Although usually-associated with abandoned card catalogs or outdated research methods, subject headings and controlled vocabularies were among the most powerful information innovations of the last century. Their near-total adoption by GLAMs across the word permitted the organization and control of a seemingly endless proliferation of knowledge. Over the past few decades, however, they have faced increasing criticism—especially over terms describing marginalized groups. Long-established systems, like Library of Congress, Dewey, or Universal Decimal Classification have been disparaged as inappropriate, misleading, or outrightly offensive. Solutions so far offered (including the use of folksonomies or tagging) have been matched by a near-equal amount of research pointing out issues with tagging and uncontrolled vocabularies that lack meaning. This presentation, which builds off a near-decade of work by GLAM professionals, will suggest that linked data vocabularies offer a possible path forward. Linked data, also called Web 3.0, is the practice of structuring information in a way that both computers and humans can easily understand, and it holds revolutionizing implications for GLAMs. Avoidance or ignorance of these questions in Web 1.0 and 2.0 had (and continue to have) dismaying effects, but rather than allowing history to repeat itself, minoritized groups staged an intervention into these new technologies, and offer a variety of options for alternative representation and organization. After sketching out the above (~2m), I will introduce the concepts, principals, and implications of linked data for galleries, libraries, archives, and museums (~3m). Next, I will focus on the use of Homosaurus, the International LGBTQ Linked Data Vocabulary as an ethical approach to subject cataloging that originates from the community described (~4m). In closing, I will mention some GLAMs that already use Homosaurus as an alternative or addition alongside conventional systems (~1m), and recommend ways to incorporate it into Omeka, Scalar, or even conventional MARC.