The Right to Disagree

On 31 October 2024, Lucy Connolly, a mother of two and a childminder from Northampton in England, was sentenced to thirty-one months in prison for a single tweet. She had posted it in the hours after three girls were murdered at a Taylor Swift-themed dance class in the northern English town of Southport. The tweet called for mass deportation and for arson attacks on hotels being used to house asylum seekers. It was visible for about four hours before she deleted it. By then it had been read more than three hundred and ten thousand times.

The tweet was ugly. It was arguably the kind of speech that reasonable people could argue ought to be caught by existing laws against incitement to violence – laws Britain has had in some form for over a century. What is striking is not that Connolly was prosecuted. It is that she received a thirty-one-month sentence, roughly the weight of some sentences for armed robbery, for a tweet she had already deleted.

The law under which she was jailed was not Britain's Online Safety Act 2023, the new statute that drew most of the international press. It was a thirty-eight-year-old public-order law, applied to a social-media post. And the philosophical cover for sentences like hers, cited more often than any actual British statute, is a footnote of about one hundred and eighty-five words that Karl Popper wrote in 1945.

I. What Popper actually wrote

Popper wrote the footnote in exile in New Zealand, at the end of the Second World War. It sits in the first volume of The Open Society and Its Enemies, attached to Chapter 7. In it, he described what he called the paradox of tolerance: that a society tolerating everything might appear to license the intolerant to destroy it.

In the version that now travels – a 2017 cartoon by the studio Pictoline, republished widely after the Charlottesville rally, and an infographic that has been on social media ever since – the paradox becomes a simple warrant: intolerance is not tolerated.

Popper's actual argument was more careful, and in important respects the opposite. He wrote that so long as tolerant societies could "counter them by rational argument and keep them in check by public opinion", the suppression of intolerant speech "would certainly be most unwise". Suppression was reserved for movements that "answer arguments by the use of their fists or pistols". The instrument he proposed, where suppression was justified, was the criminal law on incitement, comparable to incitement to murder. Not civil penalty. Not political exclusion.

Popper was arguing for a limit. He was not arguing for a warrant. He spent two volumes warning against concentrations of state and ideological power. He was the last person alive to hand governments a roving licence on what counts as intolerance.

II. The United Kingdom now

The United Kingdom shows what happens when the simplified version becomes administrative practice. In 2023, according to a Freedom of Information request published by The Times, British police arrested 12,183 people for online communications offences – about thirty-three people, every day. Of those, 1,119 were sentenced. The mismatch between arrests and convictions is itself the chilling effect. The law has become a tool of warning more than of conviction.

Six years earlier the number had been less than half. In 2017, UK police made 5,502 such arrests in the whole year. By 2023, the figure had more than doubled – a 121 per cent rise in six years. The direction is steep, and it is towards a kind of state management of speech that Britain would not, until very recently, have recognised in itself.

The prosecutions made in the wake of the Southport murders are concrete illustrations. Connolly's thirty-one months. Tyler Kay's thirty-eight. Jordan Parlour's twenty. The sentences are of a weight usually reserved for physical crimes, handed down for tweets. The laws relied upon are not new: most arrests draw on old British public-order and communications statutes from the 1980s and early 2000s, written before Twitter existed and stretched to cover it. Which is, in a way, the essay's later point – when existing laws already reach speech that incites violence, no new statute is needed. What has changed in Britain is not the law. It is the willingness to use it.

Across the Channel the same pattern. Lower Saxony alone prosecutes about 3,500 hate-speech cases a year. France tried a version of the British approach in 2020, the Loi Avia, and the Conseil constitutionnel struck it down on proportionality grounds. Britain has not yet had its own proportionality review. That is the problem.

III. The Australian drift

The same instinct has appeared at home. The New South Wales Premier, Chris Minns, defending his government's expansion of hate-speech law after a string of antisemitic attacks in early 2025, put it this way: "We don't have the same freedom of speech laws that they have in the United States, and the reason for that is that we want to hold together a multicultural community and have people live in peace, free from the kind of vilification and hatred." His government then expanded section 93Z of the NSW Crimes Act 1900 to cover specific slogans and symbols.

