Opinions

An opinion is what you think about one specific thing. "I support this exact bus fare change." "I oppose this exact zoning rule." It is different from a priority, which says what matters most, and different from a vote, which is the official decision.

Opinions are tricky. The same person can answer differently if a question is worded differently. People sometimes say what they think others want to hear. And people change their minds, often for good reasons. In My Vote Forward, an opinion is private unless you choose to share it. You choose what to share, one item at a time. You can revise it later.

This chapter explains why.

Defined terms

Opinion. A stated position, and optionally its strength, on a specific, worded question or proposal at a specific time.

Item wording. The exact text a person responded to, including any description, cost estimate, or context shown with it.

Public expression. An opinion the person has chosen to attach to their name or profile so others can see it.

Private expression. An opinion recorded for the person's own use and, if they consent, counted only inside group results.

Specific disclosure. Sharing one identified opinion on one identified item, as opposed to a blanket setting that makes all opinions public.

Social desirability bias. The tendency to give answers that seem more acceptable to others than one's true view.

Revision. A later change to a recorded opinion, kept with its date.

An opinion belongs to its wording

Wording effects are among the best documented findings in survey research. In January 2003 the Pew Research Center asked whether people favored "taking military action in Iraq to end Saddam Hussein's rule": 68% favored and 25% opposed. When the question added "even if it meant that U.S. forces might suffer thousands of casualties," 43% favored and 48% opposed ¹. Nothing about the public changed between those two versions. Only the words did.

Order matters too. In an October 2003 Pew survey, 45% favored legal agreements for same-sex couples when that question came after a question on same-sex marriage, compared with 37% without that context ¹. The earlier question changed how the later one was read.

Some questions cannot be answered cleanly at all. A double-barreled question asks about two things at once; Pew's example is asking for confidence in a president "to handle domestic and foreign policy," and its advice is to split such items into separate questions ¹. Agree/disagree formats add another distortion: Pew reports that some respondents tend to agree with statements regardless of content, and that this acquiescence is stronger when an interviewer is present ¹.

The framework draws a plain conclusion. An opinion is not a free-floating attitude. It is an answer to specific words. So:

• Every recorded opinion is stored with the exact item wording and version.
• If the official wording changes, earlier opinions are marked as answering an earlier version. They are not silently carried over.
• Group results report the wording, order, and response options used.

Detailed guidance on writing fair items is in Question design.

Public and private expression

People express opinions in public all the time: at meetings, in letters, in conversation. That is part of democratic life and the framework does not discourage it. But a recorded, searchable opinion attached to a name is different from a remark at a meeting. It can be collected, compared, and used.

The secret ballot exists because public choices invite pressure. The Australian Electoral Commission notes that before the 1850s, "people voted publicly, which left them vulnerable to intimidation and coercion" ². Opinions are not votes, and the framework does not make them secret by force. But the same pressures that once worked on public voting can work on public opinion records: an employer, a landlord, a family member, or an organized group can reward or punish what someone says. See Intimidation.

The framework therefore treats opinions as private by default and public by choice.

Hypothetical example. A teacher supports a local school funding measure and wants to say so publicly. She shares that one opinion, on that one measure. Her opinions on a transit plan and a housing rule stay private. No one can see that she has opinions on those items, let alone what they are.

Why sharing is voluntary and specific

A single "make my profile public" switch would be simpler. The framework rejects it for three reasons.

1. Consent is about a specific use. Agreeing to share one view on one subject is not agreeing to share all views on all subjects. See Consent.
2. Pressure is targeted. The harm from disclosure depends on the subject and the audience. A person may be safe sharing a view on parks and unsafe sharing a view on policing.
3. Profiles become products. A complete public opinion record for each person is exactly what campaign targeting and data brokers want. The framework does not build one and does not sell data.

Specific disclosure also means a person can see exactly what is public about them at any time, and can stop future sharing. Copies already seen or saved by others cannot be recalled, and the framework says so plainly rather than promising otherwise. See Privacy.

