting), gerrymandering, and turnout biases. moreover, the lack of aim among the voters themselves can translate to lack of alignment amongst elected officials. for instance, a charismatic authoritarian like trump or bolsonaro can rise to power by exploiting voters’ emotions, causing them to vote for politicians whose policies directly contradict their own best interests.

one ancient idea making a comeback is sortition, whereby randomly selected citizens take the place of elected officials. the idea is that “everyday people” will exceed at alignment. leaving aside any questions around aim for the moment, let us consider whether this notion of alignment has merit. random selection certainly holds the potential for alignment, provided the office is compulsory. but if it isn’t, then the legislature won’t be representative, because the random selected citizens who opt out of serving will be statistically demographically different than those who accept it—much like the aforementioned election turnout bias.

the general response to this opt-out bias is the invocation of complex algorithms which attempt to correct for it. but not only are these algorithms complex, they can only balance for the demographic metrics we choose to measure. such metrics might include age, gender, sexual orientation, religion, political party, etc. not only is the selection of the metrics itself biased, but there’s no way to accurately measure the “christian-ness” or “republican-ness” of a human being. imagine a decision made by AOC and mitt romney vs one made by joe manchin and rand paul. both duos feature a republican and a democrat, yet they are worlds apart on ideology.

and the aim of a randomly selected body is a matter of some concern as well; we might like to have representatives who are considerably smarter than average and/or have experience at making policy. i would certainly place more faith in a candidate who’s read seeing like a state and/or has deep experience studying the logistical aspects of policy in the real world.

how could we preserve the true randomness of our selection process? only by making it compulsory. we have precedent for this in juries. jurors don’t get to opt out. but this is only feasible because juries convene for a limited time, and then go back to their lives. our sortition body could operate similarly if used to elect a body of long-term officials, effectively transcribing its ephemeral ideological position into a long-term persistent record. while this might slightly decrease the representativeness (because we don’t know people’s real ideological leanings merely by interrogating them), it would tend to increase knowledge and analytical aptitude; because these qualities are much easier to objectively assess and quantify, especially when provided a deliberative context.

i admit that my analysis is speculative, and i think we would benefit from extensive experimentation with a variety of forms of sortition. i think the most important thing is for it to be legally binding rather than merely advisory. i suspect the most politically viable path here is a hybrid approach whereby existing city councils are augmented with a few sortition-based seats. the united states is a particularly ideal candidate for this pathway, given its highly unusual ballot initiative process, whereby novel electoral reforms may be enacted down to the level of granularity of a single city.



clay shentrup

advocate of score voting and approval voting. software engineer.