No, you don't really need much white space. Use thick lines to show boundaries. They do the job of visually segregating things without wasting screen real-estate.

White-space became a "thing" because of the increased use of touch-screens, such as iPads and smartphones. The space helps avoid fingering mistakes. However, if the vast majority of users are using a keyboard and mouse, it's an anti-feature. People started copying the trend without thinking about and road-testing the practicalities of it. Unlike the Kardashians, science works.

White space has its purpose - grouping and the principle of proximity comes to mind.

Of course this argument comes down to how much white space is appropriate and we may actually agree, but if you're indicating that the whole idea of white space, and other Gestalt principles, is a fad, I think you're wrong.

I think GP is just saying that white space is not the only way to apply Gestalt principles. There's hardly a clearer indication of which things belong together than surrounding them with a border.

That is true: there are multiple ways to visually split and group things. Lines, space, color, and shadows come to mind; each with trade-offs.

A down-side of space for grouping is that it uses up screen real-estate and can result in the need to scroll more. Scrolling not only takes up user time, but makes it hard to compare information on the screen, because two or more units of data may not be able to be on the screen at the same time.

For example, if you are entering employee info, and you get to the "dependents" section, you may want to know if the dependents have the same last name as the employee as a quick check of applicability. If the employee name is at the top, then it may have scrolled off the screen by the time you get to the dependents entry. In a more compact design, it's more likely both will be on the screen at the same time.

> I saw it again and again: Users memorize hard-coded paths to certain functions - simply changing the name, appearance or position of certain elements along that path was enough to throw them completely off the track.

It seems like that argues against changing it to me, at least in productivity-sensitive applications.

There could be a justification in revamping the interface if there is an expected amount of turnover or the costs of training are significantly higher, of course. Often, though, these are situations where form should follow precedent, not just function.

I'm not sure if I agree. Of ourse it's a decision that has to be weighted and carefully considered, but by avoiding slight and temporary inconveniences at all cost nothing would ever change or progress. Often people don't know what they want or need until they see it and sometimes people have to be forced to abandon the idea of faster horses.