Why are unknown or legacy references not reported as errors?
HTML parsing is intentionally compatible with legacy content. The decoder follows the browser parser: recognized named and numeric references are decoded, while unknown references remain text.