Showing posts with label interface design. Show all posts
Showing posts with label interface design. Show all posts

Thursday, September 22, 2011

Tiles, Widgets and Icons

Ever since I read Itai Vonshak's tweet about John Dvorak's article on "the serious flaw" with Windows 8, I've been vaguely annoyed by this misunderstanding of how an interface should or could work. Windows 8. Like five of these are icons. The rest are little bits of data. Sorry they are squares, but they are not icons. Windows 8 is, conceptually, at the presentation level, an offshoot of concepts we've first seen in Windows Phone. And Windows Phone (like everything else in the world) has been evaluated (by tech columnists) as a straight-up iPhone competitor. I gather that Dvorak thinks some people sat in a room in Redmond and went "I've got it! We'll make all our icons square, and blue! That'll differentiate it from that darned iPhone. We'll show those meddling kids!" I don't work for Microsoft, and didn't do any work on any of this UI. In fact, before we go further, I don't own one of these, and don't particularly have a warm place in my heart for Microsoft. But fair is fair, and I'm judging this on it's merits, as I see them. As soon as Windows Phone was announced (as I recall, even before we saw images) it was pretty clear that they were taking a quite different, not just putting their spin or skin on the "compete with Apple." I think it's a rather good approach. If you design the system (any system, of any size) to be entirely about running applications, then your application discovery method better be solid. When Dvorak talks about how he recognizes icons, or accuses Microsoft of all but switching to a DOS CLI, this reveals his misunderstanding of how mobile (and some other clever systems) can work. Instead of finding and running apps, you can be served information immediately and contextually. Skipping the bigger discussions of push messaging and contextual surfacing, Windows Phone (and I presume from the images, Windows 8) tiles are widgets (ask me later why this links to "icon"). Widgets, in this context, present little snippets of data, or even useful bits of data next to other bits of data. Immediately. Without having to explicitly launch them. This is Dan's Windows Phone, just as he normally uses it. I asked. He didn't mind me taking a photo, but this data is always in front of him. A life ticker. Aside from generally observing everything I can about mobile, I have a Windows Phone user right next to me at work. I asked him today to demonstrate, and he demonstrated exactly this use of it.
I can just open it up, and look at it, and scroll to get more information. If I need to know who did that, I look to the side and there they are.
And no, he's not a designer, or particularly a mobile guy, and I didn't set his phone up for him. I've never even touched it. Based on many other observations I have made of all sorts of mobile users, a lot of them use their devices this way. You don't drill into the app, or even launch it. You just un-sleep the screen and glance at it. Maybe, just maybe, you have to scroll a little ways. Or, I claim from all these, they want to. I still feel the same way I did about Widgets in 2007 , and increasingly think they are not just a good thing to have available, but a key attribute, and may become (like Windows is now pushing towards) the one best way to get to information. There are other platforms that have widgets. My Android handset is almost entirely widgets, using icons only when there is no good widget. There are related ways to work in this sort of mentality, like WebOS which leaves the application as a sort of widget/tile on the desktop; the set of them changes moment to moment, but they aren't just launched and then disappear when you go to launch another application. Not perfect (or even much alive) but another approach, certainly. I'd talk about Symbian, but that would stretch my "near-death OS examples" to the breaking point. My Droid2Global, with a frankly pretty boring screen of widgets. But still, I can swipe, glance and tell time in a couple places when brain addled. Why swipe, find, click, wait then read? WebOS -- Here, the TouchPad -- minimizes everything running to the desktop as readable tiles. No icon with a marker, then long-press to see what's running... Just glance, and go straight to the one you want. Forcing users to scan a grid of icons is, to me, not just a failure on mobile, but is a demonstrably secondary action in desktop systems. Think of the Applications folder on OSX, or the All Programs link in Windows. This is not the high use case, and is a simple fallback for when you need to get to something obscure, or which you cannot find otherwise. And how to get around this? Search, launch bars, recently-used application lists, and other methods to allow the user to find the most likely applications. With a distinct move towards doing this automatically for users, instead of assuming they will set it up for themselves (because they won't). Someone will probably see this as an assault on iOS. And I am fine with that. Aside from being comfortable with fanbois interpreting everything non-glowing as villainous, this is behavior I have observed; satisfaction does not always correspond to task completion, much less efficiency. And it's not evolutionary. This is a twisting path. A lot of stuff much older than this let you prioritize better than just "what is on the first page of the app listing," like say S60: S60e3 (an N95) with quick access icons, and a quick access button to access an on-screen widget! Darn it, I did mention Symbian after all. Oh, well. But since we're comparing to Apple, let's talk about another common refrain against Windows Phone – that it's over-designed. In the sense that it's designed to look good to executives approving it, to focus groups, or is just a misguided visual designer's wet dream. There are a number of ways in which I disagree with this, and find it to clearly have been well thought out by some pretty clever UX designers and IxDs. The first has even been pointed out by some who dislike it. The device has pretty notable, integrated teasers of additional function or content. Tiles are partly displayed on the edges, animations help communicate what the interface can do. Even that big dead space to the side is about leading you to change panels to the right. Another key one (there is lots of hate) is about the density of information. Those big words bug everyone, because they are not efficient. Compare the number of characters, lines of info or anything else you want to almost any other device and it comes out behind. Or does it? When I think of a lot of the use of mobile, even though context doesn't matter anymore, I see a lot of glanceable use. My co-worker, for example, will regularly use his phone on the way to a meeting. Large text and images as iconic representations helps him use this on the go. I have noted, a lot more than any of the devices I am using. I sometimes have to stop to read where the room is, on my high-density information display. While we're at it, those giant, wasteful black gutters between tiles, they seem to help reduce click errors (and improve confidence of clicking) in all environments. I have seen the design of Windows Phone disparagingly described as being like poster art. Which I say is a good thing. Posters work.
A question I assume many of you will ask is: Why isn't Windows Phone the market share winner? Well, lots of the usual reasons. Marketing, price, distribution, hardware, networks and operators, and so on. By no means does the best device win. Everyone seems to now be comfortable with OSX devices selling in their non-dominant market, so think about that. I'll tell you that a lot of these relatively niche devices (like WebOS or Windows Phone) are beloved by most users of them that I know, or have interviewed for research. They give them up due to unavailability, or lack of key apps or some other aspect outside the core interactive design paradigm. But also, we may be getting there. Android is doing rather well, and if you were inclined to, you could say it's because they support widgets, and this flexibility of layout is at least part of the success. I actually cannot find a useful "reason I chose Android" study to say that, and it's possibly untrue, but everyone else lies about statistics to prove their point. So I say that Android wins because the idle screen is more flexible and can be more mobile-optimized (I actually /do/ talk to end users who say this is why they chose it). And even more importantly, tomorrow is the future. Mobile, more than other technologies, changes all the time. And often in unexpected ways. On Thursday, a new head for HP. Is WebOS on the way back? Or this recent InfoWeek survey of mobile OSs (available at their site, but thru a subscription wall so the link may or may not be good) that shows Windows Phone perceived as just a tidge less useful to the IT professional than iOS. The future doesn't even have to be a variation on what we have today. There are OSs on the market off in other corners of the world you might not have heard of. Speaking of Itai, his Else phone prototype was a shockingly amazing piece of design in every way; I see no reason something that radical cannot make it to market sometime and surprise us all. Or maybe the web will take over, and the device OS won't matter. Or… something else. I have no idea what tomorrow will really bring. But it won't be more grids of icons.
For more on designing widgets, or just making those bog-standard lists and grids the best they can be, check out our forthcoming book Designing Mobile Interfaces. Pre-order from Amazon for a significant discount, or read the content online right now.

