It will never cease to amaze to me that Apple (under Jef Raskin who also wrote about this in The Humane Interface) observed and fixed problems with interactions like this in SYSTEM 6 in like 1993 or something - only for the same problem to have to be re-descovered and re-fixed decades later.
This looks it's describing the same technique as the last example in https://emilkowal.ski/ui/you-dont-need-animations (and that whole post is worth a read, I think, if you found this interesting).
It's such a small detail that I wouldn't have otherwise noticed (because I don't build animations, so it doesn't cross my mind), but now that I know about it, I think about it all the time.
>> Hysteresis is the dependence of the state of a system on its history. For example, a magnet may have more than one possible magnetic moment in a given magnetic field, depending on how the field changed in the past.
How is this different from the observation that, say, a non-magnetic iron ball may have more than one possible velocity in a given gravitional field, depending on whether it was recently thrown upward or dropped from above?
I greatly appreciate this attention to detail, and there isn't enough of this in the world. For example, in Visual Studio popups on hover happen so fast you can barely select anything, or you end up clicking on the popup instead of selecting a word.
My favorite is the info icons that show a tooltip on hover, but if you click them, they disappear the tip if it’s showing and never show it if it hasn’t appeared yet.
If there is a primary UI element on screen that the user can interact with and moving the mouse towards that causes a secondary, alternative, UI element to appear, the second element needs an active modifier key to bring it up. Tooltips and such are total trash and get in the way of usability. If you have something important then put it on the screen to begin with don't have it randomly jump in front of the user who is already mid action with something else.
I ran into the same problem about 20 years ago when implementing hover-based dropdown menus on a website. It was only about 10 lines of JS and a couple global variables but I was the only person who understood it. :-P
(These days I would have included more comments.)
Naturally, there was a fallback to pure CSS menus, because we weren't barbarians.
Not OP, but my guess is things like timing delay--20 years ago, CSS animations and transitions weren't a thing yet, especially since 20 years ago also puts us in the era of IE6 being a major browser.
Basically, have a global temperature. When the pointer is in something that has a tooltip, the temperature rises. Otherwise it cools (possibly at a different rate). A tooltip is shown only when the pointer is in bounds, and the temperature is above a certain threshold.
I spend a lot of time seething at how poorly modern software interaction is, and I've come to the belief that they basically solved most of the window-based GUI computing interactivity problems by around XP, and have just been gutting them since then.
My three most-hated modern trends
- Disappearing scroll bar with content that you have to scroll for
- Pills going across the top of the screen that are "most-used" filters instead of a real filtering menu
- Clickable text and non-clickable text appearing the same. I mean come on. Who the fuck is designing these things? That's the one I can't wrap my mind around. How can a self-respecting programmer put out a UI where the display and clickable elements look exactly the same? What are we doing man
I think the real problem is we don’t have a native-quality UX toolkit on the web - the built in controls are just horrible. So every app has to reinvent the wheel from scratch, or use one of a few hundred (thousand) web component libraries that have all also had to reinvent the wheel from scratch (I guess a div and some text rendering isn’t scratch per se but you get my point)
Well, Microsoft was interested in making it easy to develop Windows applications, and also in making it easy and pleasant to use Windows applications. They had reason to invest in learning how that could be done and in writing utilities that would let you benefit from their knowledge; doing that made more money for them.
Browsers are different. They don't really benefit from being written for, or from being used. This seems like enough to explain the difference. They provide some functionality, and giving your users a good experience is your problem, not theirs.
I feel like I've seen complaints that Apple has forgotten things it already knew about how to design a usable interface; that wouldn't be explained by this model.
It will never cease to amaze to me that Apple (under Jef Raskin who also wrote about this in The Humane Interface) observed and fixed problems with interactions like this in SYSTEM 6 in like 1993 or something - only for the same problem to have to be re-descovered and re-fixed decades later.
Because every single application rebuilds their own operating system
Chesterton's Fence.
Why on earth do we need this thing here?
Oh, that's why.
Raskin left Apple in 1982.
This looks it's describing the same technique as the last example in https://emilkowal.ski/ui/you-dont-need-animations (and that whole post is worth a read, I think, if you found this interesting).
It's such a small detail that I wouldn't have otherwise noticed (because I don't build animations, so it doesn't cross my mind), but now that I know about it, I think about it all the time.
I think I would describe this as an example of hysteresis. Neat! https://en.wikipedia.org/wiki/Hysteresis
Interestingly, there are more use cases for hysteresis in UI design, e.g. it was famously used in nested menus in Windows: https://www.mackido.com/Interface/hysteresis.html
I think there was also an article about it on The Old New Thing, but I can't seem to find it anymore.
Like the HN post we're commenting on, so too does this example also get regurgitated every few years.
