Follow mode replicates one person's viewport changes to everyone else

What it is

Follow mode replicates one person's viewport changes in a shared space — scrolling, zooming, jumping — live onto everyone else's screens: the leader scrolls down the document, and every viewer's frame scrolls with them. It is the mechanical implementation of "we are all looking at the same place," common in co-browsing, shared presentations, and "follow this person's cursor" in collaborative documents. Unlike the old strict "what you see is what I see" (WYSIWIS) sharing, following is voluntary coupling: by default each person owns a viewport, and following hands that viewport to someone else, reclaimable at any moment. This "can split, can join" property is exactly what separates it from screen sharing — viewers remain in their own client, with their own controls and cursors.

Why it happens

Follow mode attacks the cost of reference-frame alignment. The largest hidden expense in collaborative discussion is "where is the third paragraph you mean" — with different viewports, every act of reference must first align position before communication can start. Following turns alignment into a persistent state instead of a repeated action: the follower's viewport snaps to the leader's position, and reference collapses from "page 4, second figure" to "this one." Implementation-wise it is one-way broadcast of viewport events: the leader's scroll and zoom events are replayed in followers' clients, while input rights stay untouched — followers look at the same place but can still type and select. Precisely because the coupling is one-way, the power is asymmetric: wherever the leader goes, everyone must follow. That asymmetry is both the efficiency of following and the root of all its problems — disorienting movement, hard escapes, unaware leaders.

Studying it

• Paradigm: comparative experiments across three ways of sharing a view (free viewports with verbal alignment / screen sharing / follow mode), measuring alignment utterances ("do you mean this one?"), task completion time, and spatial reference errors; deployment logs from co-browsing systems can show how much session time is spent following, voluntary break-off rates, and post-detach return behavior.
• Variables: viewport coupling method and group size as independents; alignment utterances, task duration, reference errors, and subjective "kept up" ratings as dependents.
• Use in interface research: deciding whether a collaborative context wants following, screen sharing, or free viewports — remote walkthroughs, pair review, and read-alone contexts give different answers.
• Methodological caveat: the benefit of following depends on the share of task time that is genuinely "look at the same thing together." A task designed for full-time co-viewing overestimates its value; include phases where participants must privately inspect other regions, or the cost of detaching never gets measured.

Where it stops holding

Following holds only under a one-way attention flow: one guide, many receivers. The moment bidirectional reference appears ("look at this on my side"), one-way following fails and you need floor handover or mutual following, with sharply higher complexity. Document-like content follows cheaply (the viewport is just scroll position); infinite canvases and 3D spaces follow expensively (view direction and orientation must also sync), and follower disorientation gets worse. There is also a headcount asymmetry: one leader broadcasting to dozens of followers is fine, because broadcast creates no interaction; multiple rotating leaders demand constant handover.

Applying it

• Provide an explicit follow entry point (tap avatar, tap "follow") and show the leader how many people are following.
• While following, keep followers' local actions (selection, copy, a soft peek-scroll that does not break the follow) — do not turn following into a read-only locked mirror.
• When the leader jumps (search, outline navigation), give followers a brief positional cue so the frame does not fly away without warning.
• Verification: code the collaboration recording for "position-alignment utterances" (which page, which figure). After follow mode ships, that class of utterance should drop markedly; otherwise the coupling is not doing its job.

Related

• Same group: V3.04.2 Fast movement by the followed party disorients followers · V3.04.3 Following must be escapable at any time, returning to one's own place · V3.04.4 Who follows whom must be visible to both sides · V3.04.5 Shared viewing suits presentation, not parallel work
• Nearby: V2.02 Workspace Awareness · V5.06 Screen Sharing and Pointing
• Search terms: follow mode · viewport coupling · co-browsing · WYSIWIS