Scrolling the wheel up makes my view of a webpage go up.
Scrolling the wheel up makes the part of the wheel I'm touching go forward (toward my computer screen). It thus makes sense by analogy with my first sentence that scrolling up would make the view go forward. Zooming in is similar to making the view go forward.
Surely 'going forward' would be scrolling down a webpage? So by that thinking it would make sense for scrolling down to be zooming in. You could also see it like pulling the thing closer towards you.
(Very difficult to unlearn though, I'm not in any hurry for zoom-in to be anything other than scroll-'up'!)
When you scroll down a webpage, do you think of it as going forward? I don't think people think about it that way. They think of it as going down. I think people think of it more in terms of spatial terms than in terms of text reading direction terms.
Also consider if you see a signpost with an arrow pointing to the right. That means go right. If the signpost has an arrow pointing up. That means go forward. Or any map, e.g. physical, or gps navigation software, or videogame in-game map, up on the map corresponds to forward in real life.
For people with the alternate configuration, having making the scroll wheel move down make the content zoom in makes sense.
I have scroll wheel up mean scroll up, but 100k stars has backwards scrolling mechanisms for me. So for me, the 100k stars is inconsistent with my system settings.
> Zooming in is similar to making the view go forward.
Consider a timeline or history, which is usually thought of in terms of having an axis with endpoints "beginning" and "end", and navigation directions "backward" and "forward."
If you were to bind a chorded hybrid gesture (e.g. WheelUp/WheelDown while holding Control) to act as accelerators for going forward and back through your browser history — which direction would you expect to take you "forward" in time, and which direction would you expect to take you "back" in time? Personally, I'd expect down to be "forward" and up to be "back", because it lines up in a hierarchical sense with the timeline of reading and scrolling an individual page. You always end up scrolled to the top (i.e. the beginning) of a new page after navigating; and if you finish reading a page, you're at the bottom (i.e. the end) of the page. Because of this, in a browser with a hypothetical infinite-canvas model for history navigation[1], the destination page for each navigation would probably be laid out below the source page, such that you'd get back to the page you were on before navigating (i.e. "go back") by scrolling up past the beginning of the destination page you landed on. And likewise for scrolling down past the end of the source page, to take you forward again. (Or to navigate you to a <link rel="next"> page!)
[1] Really, you'd want to model this as links "unfurling" the new page in a way that slices the source page apart at the vertical position of the link, inserting the new page right below the link, before resuming the source page. This would mean that scrolling up would take you back to your recent-most scroll position within the source page. (But it would also mean that scrolling back down, would present a fork in the timeline: did you want to "down" as in down past the unfurled next page, or did you want to go "down" as in forward in history at the point of navigation?)
This "spatial model of time" — the model of "you're on a journey with no set destination, and you proceed 'forward in time' with each step forward into the unknown, but at any time can 'back up' in time [and thus in space] to where you were before you took that step" — is very common. It's the "spatial model of time" of an append-only log file; of a text adventure game; of a written diary; of those fancy Apple product landing pages where scrolling acts as a time-scrubber for an animation.
Compare/contrast: the timeline view in macOS Time Machine (e.g. https://help.apple.com/assets/65A8106E7C69B635140E606E/65A81...). Here, going "up" does take you "deeper" — but that's because the spatial model here is that you start at the latest version, at the "end" of the timeline, which is also the most "shallow" place; and you're "diving deeper" into history by going backward in time — which is both "upward" in a vertically-laid-out linear timeline, but also "forward and inward" in the spatial model.
This is a different "spatial model of time" — it's a model of "historical time as an investigation, where you start in the present, looking further back with each step you take forward, until you reach the earliest time, which is a terminal point you cannot pass." This model is not common in HCI, but rather is very case-specific.
If you consider the user story of someone exploring a 3D space as a "spatial model of time" — then you would expect it to be an "unbounded journey forward in arbitrary direction by steps"; not an "investigation that retraces a history step-by-step to a terminal origin point." You would expect that the "entry point" to the 3D space is equivalent to the top of a webpage, or the start of a timeline.
---
That being said — if you replace the scroll-wheel with a flight stick, then it becomes seemingly natural that "up means dive", because that's what planes do. But even that's a corruption — before control sticks, there were only control yokes; and yokes have an actual Z input axis. You push the yoke deeper into the cockpit, away from yourself to dive; and you pull the yoke back toward yourself to climb.
The mapping of a control yoke's literal "make the yoke be further away from you" to a stick's "tilt upward" [where, therefore, the tip of the stick is further away from you] may seem principled, but it's entirely arbitrary; they could have just as well followed a different line of logic, and had control sticks interpret "tilt down" as "tilt the nose of the plane down, i.e. dive." (Both yokes and sticks were designed originally to operate with direct mechanical linkage. So there was no choice for how the yoke needed to work — the Z axis is the only axis of a yoke that you can push/pull on, and you need to be pulling on something to cause the linkage to in turn pull on the tips of the elevators. [Think bicycle brakes.] But a tilt in any direction on a flight stick can be made to generate "pulling" force away from center — and so any tilt-direction of a stick could have been mechanically linked to pulling on the elevators.)
And that's why so many people think inverted-Y-axis viewpoint controls are dumb. Even when planes did it, it was for no good reason :)
When I say forward, I'm thinking of forward in terms of 3D space, not forward in terms of time.
Thinking about time doesn't bring clarity. As you point out, you could lay out a timeline such that forward in space means forward in time (text adventure game). Or you could lay out a timeline such that forward in space means backward in time (macOS Time Machine).
The 100k Stars zoom has no notion of time, just of space. When thinking about space, up generally corresponds to forward. Think of a signpost with an arrow pointing right, that means go right. If a signpost has an arrow pointing up, that means go forward. Or any map, e.g. a physical map, or gps navigation software, or a videogame minimap: in those up means forward.
Our maps are meant to be read from above. If we lay a map on a table, we look down at it, and forward means forward. If we then pick up the map, so that our face faces forward instead of down, now up means forward.
If we had a map meant to be read from below, and we stuck the map on an upside-down table hovering over our heads and bent our necks backward to read it, then if forward means forward on that map, when we take the map off the hovering table and stop bending our necks back, up would mean backward. That's not how we make maps though.