Find modern examples by Googling "menu safe triangle." Here's one from 5 days ago https://www.framer.com/blog/cursor-trajectory/
> I asked the Framer Agent to fix it, and within minutes it had built and applied what I now call trajectory-based menu aim
>> Hysteresis is the dependence of the state of a system on its history. For example, a magnet may have more than one possible magnetic moment in a given magnetic field, depending on how the field changed in the past.
How is this different from the observation that, say, a non-magnetic iron ball may have more than one possible velocity in a given gravitional field, depending on whether it was recently thrown upward or dropped from above?
I greatly appreciate this attention to detail, and there isn't enough of this in the world. For example, in Visual Studio popups on hover happen so fast you can barely select anything, or you end up clicking on the popup instead of selecting a word.
My favorite is the info icons that show a tooltip on hover, but if you click them, they disappear the tip if it’s showing and never show it if it hasn’t appeared yet.
Opus Magnum suffers from this. It feels like punishment for taking the most direct route to get the cursor where I want it.
If there is a primary UI element on screen that the user can interact with and moving the mouse towards that causes a secondary, alternative, UI element to appear, the second element needs an active modifier key to bring it up. Tooltips and such are total trash and get in the way of usability. If you have something important then put it on the screen to begin with don't have it randomly jump in front of the user who is already mid action with something else.
I ran into the same problem about 20 years ago when implementing hover-based dropdown menus on a website. It was only about 10 lines of JS and a couple global variables but I was the only person who understood it. :-P
(These days I would have included more comments.)
Naturally, there was a fallback to pure CSS menus, because we weren't barbarians.
what did the JS provide that the CSS solve wasn't the only solve?
Not OP, but my guess is things like timing delay--20 years ago, CSS animations and transitions weren't a thing yet, especially since 20 years ago also puts us in the era of IE6 being a major browser.
Chris Coyier also posted a version using only HTML/CSS, no JS: https://blog.master.dev/delayed-then-instant-tooltips-with-h...
Basically, have a global temperature. When the pointer is in something that has a tooltip, the temperature rises. Otherwise it cools (possibly at a different rate). A tooltip is shown only when the pointer is in bounds, and the temperature is above a certain threshold.
What about using mouse velocity? If the mouse is changing position at a certain rate, don't show the tooltips.
i only show tooltips if you wiggle the mouse cursor over an element
On a Catholic church site I designed you have to do the cross pattern with your mouse and say 3 Hail Marys to see the tooltip.
There is no contact form. Only a Confession Textarea
At least the cookie popup comes with wine
Windows taskbar has done this forever and I never once noticed it consciously. Good catch.
Feels like the delay also gets implemented into displaying the tooltip on tap on a touch interface which is probably an error.
I was disappointed to see the new web popover api doesn't have this basic usability feature built in
Is this post text claude-generated? Very creepy to read around the solution section. It reads like the agent report.
Hilarious. Web devs discovering UX from Win95. What's next, 3d buttons?
I know right? It's like interaction design came and went over a period of 30 years and now we're back to the beginning again.
I spend a lot of time seething at how poorly modern software interaction is, and I've come to the belief that they basically solved most of the window-based GUI computing interactivity problems by around XP, and have just been gutting them since then.
My three most-hated modern trends
- Disappearing scroll bar with content that you have to scroll for
- Pills going across the top of the screen that are "most-used" filters instead of a real filtering menu
- Clickable text and non-clickable text appearing the same. I mean come on. Who the fuck is designing these things? That's the one I can't wrap my mind around. How can a self-respecting programmer put out a UI where the display and clickable elements look exactly the same? What are we doing man
I think the real problem is we don’t have a native-quality UX toolkit on the web - the built in controls are just horrible. So every app has to reinvent the wheel from scratch, or use one of a few hundred (thousand) web component libraries that have all also had to reinvent the wheel from scratch (I guess a div and some text rendering isn’t scratch per se but you get my point)
Well, Microsoft was interested in making it easy to develop Windows applications, and also in making it easy and pleasant to use Windows applications. They had reason to invest in learning how that could be done and in writing utilities that would let you benefit from their knowledge; doing that made more money for them.
Browsers are different. They don't really benefit from being written for, or from being used. This seems like enough to explain the difference. They provide some functionality, and giving your users a good experience is your problem, not theirs.
I feel like I've seen complaints that Apple has forgotten things it already knew about how to design a usable interface; that wouldn't be explained by this model.
The pure CSS version seems more interesting to me: https://blog.master.dev/delayed-then-instant-tooltips-with-h...
Perhaps I'm just an old grump, but I prefer the No Delay version in the example.
I like predictable, consistent, and fast.
Same. To me exactly delayed version is annoying. Non-obvious rules which guide behavior make it feel like there's a bug.