On its face this is a decent and humane position. Who could object to community peace? Who wants to be the person defending hatred?

That is how every slide into state control over speech begins. Chris Minns is not malicious. He is not a would-be tyrant. He is, without quite noticing, making the same argument quieter regimes have made before him: that the state is better placed than its citizens to decide which speech is compatible with peace, and which is not. Once a government gives itself that power, the next one keeps it. The ratchet turns one way only.

Minns is not alone. Victoria's Justice Legislation Amendment (Anti-vilification and Social Cohesion) Act 2025 created a three-year incitement offence and a five-year threat offence, with a "genuine political purpose" defence whose scope no one has yet tested. The federal Misinformation and Disinformation Bill 2024 was withdrawn in November 2024 after a Senate committee majority recommended its discharge. Credit where credit's due: on that occasion Canberra listened. But the instinct behind the bill, that government-authorised content would be exempt from its own misinformation test, will not disappear. It will return under another minister and a different name.

Premier after premier, bill after bill, parliamentary committee after parliamentary committee, now makes a version of the same argument in different accents. Australia is drifting, and the drift is polite, well-intentioned, and comprehensively mistaken.

The ground on which the drift is happening is shallow. Australia has never had a First Amendment. Our free-expression protections sit in the common law and in a narrow implied freedom of political communication that the High Court read into the Constitution in 1992. Both are easier to erode than to build. Both are eroding.

Section 18C of the Racial Discrimination Act 1975 makes it unlawful to "offend, insult, humiliate or intimidate" a person or group by reason of race, colour, or ethnic origin. Section 18D protects artistic work, scientific debate, and fair comment on matters of public interest. The balance is meant to hold between abuse and argument. It has not always held. Eatock v Bolt (2011) found against Andrew Bolt over two newspaper columns. Faruqi v Hanson (2024) required Senator Pauline Hanson to delete a tweet directed at another senator. The Federal Court rejected Hanson's implied-freedom challenge.

The thing worth defending in Faruqi v Hanson is bigger than Hanson's tweets, and more important than them. It is the principle that elected politicians should be free to say sharp, even offensive things about one another's politics without a court telling them to delete. That is exactly the speech the implied freedom exists to protect. The right answer to Hanson's tweet is Faruqi's rebuttal, in the parliament and in public. A court order to delete is an answer the law should be very slow to give. It was too quick here.

Allegra Spender, the independent Member for Wentworth, tabled her vilification amendment in February 2025, ten months before the attack at Bondi in December. Her concern was real. The rise in antisemitic incidents in Sydney and Melbourne through late 2024 was serious and well documented. But the amendment was aimed at speech, and the problem at Bondi was violence. No vilification law stops a terrorist with a weapon. The amendment could not have prevented Bondi, and it will not prevent the next attack of that kind. What it does is what symbolic legislation always does: it lets an elected representative appear to be acting on a problem her law does not reach.

After Bondi, legislation that had struggled in early 2025 acquired new political momentum. Grief did the work of debate, and a bill drafted in one environment was carried, after the fact, as the answer to a different one. The tool for catching speech sharpened. The tool for catching violence was not reconsidered.

The distinction worth drawing, and defending, is simple. Speech that incites violence is punishable, and Australian law has caught such speech for a long time. Speech that expresses opinion, including ugly opinion, should remain legal. The role of government is to hold that line narrowly, prosecute the first, and leave the second to the immune system of public argument.

IV. Why the harm test begs the question

Every expansion of speech law since the mid-1990s has been argued on the basis of harm. It is a serious case. It deserves a serious answer.

