Representative Democracy is an Interface
Some reflections on my conversation with G. Elliott Morris, all while reading a book literally called Oligarchy.
Last week, I had the pleasure of getting to meet and talk to (and record for your enjoyment) a conversation with
. You can listen to the whole conversation here:We talked about a lot of things. I want to talk to you about one methodology thing, and two more high level things that address What Public Opinion Polls Are For.
The Methodology Thing
Elliott said something that I’m sure will rankle some — that he isn’t sure there is such a thing as a “gold standard” for polling methodology or mode anymore.
If you’re not familiar with the concept, the “gold standard” for polling methods encompasses a few things, the main components of which are:
A probability based sample — using address based sampling (ABS) or random digit dialing (RDD) to ensure that all people have an equal chance of being screened for a survey, rather than opt-in samplec.
Phone based/live interviews — actually talking to a person on the phone, rather than directing them to an online survey form, either via desktop or mobile.
Large sample sizes — big enough to apply weights without distorting the results, and provide cross tabs by key demographics
You can see just from this list where the trouble begins — people don’t answer the phone as much anymore, live interviews cost more the longer your interviewers are on the phone and the more people you have to call because people hang up. The expense of getting to 1000, 1200, or 1500 completes is extremely high — and that’s without paying people to participate, which also is counted as part of the gold standard. You don’t want people telling you their opinions just because they’re getting paid, the thinking goes. You want them to do so out of some sense of civic obligation.
And we have organizations that are doing their best to adhere to the gold standard — Pew Research, The New York Times, and other public pollsters invest a lot of time, effort and money in fielding surveys using these “best known methods”.
But in 2024, the gold standard underperformed pollsters who simply guessed that more Republicans would turn out — and they tend not to tell us much about their methods so who knows what they’re doing with weights, methods, modes, etc. (Elliott wrote about this at some length — a great reason to subscribe to
… I even have some gift links, I think, so hit me up and I’ll see what I can do.)I said during our conversation that I could be persuaded that it might still be worth it to use the gold standard when trying to predict electoral outcomes, but that it might not be as necessary when simply trying to understand public opinion. After talking to Elliott, I’m prepared to abandon the old gold standard altogether.
If I had to think through what the scorecard for a new gold standard would be, I think I’d include a few things that maybe haven’t been as emphasized:
Methods that overcome (even partially) partisan non-response, and general non-response. I’m interested in methods that can get even apolitical, unengaged people who might nevertheless vote to participate.
Methods that reduce respondent fatigue. Shorter surveys with more relevant questions that people actually want to answer, delivered in whichever format people are most comfortable with.
Online sample quality controls to help ensure the person completing the survey is who they say they are. I’ll once again refer you to my two-parter with Rich Ratcliff about sample quality and online survey fraud.
But to be honest, I think there are plenty of great tools available for much less expensive polling on public opinion, where we’re really trying to understand people’s relationship to politics and policy, and not predict electoral outcomes. And I agree with Elliott that there is room for more experimentation with methods in the name of gaining greater understanding of what people believe and want, and why.
What is Public Opinion Polling For?
Elliott put forward a point of view on the role of public opinion polling that others have also articulated — that it is important for all of us to know what our fellow citizens think and want, but it is possibly even more important for our elected representatives to know what we think and want. The theory here is that an informed, educated body of representatives would work the will of the people based on some evidence of what “the people” actually say.
Just as important — the people should have some influence on the kinds of questions that get asked. The questions should be reflective of people’s real ideas, values, beliefs and concerns — not the frames selected by the media, the politicians, and the lobbyists and interest groups and donors.
And you know, I agree. I think there’s shockingly little democracy in this democracy, and I think anything that provides a vehicle for democratic participation is good. Which is one of the things I like best about Strength in Numbers taking suggestions from readers about questions to ask — let the people decide what to ask the people.
But as I said in our conversation, this idea — and his proposal from his book Strength in Numbers: How Polls Work and Why We Need Them to have more poll-based public interest groups — both sounds good and … makes me feel like I’m going insane.
On paper, the Constitution provides the people with a representative: a Congressman. And yet, implicit in the need for public opinion polling is that this representation is a fiction. The representative does not directly engage most of his or her constituents. Increasingly, we see members of Congress abandoning the town hall in favor of the private fundraiser (like this one, that happened just up the street from me recently). Congressional offices don’t conduct many polls of their constituents that aren’t for the purposes of a campaign or a fundraising drive. And they rely heavily on surveys conducted by interest groups, who have the resources to fund their own research and polling, if not always the skill or intention of doing so accurately.1
When conducted by news organizations, polling is often intended to augment or stand in for traditional reporting. As we discussed in the most recent episode, newsrooms will pick the story first, then add a question to a poll, then report the results — which can be overly deterministic, or serving of an editorial or reportorial stance, rather than an open-ended fact-finding mission.
As a result, the sense I have is of polls serving as a buffer, an intermediary between policy makers and the people whose lives are affected by the policy. If ordinary citizens only have the representative-less town hall where they can talk to an empty chair, the online form letter, the voice mail, the postcard, and the opinion poll — then the representative is quite clearly representing someone else, some other set of constituents that have more of the representative’s attention.
The idea that ordinary citizens need an interest group to poll their opinions and represent them to their representatives is in some sense deeply perverse.
But of course we do not hold national public referenda on policy — we elect representatives to represent us, our interests, our communities, our families, our futures. How are we meant to communicate those interests? Of course, there is letter-writing, phone-calling, district office-visiting, town hall-attending, and DC-visiting to do.
