> In this specific case, it should be clear that you can select multiple options from context or by text, or it should be encouraged to try it out (i.e. easy to undo).
It seems like you're saying that because there isn't anything inherent in the visual difference, it is better to remove the difference entirely and force everyone through a less efficient mechanism with higher cognitive overhead.
> If you need to rely on the visual difference of the icon, then you've already lost more than 50% of the people who don't realise this.
Agreed, but how does having a visual difference mean forcing people to rely on it? The visual difference is purely an augmentation, and one that kicks in early enough that it prevents people from forming an incorrect mental model. At least for some percentage of people—you're saying 50% here, I would hazard at least 95% (of people at least a little familiar with using graphical interfaces). And the remaining 5% aren't being left out in the cold, they can always experiment.
> It's really just a property of a list of checkboxes, how many you can select, not an entirely different class of UI component.
I completely disagree, and in fact I think this is probably the fundamental error. You are talking from an implementation point of view. From the point of view of a user, they are being asked to input very different things. They are picking from a set of options, or they are accepting/rejecting each item in a list of things. It includes a distinction between one and multiple, and I hope you don't think those are handled the same way in our brains. ("Sorry, dear, I thought she was one of my wives.... oh, right! You're the only one I have! I guess I forgot again.") The fact that they can both have a superficial manifestation as a list of options makes it more important, not less, to visually distinguish them. Radio buttons have more in common with a single-selection dropdown than they do with a list of checkboxes.
It seems like you're saying that because there isn't anything inherent in the visual difference, it is better to remove the difference entirely and force everyone through a less efficient mechanism with higher cognitive overhead.
> If you need to rely on the visual difference of the icon, then you've already lost more than 50% of the people who don't realise this.
Agreed, but how does having a visual difference mean forcing people to rely on it? The visual difference is purely an augmentation, and one that kicks in early enough that it prevents people from forming an incorrect mental model. At least for some percentage of people—you're saying 50% here, I would hazard at least 95% (of people at least a little familiar with using graphical interfaces). And the remaining 5% aren't being left out in the cold, they can always experiment.
> It's really just a property of a list of checkboxes, how many you can select, not an entirely different class of UI component.
I completely disagree, and in fact I think this is probably the fundamental error. You are talking from an implementation point of view. From the point of view of a user, they are being asked to input very different things. They are picking from a set of options, or they are accepting/rejecting each item in a list of things. It includes a distinction between one and multiple, and I hope you don't think those are handled the same way in our brains. ("Sorry, dear, I thought she was one of my wives.... oh, right! You're the only one I have! I guess I forgot again.") The fact that they can both have a superficial manifestation as a list of options makes it more important, not less, to visually distinguish them. Radio buttons have more in common with a single-selection dropdown than they do with a list of checkboxes.