The claim runs that offensive speech causes real harm: that it silences people, wounds them, produces psychological injury. There is something to the argument at the edges. Direct threats and sustained harassment campaigns do produce measurable effects, and the criminal law has recognised this for a long time. But the claim should not be accepted for the wide middle ground of political and religious speech, where the loudest calls for new law tend to focus.

Christopher Hitchens used to say that taking offence is a choice. Douglas Murray has made a similar argument in more recent form. Offence is something the listener decides to feel, having heard the speech. The speech is speech, and the listener is an adult. To be offended is not the same as to be injured. Collapsing the two lets the most easily offended listener set the limits of public debate. That is not a neutral legal threshold. It is a veto.

Once harm is accepted as the test, the test itself becomes unstable. The Victorian 2025 Act turns its criminal threshold on "revulsion" and "severe ridicule". A law whose trigger is revulsion is unpredictable in application and wide in its chilling effect.

Imagine a Sydney bus driver posting on Facebook a sharp joke about a minister's speech at a religious gathering. Some readers find the joke disrespectful. Is it "severe ridicule"? A jury on one day might say yes. A jury on another day might say no. The law is the same. The outcome is a coin toss. The bus driver learns the lesson, and the post never gets made. That is the chilling effect. The argument never happens. The debate that might have changed someone's mind never takes place. The country is quieter, but it is not any wiser.

Isaiah Berlin drew a famous distinction in 1958. Negative liberty is freedom from coercion, the right to be left alone. Positive liberty is the state creating conditions for a citizen to flourish, a larger and more active job. When speech laws mix the two up, the state gives itself the power to manage how citizens feel by managing what can be said to them. That is a much bigger power than Australian governments have usually claimed. It is the power the current drift is handing them, by inches.

Aleksandr Solzhenitsyn, in his 1978 Harvard commencement address, named this mechanism in a single sentence. "Without any censorship," he told the audience, "in the West fashionable trends of thought and ideas are carefully separated from those which are not fashionable." His point was that conformity does not require a censor. Polite opinion, media convention, and social pressure produce it on their own – the fashionable ideas occupy the centre, and the unfashionable ones are quietly edged to the margins, without anyone having to ban anything.

What a harm-based speech law does is take that cultural separation and give it the force of law. What was the difference between saying something in polite company and saying it in the pub becomes the difference between saying it and being charged. The filter is no longer social. It is enforceable.

V. The immune system is a better metaphor

An immune system is a better metaphor for an open society than a walled garden. The defining feature of an immune system is that it learns by contact. It cannot tell self from foreign without exposure.

Societies that argue with their extremists develop antibodies. Societies that curate them out of view do not. You cannot develop antibodies to an argument you have never heard. A society that closes its ears to hard ideas is not safer. It is weaker.

The evidence sits in the arguments we have already had.

In 2009, the BBC put Nick Griffin on Question Time, its flagship political programme. Griffin was the leader of the British National Party, at the time Britain's main far-right party, which had just won its first two seats in the European Parliament. The BBC's decision to give him the platform was protested loudly. Some argued that letting him speak on national television would legitimise him. What actually happened was the opposite. In front of an audience of eight million, Griffin was pressed on his record of Holocaust denial, on his views on mixed-race marriage, on Islam. He was evasive. He contradicted himself. He looked small. The BNP's polling collapsed within months, and the party has never been a serious electoral force since.

In January 2023, Konstantin Kisin spoke for the affirmative in an Oxford Union debate on whether "woke culture" had gone too far. The motion carried, 89 votes to 60. Kisin won by addressing his opponents directly, treating the debate as a conversation rather than a performance, and offering a positive counter-proposal. The speech has been viewed over a hundred million times.

In each case, the society met the argument in the open and carried it. That is what an immune system does. What it cannot do is operate in the absence of exposure. Keep the pathogen out of sight and the body never learns.

VI. What an open society looks like under strain

The test is what happens when the society is genuinely tested.

