According to DEV Community, a developer building a site of free browser minigames found that dragging vertically on a game board stopped the page from scrolling on phones. The board covers 60 to 70 percent of the screen, so reaching the guide and leaderboard below was awkward. The cause was the developer's own code: 37 components set touch-action: none on the whole board to make touch controls work.
The key rule is that touch-action is an intersection. The browser combines the values on the touched element and all its ancestors, so a child cannot turn scrolling back on if a parent has set none. Setting pan-y on an inner canvas did nothing. The fix was to leave the board container unrestricted and place touch-action: none only on a central overlay, called a PlayZone. The top and bottom 12 percent are left alone, so dragging there scrolls the page.
The developer also sorted the games by input style. When scrolling begins, the browser fires pointercancel, but pointerdown has already fired. Tap games that act on pointerdown would register an unwanted input, so touches starting in the edge bands are now ignored. Drag games must treat pointercancel as a cancel, not a commit. WordSearch checked answers on cancel and Unblock committed a block move on cancel, and both were fixed.
The developer tested by sending emulated touch events through Playwright and measuring scrollY. Across 35 games, swipes in the play zone scrolled 0px, while swipes from the edges scrolled roughly 75 to 128px. The author lists limits: no testing on real iOS Safari or Android Chrome hardware, drags starting on edge rows in WordSearch, GemMatch and Nonogram get cancelled, and the band sizes are design choices rather than data-driven.
Why it matters
Blocking touch gestures on a large element can quietly trap visitors on mobile. If a big interactive area covers most of the screen, people may be unable to scroll to the content below it. That can hide guides, forms or links you want them to reach.
The article also shows that a fix can create new bugs. Changing which gestures the browser handles alters what events your code receives, so handlers that assumed a gesture always finishes may misbehave. The author is clear that emulated tests are not a substitute for checking on real phones.
Source: DEV Community. This brief summarises their reporting; read the original for the full detail.