Political Research Ethics, Transparency and Regulation

Everything in the earlier chapters happens inside rules, and those rules have been rewritten at speed. The European Union adopted a regulation on political advertising that requires a paper trail for every political ad, and the biggest platforms responded by leaving the market. American regulators confirmed that AI-generated voices fall under old robocall restrictions, and dozens of states passed laws on deepfakes in elections, some of which courts have already struck down. Betting markets on elections expanded fast in the United States, and a legal fight followed over who has the power to allow or forbid them.

The polling profession faced its own questions. Were pollsters copying each other's numbers to avoid standing out? Should they disclose more about their methods? Should countries keep banning the publication of polls in the final days of a campaign? At the same time, the data that researchers once used to study political advertising and foreign influence became harder to reach, partly because of the very transparency rules meant to open it up.

Why this matters to you

If you work in a campaign, these rules decide what you may do with data, advertising, AI and polling, and what the penalties are when you get it wrong. If you are a regulator or lawmaker, this chapter sets the evidence against the rules: where they have worked, where they have had consequences nobody intended, and where nobody has measured anything at all. If you are a researcher or a journalist, it explains why some of the records you relied on have gone dark and where to look instead. And if you are a citizen, it tells you what you are entitled to know about the political messages aimed at you.

What you'll find inside

The chapter opens with Europe's political advertising regulation and the platforms' exit, then follows the transparency paradox: the public archive of political ads went dark just as the law requiring transparency arrived. It covers the rules on deepfakes, AI disclosure and robocalls, and the dispute between federal regulators and states over election betting. Two pieces look at polling itself: the evidence on whether pollsters herd towards each other and what disclosure can and cannot reveal, and the countries that ban late polls. You will see what the evidence says about the reach and the effect of foreign influence operations, which are two very different things, and how data protection law treats information about someone's political views. The chapter closes with two framing pieces: the ethics of testing persuasive messages on real voters, and the oldest question in the field, whether polls are good for democracy.

The honest note

The characteristic caveat of this chapter is that the rules have run ahead of the evidence. Several new laws rest on harms nobody has measured, and several measured harms have no rule. Transparency mandates have, in places, removed data that researchers used to have. Enforcement is often thin: some of the most discussed rules have produced few or no decisions yet. And much of the evidence here is legal text, court rulings and official records rather than experiments, so it tells you what the rules say more reliably than what they do. The articles keep those two apart, and they report plainly when the effect of a rule is unknown.

The ten pieces in this chapter

• P6-01 · The TTPA Regime: Transparency by regulation, exit by platform
• P6-02 · The Transparency Paradox: The archive went dark when the law arrived
• P6-03 · Rules for Synthetic Politics: Deepfake laws, disclosure duties, robocall bans
• P6-04 · Who Gets to Price an Election: Prediction markets vs states
• P6-05 · Show Your Work: Poll disclosure and the herding question
• P6-06 · Poll Blackouts: Countries that ban late polls
• P6-07 · Influence Operations, Measured: Reach is not effect
• P6-08 · Political Data and Privacy: Sensitive by definition
• P6-09 · The Researcher in the War Room: Effective and ethical, in politics
• P6-10 · Do Polls Help Democracy?: The oldest argument, with new evidence