Social desirability

Even in private, people do not always report what they think. Pew lists social desirability among the main sources of error in surveys: respondents tend to understate behaviors such as alcohol and drug use, tax evasion, and racial bias, and to overstate church attendance, charitable giving, and likelihood of voting ¹. Pew's own questions about voting are written to make not voting easy to admit; one version first notes that circumstances may have prevented a vote and then asks "did things come up that kept you from voting, or did you happen to vote?" ¹.

Two implications follow for the framework.

• Privacy improves honesty. If people know an opinion is private by default, they have less reason to perform for an audience. This is a hypothesis, not a measured result for this framework. See Evaluation.
• Public opinions are a different data set. Opinions people choose to share are likely to differ, on average, from those they keep private. Public opinions should never be treated as a sample of all opinions.

Revision is normal

People change their minds. They learn the cost of a proposal, hear an argument, or see how a plan affects someone they know. A framework that treated a change of view as inconsistency would punish exactly the behavior that informed participation is meant to encourage.

There is research evidence that informed discussion changes views. Deliberative Polling, originated by James Fishkin at Stanford in 1988, polls a random sample, gives participants balanced briefing materials and discussion with competing experts, and then asks the same questions again. The Stanford Center for Deliberative Democracy reports statistically significant changes in views after deliberation; in one South Korean example in 2011, agreement that unification was "unnecessary" moved from 72% to 91% ³. Those are reported results from a specific method, not a prediction for this framework. They show that opinion change after new information is common and measurable.

In the framework:

• Opinions can be revised at any time before the item closes.
• Each revision is stored with its date. The person can see their own history.
• Group results use each person's latest opinion as of the release date, and the method is stated.
• A changed opinion is never flagged publicly as a reversal.

Representatives change positions too. How that should be explained is covered in Representation.

Evidence

• Research results: wording, order, double-barreled items, and acquiescence all change survey answers ¹.
• Research result: social desirability distorts self-reports on sensitive topics, including voting ¹.
• History: public voting exposed voters to intimidation and coercion ². See The secret ballot.
• Reported research practice: structured deliberation is associated with measurable opinion change ³.
• Design proposal: voluntary, specific disclosure reduces pressure and improves the quality of what is shared. This is untested.

Competing interpretations

"Opinions should be public to be accountable." One view holds that a democracy needs people to stand behind their views, and that private opinions encourage irresponsibility. The framework answers that public expression is fully available; what it refuses is a default that exposes people who did not choose exposure.

"Opinions are too unstable to be worth recording." A skeptical view holds that many survey answers reflect the question more than any stable attitude, given how much wording alone can move them. If so, recording opinions may add noise. The framework responds by tying each opinion to exact wording and by not treating opinions as decisions. Whether recorded opinions carry enough signal to be useful is an open empirical question.

"Allowing revision invites manipulation." If opinions can change, coordinated campaigns could flip them near a release date. The framework treats release timing, cutoffs, and detection of coordinated behavior as governance problems. See Collective priorities and Published releases.

Application to this framework

• Every opinion is stored with its exact item wording and version.
• Opinions are private by default. Sharing is opt-in, per item, and visible to the person.
• No blanket "public profile" setting exists.
• Revisions are allowed and dated. Group results use a stated cutoff rule.
• Public opinions are never presented as representative of all opinions.
• An opinion is never treated as a vote, and no attempt is made to link a named person's opinion to their ballot. See Opinion and choice.

Unresolved questions

• Does private-by-default actually reduce social desirability bias in this setting? Untested.
• What is the right cutoff for counting revisions before a published release?
• Should strength of opinion be recorded, and if so on what scale? See Question design.
• How should opinions on an earlier version of a proposal be shown once the wording changes?
• How can coordinated, inauthentic opinion campaigns be detected without surveillance of individuals?