But we have to compete with donors, interest groups, industry lobbying firms, the culture of the Congressional workplace, and whoever is President at the time for attention. They have whole staffs who do this work full time. Most ordinary people do not. So, very often, we lose on policy, even if we get the candidate we voted for. Those who do this for a living? They lose a lot less.
So then we have to wonder what the elections were for, exactly.
Could polls help us do more democracy?
I have started to think about our system as fundamentally a civil oligarchy with a representative democracy interface.
Representative democracy is, seen this way, the mechanism by which we, the People, interface with various elites. After all, elites own businesses, run the government, control the flow of capital and wealth and status; we, the People, provide labor, customers, taxes to fund government, and investment in their businesses. There has never been a government that long lasted without the consent of the governed, after all. So we use representative democracy as a consent interface that holds up the social contract via, chiefly, the rule of law.
But as anyone who has done usability research, or studied human computer interaction, or who studies systems and their affordances and entailments can tell you: interfaces are designed, and chiefly for the benefit of the party that designs the interface2.
Who designed the interface? We all know from grade school that when the Republic was first founded, only landed white men over the age of 21 had the right to vote. Even then, they chiefly voted for their Congressmen. The Senate was typically selected by governors or legislatures. The President, as we all know, but somehow forget every four years, is not directly elected via popular vote, but through the states via an Electoral College. The Supreme Court is nominated by the President, and confirmed by the Senate, and serves for life.
The founders, in their infinite wisdom, did not imagine themselves representing the will of the people in the sense of first finding out what the will of the people was, and then serving as zealous advocates for that will. Rather, they saw themselves as embodying the will of the people through their election or selection, custodians on behalf of a naive public, who should not be allowed to weigh in on all matters directly. They would act in our interests as benevolent interpreters of what we ought to want. They would trust that we would choose them wisely, so that they might govern with their own wisdom.
Sure, we do directly elect Senators now, yet many states endow governors3 with the power to appoint replacements when one dies in office or resigns or goes to prison. The Electoral College continues to privilege the representation of land over the representation of the people. And from Buckley v. Valeo (1976) to Citizens United v. Federal Elections Commission (2010), the majority of the Supreme Court has privileged cash above all else.
So if that’s the interface, what is the system?
The system is a modern form of civil oligarchy (forgive me, Elissa Slotkin). Civil oligarchy relies on the rule of law, institutions and governments to protect the property of, and settle disputes between, the wealthy. This way, oligarchs do not need to have their own armies, or engage in violence to protect their wealth. Instead, they vest a monopoly on violence in the state. In exchange for this protection of wealth/property, the oligarchs are willing to submit to some regulation, indulge civil rights protections for the non-wealthy, and to pay some amount of tax in order to fund the government and institutions that protect their wealth. But any taxation or regulation in excess of what they consider to be absolutely necessary — because it might cut into their wealth in a way they believe is excessive — will trigger their political engagement in the project of protecting their wealth.4
Once I encountered this concept, the system itself made more sense to me.
So did the interface.
We get to participate in the selection of the administrators of that government because we are incidentally protected, and because the governors need the consent of the governed for their own safety and success. But the system is not really designed for us. It’s designed to keep wealthy, propertied people from stealing each other’s stuff5.
And I’m not entirely certain that a polling-based interest group will improve this interface, nor that it will change the system.
Then again…
What if a true polling-based interest group, one genuinely invested in the project of representing people’s real values, goals, preferences, beliefs, ideas, etc., one divorced from the demands of a few wealthy donors, could conduct and publicize its research not only to elected officials and those who influence them, but could also hold up a mirror to The Public, allowing us to really see ourselves as we are, and not merely how we are framed for partisan or narrative purposes?
What if the evidence gathered by polling-based interest groups was there to help us all see how out of step the public narrative, and the parties, are with The Public? What if this was used to motivate, organize, and mobilize people to compete with wealthy and powerful policy-demanders? What if it was used to, if nothing else, try to true up what you think you’re voting for when you pick your team jersey, with what you’re actually voting for in terms of policy effects?
I suspect that the kinds of organizations that would be required in order for these public interest groups to amass the power, influence and resources necessary for such an effort would be member-driven and member-funded, organized, active, embodied as well as virtual, local as well as national, and steered not by a leadership built on team blue or red, but on genuine curiosity about and willingness to advocate for, the public and their interests, no matter what the money people say.
You know, these characteristics sound really familiar…
They sound kind of like a union6.
For an example of this, check out my conversation with Molly White about the dodgy polls conducted by Fairshake and other crypto interests — and her most recent reporting on their investment in recent crypto legislation and the way they’re already throwing their weight around ahead of the midterms.
For example, think about the LLMs you might be using. The chatbot is so friendly, so patient, so affirming. That window, the voice used by the LLM to organize its response to your query, all of it is designed to feel simple, conversational, trustworthy, supportive. It doesn’t matter what the underlying model and training corpus looks like, right?
My birth state of Oregon recently granted that power to the governor.
The social contract is, roughly, “if the wealthy create economic growth and subsidize a social safety net, the working class will not get so desperate they come for us”. The rule of law is used to protect that social contract, but its main purpose is to protect wealth (and prevent billionaires from becoming war lords, at least for awhile).
To be specific, it was designed to keep a monarch an ocean away from seizing the land and holdings of propertied men.
Or, you know, an independent newspaper.