Take Ridd v James Cook University, decided by the High Court in 2021. Peter Ridd, a marine physicist at the university, had publicly criticised the quality of some Great Barrier Reef research, and had not been polite about it. James Cook sacked him for "uncollegial conduct". Ridd fought the sacking all the way to the High Court. The Court upheld his dismissal, narrowly: it found the specific contract clauses the university had used were valid. But in the same judgment it said something much larger. Australian academic freedom, the Court held, includes the right to criticise colleagues' work even in "discourteous or intemperate" language – because that is what intellectual disagreement sometimes requires. Blunt speech in a university is not a bug. It is what a university is for.

In Lattouf v Australian Broadcasting Corporation [2025] FCA 669, the Federal Court found that the ABC had unlawfully terminated the employment of Antoinette Lattouf for reposting a Human Rights Watch video about Gaza. Compensation was $70,000. The penalty was $150,000. The full cost to the broadcaster was over $2 million.

These are not comfortable cases. They are institutions working, slowly, expensively, painfully.

Václav Havel described the other half of the picture in his 1978 essay The Power of the Powerless, written under communist rule in Czechoslovakia. A greengrocer puts a sign in his shop window: "Workers of the world, unite!" He does not believe it. No one has strictly ordered him to display it. But it is expected, and everyone – the greengrocer included – knows the regime would read the sign's absence as a political statement. So the sign stays.

Havel's insight was that a regime of that kind is not held up primarily by its own force. It is held up by millions of small acts of submission: the sign in the window, the slogan repeated without belief, the uncomfortable opinion left unsaid. The greengrocer simply declining to put the sign up is a small act. Multiply it enough times, Havel wrote, and what he called "living within the lie" begins to crack.

An open society is not preserved by its courts and tribunals alone. It is preserved by its greengrocers. By citizens who, in private and in public, decline to say things they do not believe.

VII. A simple test for any new speech law

Before passing any new law that limits what Australians can say, Canberra should answer two plain questions.

One. Can we just argue it down? If something is said that is wrong, dangerous, or shameful, can ordinary public debate deal with it? Journalism, counter-argument, public pressure, people refusing to buy the newspaper, employers refusing to hire the speaker. The normal tools of a free society. If yes, no new law is needed.

Two. Is it already against the law? Incitement. Threats. Harassment. Defamation. Public-order offences. Workplace protections. Consumer law against false advertising. Most genuinely dangerous speech is already illegal, and has been for a long time. If one of the existing laws catches what is being said, no new law is needed.

Only if the answer to both questions is no, and only then, is there any case for a new restriction. And even then, only a narrow one.

Call it the Mill-Popper test. Mill, in On Liberty (1859), argued that true ideas survive bad ones so long as they are allowed to fight in the open. Popper argued that open societies stay open by keeping that fight public. The test is the dull check that makes both possible.

Applied honestly, it clears most of the territory in dispute. The federal Misinformation and Disinformation Bill would have failed it. Section 18C, in its "offend or insult" limb, fails it. The Victorian 2025 Act fails it at its "revulsion" and "severe ridicule" thresholds. The UK Online Safety Act fails on most of its content, and passes only on its narrow threats-and-false-communications provisions.

VIII. Enshrine it

Procedural discipline is not enough. A test a Canberra of 2026 agrees to apply can be quietly set aside by the Canberra of 2031. What survives a change of government is what sits in the Constitution.

Australia's protection of free expression is a shield and not a sword. The implied freedom covers federal political speech only. It does not protect artistic expression, religious expression, commercial speech, scientific argument, or the ordinary conversation of citizens. It has not prevented the drift this essay has described. It will not prevent the next bill drafted in the wake of the next attack.

It is time to enshrine free expression as a positive right.

Which model should we reach for? The natural instinct is to look at countries like ourselves – Canada, the United Kingdom, Europe. That instinct is wrong.

