Why Australia’s LGBT Census Is a Bad Idea

In August 2024, the government of Australian Prime Minister Anthony Norman Albanese, together with the Australian Bureau of Statistics (ABS), declined to include questions on sexual orientation and gender identity in the national census questionnaire, citing a reluctance to provoke a “divisive debate.” Activists and some parliamentarians saw this move as a betrayal, accusing the authorities of attempting to make minorities statistically “invisible.” Under intense political pressure, the cabinet backed down: on August 11, 2026, the census went ahead with an expanded list of questions.

Although supporters of the initiative are celebrating a victory in the fight for equality, this case goes beyond the protection of minorities. Behind the discussions about the proper allocation of resources lies the problem of the limits of state curiosity and the boundaries of personal information that a citizen is obliged to disclose at all.

Proponents of expanding the questionnaire rely on practical necessity: the state distributes taxes for healthcare, education, and social housing. Without data on the numbers and geographic distribution of LGBT people, they argue, it is difficult for officials to fund specialized clinics or targeted psychological support programs. Furthermore, cross-analysis — comparing sexual orientation with income levels, education, or dismissal rates — helps to reveal social inequality.

Before criticizing, let us recall history. The census has always been an instrument by which the state makes the population “legible.” Initially, it solved simple fiscal and administrative tasks: counting people to allocate seats in parliament, collecting taxes, and registering men fit for military service.

However, as government powers expanded, the census turned into a mechanism for detailed classification. Residents began to be divided by income level, ethnic origin, education, housing type, and now they have reached sexual orientation.

Analysts at research centers, such as the Foundation for Economic Education and the Cato Institute, have long pointed out the nature of this process: the bureaucratic apparatus needs standardized categories to manage people conveniently. The more detailed the population is classified, the easier it is to control, and by agreeing to hand over intimate details to government agencies, society inevitably cedes part of its own autonomy.

Collecting such data also undermines intra-family privacy. In Australia, the general census questionnaire for everyone living under one roof is traditionally filled out by one person — usually the head of the household. Although the law allows any family member to request an individual form to keep their answers secret, this step requires courage and conscious, proactive actions.

If a teenager is hiding their identity, the general form will most likely be filled out by their parents, who may not know the truth or may intentionally distort it. The very fact of requesting a separate form immediately betrays a desire to hide something. In families with homophobic views, attempting to use a state protection mechanism can only provoke conflict.

At the same time, the need for social statistics does not mean that information should be collected specifically through a universal census. The basic principle of information security — data minimization — states that the collecting party is obliged to request exactly as much information as is strictly necessary for a specific task. To assess the demand for specific medical services or measure the level of discrimination, agencies do not need to link the intimate preferences of every resident of the country to their home address.

For decades, sociological services and independent institutes have successfully solved such tasks using representative samples. Voluntary and truly anonymous surveys of tens of thousands of people provide a sufficient picture for budget planning. The Australian government, however, has chosen to use a sledgehammer where a scalpel is needed.

In response to fears of data leaks, statistical agencies usually promise anonymization of questionnaires: names are supposedly securely separated from answers, and researchers work exclusively with aggregate numbers. However, in the era of big data, absolute anonymity remains a technical illusion. The more characteristics — sex, exact age, postal code, profession, ethnicity, and now sexual orientation — are combined in one record, the more unique and vulnerable the digital profile becomes.

The scale of the problem was vividly demonstrated by the United States Census Bureau. During an internal experiment, analysts took the officially published tables of the 2010 national census and, using methods of mathematical reconstruction combined with commercial databases, reliably de-anonymized the records of tens of millions of Americans.

Today, with the development of AI, disparate information is cross-referenced even faster and more accurately. Adding an intimate variable to it multiplies the risks: any successful cyberattack on government servers will create a ready-made list of LGBT people. Considering that even in Australia there are radical groups actively looking for targets, a leak of such archives could lead to real crimes.

World experience shows that information collected with guarantees of confidentiality has repeatedly turned against citizens during political crises. A classic example is the American census during World War II. Shortly after the attack on Pearl Harbor, the US government forcibly sent tens of thousands of citizens of Japanese descent to internment camps. To find them, the military turned to the Census Bureau, which handed over the information without hesitation.

History is unpredictable. No one can guarantee that the data collected today will not one day fall into the hands of a radical cabinet of ministers. By cataloging minorities, the modern state, even with good intentions, leaves its successors a ready-made infrastructure for potential persecution. Practice shows that the only fully secure database is one that was never created.