Wednesday, September 14, 2011

Interactive Criticism and the Cult of Clean Design

This discussion of minimalist posters got me thinking of some of the design problems I have been encountering lately. And that got me to thinking that I have been seeing a variation of this for years and years.

 Unlike much of the graphic arts scene, it seems that interactive design hasn't gotten over it's trend of simplicity and whitespace for their own sake. If anything it seems to get worse year after year. While I won't quite blame Apple – and everyone else who follows that design ethos of spartan minimalism in everything – this is the design I am talking about.

 But the problem isn't really that we're stuck in a design rut. We are, but it's much more that not everyone evaluates everything the same way, with the same depth. People use a cool new product (on their iPad), then come to work the next day and say we should make a product as clean and easy to use as this.

 But "clean" and "easy to use" are different things. In the same way that "look and feel" are two different phrases (hence the "and"). "Clean" rapidly becomes misinterpreted as "low-featured." "Easy to use" becomes misinterpreted as "easy for every single person in the world to use." "Simple" turns into "simplistic," and we cut features not to get to core principles, but because they are visible.

We cut features not with a knife, but with an axe. And rapidly, any feature is worthy of being assailed not on it's merits, but on it's immediate visual appeal in a wireframe, or comp. But it's a false argument. Anything, – anything – can be assailed as cluttered, complex, or any number of adjectives that are, really, meaningless. There is no way to argue against "you don't want to confuse the user..." on it's merits, because there is no internal logic to this argument; it has no merits, really.

I routinely want to challenge such arguments and bring us back to principles, or to the scientific underpinnings of the practice. "Hard for which users?" or "But this feature is required to meet objective 2…" or "But the research proved that 100% of the participants prefer…" So maybe we can win the argument by counter-attacking their premise, and getting to the core of the issue like this. At least on the face of it.

