To the New York Times Reader designers,
The Reader is exciting! I think that it's the most significant step that newspapers have taken on the web, since the building of the first newspaper websites. Congratulations.
However, I'm concerned about one feature of the interface: its confusing spatial logic. The Reader is already a spatialized way of looking at a collection of documents. And it already has a definite logic of the spatial relations between various documents. But the spatial system the designers have chosen is, unfortunately, hard to intuit and hard to rely on, because it violates conventional spatial reality; it violates user expectation; and it seems not entirely consistent within itself, at least logically. With some simple changes to the way that pages appear to be connected to one another, you could improve the Reader greatly.
There are two sources of spatial feedback built into the interface you have: which direction a page scrolls in from during the animated page transitions; and which arrow key direction (or arrow button direction) the user applies. These two spatial determinants combine -- or might -- to give a user a gut-level sense of "where they are". (Or, they interfere, or and the user doesn't have that sense.)
To walk through how the current spatial logic works (or doesn't), take the experience of reading a multi-page news article. The software seems to encourage people to use the arrow keys to navigate in this case, since you made the "page down" arrows so small and hard to hit with the mouse. That's a reasonable design choice. The DOWN arrow takes me farther "down" into the article.
(Personally, I would suggest that the article extend to the RIGHT instead (perhaps even pulling in *one column at a time* as it did so, instead of flipping whole pages at a time -- conserving the user's visual context at all times. Please try out the Macintosh text-reader called "Tofu" to see what I mean. http://amarsagoo.info/tofu/index.shtml))
However, your designers chose DOWN, so down it is. At least the direction of movement matches the arrow key, thus far.
Once I reach the end of article A, pressing the DOWN key takes me to the next article, article B. Firstly, that's somewhat unexpected, since the down icon was dimmed. Secondly, despite my pressing the DOWN arrow, the article moves in from the RIGHT, as though I had pressed the right arrow. Spatial factor 1 (animation) didn't match factor 2 (arrow key).
If the article had scrolled in from the bottom, I might think of myself as continuing to go "down" the stack of articles. But instead I need to think of myself as going "right", across a row of articles. Evidence for this is that if I press the LEFT key, I go back to (the first page of) article A. So article A is "to the left of" article B, in your layout. (What is the relation of these two articles to their section header page? As it turns out, I have to go from article B, through article A, to get back to the section header -- which is too bad. If the section header was "above" all its articles, then pressing UP from the first page of any article would get me back there.)
Back at article B, though. If instead of pressing LEFT, I press the UP key (the reverse of the DOWN key that brought me from the end of article A to the start of article B), I go back to . . . the start of article A. Not the end. But wouldn't it make more sense to go back to the end? All I'm trying to do is to reverse the action I took -- I went DOWN, now I want to go UP -- but that didn't get me back to where I was. That's unintuitive.
(Also, the page scrolled in from the left; consistent with your layout, but disobedient of my directional input.) If I really do want to go back to the last page of article A, I now have to scroll forward through the whole article again. (Or pick up the mouse and click the Back button.)
It may sound like a lot of small details, but I think these details add up, and either support users' spatial intuitions, or muddle those intuitions. The principle of coherency would suggest at least the following rules of engagement:
1. Actions should be reversible. The opposite of left should be right; the opposite of up should be down, like in the real world.
2. The direction of animation (spatial factor 1) should always match the direction the user asked to go in (spatial factor 2). To support users' spatial intuitions, we should avoid surprising them about where they will end up when they do something.
It would be better to temporarily disable a key than to map it to a non-intuitive action. (You do this on the front page, in fact; the UP arrow does nothing there, which is good. Unfortunately, on that page, the left arrow animates the page over to . . . itself! Why?)
For coherency, the "back" and "forward" arrows suffer some of the same problems as the physical arrow keys, because they seem to suggest a spatial direction -- forward and back. (Back always points left.) When I press the "Forward" arrow, the new page slides in from the right, which does match intuition. But when I press the "Back" arrow, which points LEFT, the page again slides in from the RIGHT -- exactly the opposite of expectation. Maybe the "back" button should follow the user's trail -- through the space of documents -- exactly according to the path they followed, whether they went up, down, left, or right. That would be a powerful wayfinding device that I think even the most casual users would notice and appreciate. Perhaps the button icon could point in the direction that it would actually take the user. The words "retrace steps" or "go back" would have to be written underneath the button. The forward button might get similar treatment.
By releasing the Reader as a version of the New York Times, you are trying to (continue to) transition readers away from a physical object which has, among other things, strong spatial logic (a product of careful human design) and perfect spatial consistency (a product of physics). To win over users who are not computer experts, if you give users a spatial design, you should make sure you don't betray users' built-in spatial intuitions, which a real newspaper never does. The fact that software frequently undermines users is one of the reasons many people don't understand, for example, their own file system. (Please check out John Siracusa's critiques of the current Macintosh operating system interface on arstechnica.com for detailed arguments in favor of (consistently) spatial software. http://arstechnica.com/articles/paedia/finder.ars/3)
If you were inspired to test spatial effects (and are blessed with testing staff and budget), a couple of simple experiments could reveal users' intuitions about the interface (or lack thereof). Have study participants read the news at their own pace and discrection, but interrupt them at various points, to ask them "If you press the right arrow {or back button, or space bar, or whatever} right now, what do you think you will see?" Measure how often users are wrong. You might also try think-aloud studies, where users talk about their browsing intentions. Try to get them to say aloud, "I'm going to go to the next article now", or, "I'm going to go back to the front page now". See exactly how many mistakes they make; how often do they do what they're trying to do on the first try? My claim is that in the current setup, there will be many mistakes.
I personally think a good spatial setup would be if articles were purely horizontal -- they extended to the right, instead of downwards. The right arrow takes you through an article to its end. Then it doesn't do anything -- it's disabled. The next article is always "below" the one you're reading now. The down arrow takes you to that "next" article, which slides in from the bottom. When you get to the bottom of a stack of articles, the down arrow doesn't do anything -- it's disabled. Going LEFT from the first page of an article -- any article -- takes you to the section heading. (Perhaps there's a special indicator on the left hand edge of the screen that makes this clear.) I think this would be an arrangement (one of many possible) that users would intuitively grasp, even if they did not consciously ascertain -- and this could be seen in the results of the experiment outlined above.
But barring that suggestion and sticking closer to your current arrangement, I contend that simply imposing some order and consistency on what you already have would still yield measurable benefits.
sincerely,
BLWH
P.S. Other notes:
1. The Macintosh user community would have been a natural fit for this application -- please don't leave them out in the cold entirely. I realize the appeal of developing this application with WPF, but if the downside is the total exclusion of a prime user base, the price is too high.
2. Are you getting any positive feedback on the "Show What's Read" widget? I would be surprised if it turns out to be valuable. Having to exert so much fine motor control over the mouse to hover over such small targets, just to read a handful of headlines (one per tiny target) seems like meager benefit for such (relatively) high UI effort. Couldn't you simply write the headlines in a small font and pack them together somehow? Then the UI effort would be zero, and information density would go up. If I want to read one, I move my mouse to only one tiny target: the one that I want.
