The mechanics of the shift
Scroll used to perform one function: translate the document vertically so that content outside the viewport came into view. It was a navigation primitive, the digital equivalent of moving a sheet of paper under a reading lamp. Whatever you saw was a direct, proportional result of where your thumb or wheel was positioned. Stop, and everything stopped. The mapping was one-to-one.
The first break from that contract arrived with parallax — background layers moving at different speeds from foreground ones, implying depth while the user scrolled normally. Mildly deceptive, but the core exchange was preserved: scroll position still controlled what you saw, and reversing the scroll reversed the effect.
What followed was categorically different. Scroll-linked animations began driving sequences: elements entering at precise breakpoints, text assembling word by word, a diagram constructing itself through a choreographed chain of keyframe states, all triggered by scroll progress rather than time elapsed. The scroll bar became a playhead. The document became a timeline.
What the reader actually loses
The word that matters here is control. In the older model, control was the reader's entirely. They could jump, skip, reverse, linger. Speed of reading equalled speed of scroll; there was no mismatch between cognitive pace and display pace.
Scroll-as-projector breaks this in two ways. First, the content is gated behind physical travel — you cannot see paragraph three until you have scrolled through the animation that precedes it, regardless of whether you want to watch it. Second, the sequence only makes sense at a particular scroll velocity. Move too fast and you miss a staged transition mid-state, catching a diagram half-assembled and semantically broken. Move too slowly and the content dribbles out at a pace that feels narrated rather than read.
The result is a page that behaves less like a document and more like a presentation deck someone else is advancing. The reader has not lost information; they have lost agency over the rate at which information arrives. For editorial content, this is a meaningful regression. For a product that uses scroll-linked animation to explain a physical mechanism or a data transformation, the constraint is at least purposeful — the sequence is the explanation.
The problem is that the technique rarely observes that distinction. Scroll-linked animation became a gesture of production value rather than a solution to a specific communication problem. Movement nobody asked for is one cost; gated content is another, and the second one is harder to test for because it doesn't file an accessibility complaint.
The frame rate trap
There is also a technical cost that compounds the experience one. Scroll-linked animation is driven either by JavaScript listening to scroll events, or — more performantly — by the CSS animation-timeline: scroll() mechanism that modern browsers now support natively. The native route runs off the main thread and is substantially smoother. But even there, the animation is only as smooth as the frame rate the device can sustain under layout pressure. On a page with multiple scroll-linked sequences running simultaneously, mid-range hardware begins to drop frames, and a dropped frame in a scroll-linked animation reads not as a brief stutter but as a jump cut — a visible discontinuity in a sequence that was supposed to be seamless.
Purposeful, narrowly applied, scroll-linked animation genuinely earns its place.
The projector metaphor, then, is accurate beyond the rhetorical. Film projectors also dropped frames, and the artefact was the same: a visual skip that broke the illusion at precisely the moment it was most needed. The difference is that a film editor knows the projector will run at 24 frames per second. A scroll animation designer has no such guarantee, and the audience's hardware is the variable nobody controls.
Purposeful, narrowly applied, scroll-linked animation genuinely earns its place. The question to ask first is whether the sequence is the argument — whether the thing you are explaining only becomes clear in motion and in order. If the answer is that the content would survive linearisation, the projector should be switched off and the page returned to the reader.