CSS Working Group explores “interest” states with triggered delays
While the CSS Working Group’s latest teleconference minutes covered the usual modules — Values, Easing, and Selectors — the most notable discussion involved a proposal that could change how we trigger UI elements in response to user intent. The idea originates from the OpenUI proposal, the same body that pushed forward the Popover API and the customizable select element.
The core concept, known as an “interest” state, would let developers specify a delay that determines when an action fires after a particular interaction — for instance, holding a pointer over a <button>, a long tap, or focusing an element. The proposed use case is intuitive: when a person hovers over a trigger, they are signalling interest, and a tooltip or complementary interface could appear after a prescribed beat rather than instantly.
So. Does this belong in CSS? Or should it be elsewhere? Does the approach make sense? Are there better ideas? Most interested in the last.
The minutes record debate on whether this is best situated in CSS, HTML, or JavaScript. One reviewer notes that JavaScript already allows developers to inspect media query states and alter the document, so this could be handled outside CSS entirely. Others argue that CSS’s existing ergonomics make it the natural declarative home for such timing logic.
A separate line of discussion focuses on whether developers should control the delay interval at all. The original rationale for a delay is to prevent hover from being instant — this aligns closely with platform-standard behavior. If custom delays are disallowed, CSS could adopt a consistent set of keyword values that map to platform conventions. That raises an open question: while 800ms seems more direct, a keyword such as short or long could better reflect or approximate the user agent’s native timing, preserving the intent without exposing raw numbers.
The group is aware that web authors already manage similar behaviors using animation-delay and transition-delay, often applied globally via the universal selector or a prefers-* media query. That existing practice shows there is a real need for standardized, ergonomic control over interaction-based latency.
Whether the final design uses explicit time values, a new pseudo-class, or a dedicated property remains undecided. As one participant summarizes: “I think this sounds reasonable and I’d like to explore it. Unsure if this is the exact shape, but this space seems useful to me.”