Canada has a Charter of Rights that guarantees free expression. It also has a proposed Online Harms Act that would fine platforms millions of dollars for permitting broadly defined "hateful content". Britain has the Human Rights Act and the European Convention on Human Rights sitting above it, and is still arresting 12,000 people a year for online speech. The reason these documents have not held is structural. Each contains a "reasonable limits" clause that lets a government trim the right, so long as the government can argue the trimming is reasonable. And governments almost always argue the trimming is reasonable. A right that hinges on reasonableness, defined by the government of the day, is not much of a right.

The United States is not an advertisement for itself on everything. America's campaign-finance settlement and its corporate-speech doctrine are cautionary tales, and Australia should not copy them. But on the central question – is a free society safer when it errs towards letting hard things be said, or towards stopping them? – the First Amendment's answer has held up better than any other. It is a plain, strong, positive statement: Congress shall make no law abridging the freedom of speech. Two and a third centuries later, it is still there. No other English-speaking common-law tradition has anything that matches its durability.

Australia does not need to become America. But on this one question, we should learn from what works. A positive right to free expression, plainly worded, with narrow and clearly defined exceptions – incitement to violence, direct threats, defamation, commercial fraud – instead of an open-ended "reasonable limits" clause. Something durable. Something that forces every future government to argue against it, not around it.

Enshrining that right would require either a constitutional amendment, which is hard in Australia and rightly so, or a federal Human Rights Act of the kind the ACT and Victoria already have at state level. The statute is easier. The amendment is stronger. Either would anchor free expression as a positive right the citizen holds, rather than a side-effect of some other rule.

Free expression is a cornerstone of Australian democracy. Cornerstones are not held in place by convention. They are laid.

IX. What an open society looks like

Australia has been an open society at its best. It has trusted its citizens with disagreement. It has argued its way through hard speech. It has not, on the whole, legislated the emotional temperature.

We are drifting. The Mill-Popper test, applied seriously, would stop the drift. A positive right to free expression, properly drafted, would make the stopping durable.

Solzhenitsyn's minimum moral act, for the writer and the citizen living through a society's quiet decisions about what could be said, was the refusal to knowingly support lies. That refusal is available to every Australian, every day. It is the starting point. It is not the destination.

---

Sources

• Karl Popper, The Open Society and Its Enemies, Volume I: The Spell of Plato, Routledge, 1945.
• John Stuart Mill, On Liberty, 1859.
• Isaiah Berlin, "Two Concepts of Liberty", in Four Essays on Liberty, Oxford University Press, 1969.
• Aleksandr Solzhenitsyn, "A World Split Apart", Harvard Commencement Address, 8 June 1978.
• Aleksandr Solzhenitsyn, "Live Not by Lies", 12 February 1974.
• Václav Havel, "The Power of the Powerless", 1978.
• Yascha Mounk, "Karl Popper's Paradox of Tolerance, Correctly Understood", Persuasion, 8 August 2024.
• The Times, Freedom of Information analysis of UK online-communications arrests 2023, published April 2025.
• R v Connolly, Crown Court at Birmingham, 31 October 2024.
• Eatock v Bolt [2011] FCA 1103.
• Faruqi v Hanson [2024] FCA 1264.
• Ridd v James Cook University [2021] HCA 32.
• Lattouf v Australian Broadcasting Corporation [2025] FCA 669.
• Parliament of Victoria, Justice Legislation Amendment (Anti-vilification and Social Cohesion) Act 2025.
• Parliament of Australia, Communications Legislation Amendment (Combatting Misinformation and Disinformation) Bill 2024, withdrawn November 2024.
• Parliament of Australia, Member for Wentworth, motion on antisemitism and vilification amendment, 4 February 2025.
• Konstantin Kisin, Oxford Union debate, 12 January 2023.
• BBC Question Time, Nick Griffin appearance, 22 October 2009.
• Conseil constitutionnel, Décision n° 2020-801 DC (Loi Avia), 18 June 2020.
• Canadian Charter of Rights and Freedoms, Part I of the Constitution Act, 1982, ss. 1 and 2(b).
• European Convention on Human Rights, Article 10.