Comment by pketh | Hacker News Reader

Comment by pketh | Hacker News Reader

Parent Full thread

> I do like the idea of using colors to indicate function, but I would be concerned that they draw attention to themselves > with colored buttons and doodads everywhere

I agree, functional colors are not a solution that work for everyone, and also requires a lot of diligence and restraint to not turn into a birthday cake mess (something harder for large corps like reddit)

> really struggle with the sort of tiny buttons shown in this GUI

It's hard to develop a UI that works for touch and doesn't look clumsy when you've got a mouse. The solution we use here is to visually size the buttons for desktop, but have an invisible padding around them be clickable

> Returning immediately can feel flimsy and unreliable, like "did it really happen?"

The way we address this is to show your comment appended to the comments list, or to show your post added to your journal. In this case we're optimizing for the joy that comes from the immediate feedback of I pushed submit and now my content is real.

There's def room for more fun florishes in some apps like confetti etc when you do something, but those are a better fit for games or for rarer interactions imo (ideally leaving comments is a common action that some will do multiple times a day)

Using colors to convey state/semantics is also a big accessibility no-no, fyi.

If it's only way you convey it.

Using colors to further emphasize state or semantics is pretty great. Traffic lights is a good model of how that can work.

> If it's only way you convey it.

Exactly. That's why your traffic lights example is puzzling to me? They're a notorious example of a case of something that can very easily be (and often is) ambiguous.

How is a traffic light ambiguous? The stop and go lights are physically distinct lamps. The top light means stop, the bottom light means go.

It's not obvious that the position is meaningful or what meaning it has. That's why if you look at the history of traffic lights you'll see many variations, including pictograms to complement the colors (e.g. character walking / standing still, "STOP" "GO" text, etc. etc.).

You can see that stop means to stop by looking at what happens when the light turns read.

It's not obvious that "STOP" means to stop either, or a "walking" character means to walk if you don't speak the language or are unfamiliar with cartoons. When you learned the language, you connected the meaning to the word, because you heard it in connection to someone stopping, so now the word stop is a metaphor for that time someone stopped.

This is exactly the same way you assign meaning to any sign, including a stoplight. That's just how semiotics works, all language is metaphorical, no sign can carry meaning without a context.

That's 100% true in the abstract sense that meaning is conventional/contextual, but that doesn't mean all conventions are equally intuitive.

The rule "Top means stop, bottom means go" is specific to traffic light context. The meaning of the word "Stop" is much broader as it is used in more contexts. (Of course, some people can't read, which creates a bunch of other problems).

Maybe a better way to illustrate my point: I think traffic light are better at highlighting the problem of how to properly do semantic redundancy than at actually solving that problem (and that this is in part because colors are not accessible).

FYI, I'm not some kind of accessibility zealot, I think it's just an interesting design problem.

Yup and in this case everything still has icons and labels. The color here is more akin to optional progressive enhancement

But the icons are for individual actions rather than the semantic grouping the colors convey, no?

Yes and those actions can be understood independently of their grouping

Accessibility should rely on an alternative adapted representation, not justify a lowest common denominator approach where everybody loses.

Of course, the point wasn't that the colors should be removed entirely, only that they're only a partial solution.

Reply on news.ycombinator.com