> hard (and therefore slow) to use.
You assess the use case, as in each case the user goes from want to finish. If it's slow and hard, you fix that.
1 click is the fastest.
An obvious button that does exactly what you want is the easiest.
There is nothing slow or hard about an interface optimized based on a complete understanding of what the user wants to the point that they can distill it in one button.
What the user wants is important. Understanding how the user works in order to get there is also important. The most efficient path for a person is not necessarily the one with the fewest clicks/shortest mouse travel.
Mouse clicks and distance are two of the most basic metrics. And straight forward. So why wouldn't you optimize there?
But what you're referring to is that which can offset and add to those metrics. "Mouse click" in and of itself means nothing in the context of the program. So it's the "other stuff" that pushes against the goal being accomplished in 1 click.
So the answer is, considering that "other stuff", what is the most efficient, as in, fewest clicks and shortest distance possible?
The gripe that should be had is not even optimizing on this level, and even worse, prioritizing "fancy" animations as if that's a good thing and as if that's your main job.
> The most efficient path for a person is not necessarily the one with the fewest clicks/shortest mouse travel.
The most efficient path for that person is necessarily the one with the fewest clicks/shortest mouse travel (for that person).
And that is the optimization process of 1 use case (user usage case path).
The goal of UI is to optimize efficiency across most use cases (and hence users) as measurably possible.
(all the fancy animations can be added after that, so you can bill for those hours, and add an obvious "animations off" button, and you'll literally make everyone happy).
That's the McNamara fallacy: https://en.wikipedia.org/wiki/McNamara_fallacy Basically: enemy bodycount is easy to measure and thus you should optimize for it. Conversely if you can't measure it, it must not be important. The fallacy is that enemy body count is not wrong per se, but oversimplified.
I don't disagree with focusing on making most use-cases faster, but treating a human like a machine ignores things like error rates, understanding, lasting impressions, and ease of learning, which are important for users who have a choice in software. If you have vendor lock for your software, you don't need to optimize at all so it's kinda moot.
I literally just stated how you should observe "all others". With everything important considered, you should then "count your bodies" and optimize. McNamara would agree with that.
Most UIs don't. Just take any app or program you use daily, and imagine better paths to the goals you repeat on a daily basis. If you're a UI person, you should be focused on optimizing these paths.
You might, but to do so based solely on those metrics is a mistake. That's because those metrics don't fully capture what makes a UI efficient for human use. Other things, like cognitive load, are more important.
Does it matter if it only takes two clicks to do a thing if that usage path isn't one that meshes with the way I think? I think not, because it means that I'll spend additional time thinking about how to accomplish the task rather than just accomplishing it.
> and even worse, prioritizing "fancy" animations as if that's a good thing and as if that's your main job.
I 100% agree with you there! Animations are commonly misused and dramatically overused, and I think that more UIs would be improved by omitting them.
> The goal of UI is to optimize efficiency across most use cases (and hence users) as measurably possible.
If we're talking about "efficiency" in terms of "minimizing the use of the mouse", then I disagree. If we're talking about "efficiency" in terms of what helps a person to accomplish a task in the best way possible, then I agree -- but focusing mostly on those metrics excludes so much other important stuff as to give a distorted picture of the situation.
> You might, but to do so based solely on those metrics is a mistake. That's because those metrics don't fully capture what makes a UI efficient for human use.
What do you mean? Out of `n` total possible actions, we can create peak UX where every action needs AT MOST only one click, and `log4(n)` [1] pixels of mouse distance.
You say your website has one googol (`10^100`) total possible actions a user can make? Fear not, my friend. Your user needs to move their mouse only a measly 177 pixels in the worst case (assuming Python's `math.log(10 ** 100, 4)` is correct).
Let me enlighten you in The Way Of UX:
1. The user clicks anywhere. This is zero mouse movement (this is important). You can choose if it's on right-click, or on left-click-while-pressing-shift, or some similar condition. It's important that mouse is not moved.
2. On doing that, a "radial menu" appears around the cursor, with four choices: "up", "right", "down", and "left". The reason for being only 4 choices becomes obvious in the next point.
3. If the user moves the mouse ONE PIXEL in any direction, that option from the tree is chosen. You can display text labels if you want, as big as you want, as long as you don't mess with the "moving the cursor literally one pixel selects de option"; remember we're optimizing for mouse distance here.
4. When you reach the final option of the radial menu, the option is automatically chosen without needing to click.
I fail to see what is wrong with this. This is The Way Of UX. This is what optimized perfection looks like.
/s
---
Still, we must not miss the forest for the trees. Something like this would be nice for a keyboard-driven menu.
That means now we need to measure total hand movement distance!! We must not forget that the least movement we do, the most accessible we are to most people.
We need to compare between the total movement a hand needs to do while moving the mouse one pixel in any direction, and the total movement it needs to do to pick different arrows.
It is imperative that we optimize for the least movement possible to achieve peak UX, so a full study is needed to compare both options; without this most important study, we'd be building a literally unusable interface, since it would not be peak UX.
/s for real this time.
---
[1]: Assuming my math is correct; which it might not be.
Not solely.
> "efficiency" in terms of what helps a person to accomplish a task in the best way possible
Correct. THEN optimize the metrics.
I didn't realize there was a myth that some users prefer slower. They don't. If they could be done they'd rather be done. Especially if this involves work. Most users that use software at their job didn't choose that software. The least we can do as engineers is let them go home to their family sooner.
[...]
> And this can be scientifically assessed by taking all use cases and refactoring over them.
And I can't count the number of times I've seen applications do this, resulting in a terrible UI that is hard (and therefore slow) to use.