Because really, we're missing a key point by calling it a design issue. It's a communications issue, or maybe (as Alexandra Lange responded in the comments on that Design Obsever article), an issue of design criticism. I admit that I would not have made any of these connections without her article. As practitioners of IxD, UI and UX, we try to make decisions based on provable knowledge. You could even characterize it as a scientific approach to design.

But it's not always true. Opinions still abound in discussions amongst ourselves. About interpreting what actually happened in a test, or what situation we are really looking at, so what heuristic should be applied. If you were to watch a design (or art) critique session, you might think it's just a conversation, but really this is a fairly formal construct, with rules that make it work to everyone's advantage.

You cannot complain about a work without merit; "I don't like it" doesn't fly. You better have a good, reasoned argument as to what you don't like, in what manner you don't like it, and maybe even suggest a solution. It better address the intended audience, or the purpose of the object. Assuming use outside of this may be irrelevant.

 As the creator, you better accept all criticism, and I don't mean passively; you have to engage in the conversation, to make sure everyone understands each other, and so group discussions can help reach consensus on the best course of action. Note that I didn't say "solution." One individual still ends up cutting the wood, or painting the line, or placing the pixel. Iterative critique gets it closer to the intended goals. Mostly, you cannot perform design criticism alone, or as the only proper practitioner of it; everyone has to participate and follow the rules. 

 Interactive design doesn't have any of this. Sure, your studio might work fine. And I have absolutely been on teams – or led teams – that worked exactly like this and had terrific group design critiques (and terrific products came out the other end). But as a whole discipline we do not have a method of critique. We argue, cajole, express opinion, refer to previous solutions, and our favorite products. We make deals, and end up splitting the difference so we can stop arguing and get to the next point. We decry ornament (especially when we bow to the altar of Clean Design) and insist we evaluate based on interactivity and process.

All while we draw comps and complain about crowding objects, and demand white space be added or everything align to the grid. We carry out common practice, and build what is easy, instead of finding and defining what is truly, demonstrably best practice. We do not share, or tell anyone else about our design solutions. Maybe not even the guy in the next pod, but almost never to the design community as a whole. We design in isolation, take criticism as insults to our work, and believe there is design that cannot be improved upon. We perceive openness to opinion, fuzzy concepts or acceptance of change as weakness. We are simply confused (and maybe assume it's a trick or insult) when a designer offers to change to an alternative design when a good point is raised by others. And we cannot fathom why our clients do not respect interactive design and user experience as evidence-based, consistently-applied, professional fields, and express sorrow every time they insist we change something based on whim.

 We need an ethos of design criticism. Not a process. The existing ones for fine art and design work fine. And we must not stop disagreeing with each other. We must not stifle creativity, or reduce iteration, or work more in isolation from fear or as a res Of course you know that arguing with the client never really works. I said that up above because that's what I want to do, but another approach is needed.

 So, if you agree, what next? Well, it's simple. Do this. Tomorrow. Okay, it's the end of the week. So, Monday. The next time you bring the team together, lay out the ground rules and tell everyone this is how we're going to talk to each other. In several ways – but especially in becoming a seasoned practitioner of art and design criticism – going to art school was maybe the best education I could have gotten. I've been thinking for a long time we need to teach not just what you should know, but how to do it and how to do it with a team. Not every field of study or school does this. You can make sure to teach your junior designers the skills they need, and not just assume they will pick it up on the way. And you can try to get the right feedback from everyone.

When you walk to someone's desk, and they give feedback, ask them to justify it, very precisely. Nicely, but get to the heart of it. Anyone can answer these questions if pushed to it. Not naturally, but they can. Even your clients. If you are told "why not more like how Amazon does it?" then pull up that part of Amazon, and ask in what manner this solves the need. Ask how this meets the goals of the product, ask questions you know that there are answers to, or which they will enjoy answering. As much as it's been talked about, there is no certifying organization for IxD.

As much as some people get to write books (now me!) and get followed by thousands of people every day, there is no one guru who can change the whole field. We all do it, every day, as we work with each other and improve interactive products. If you like this idea, do it. If you have a better one, do that. But if anything ever bothers you about how our field works, talk about it, find a solution, and put it into practice. This is how we improve ourselves, and our little part of the world.

Thursday, April 24, 2008

"Lo"? What the !@#$ does that mean?!"

this is the promo shot, not actually my thermostat
One of those standard things I think all homes should have is a smart thermostat. We bought this one shortly after moving in. Though I agonized trying to find a reasonably easy-to-use one that fit the space, I failed. Its sort of junky. Do not buy it.

One thing is that its sort of hard to manually control. There is no real "fan-only" setting, for example. You end up working around the software a lot, to trick it into things.

Lately, its been worse. Its performing uncommanded actions. You set it to manual, click over to the air conditioner (which is still "unplugged" so will not engage) and if you wait a bit you can see it go "nope, you need heat." The little heat-mode icon lights up, and then the wavy lines that mean its making heat light up. Go to the basement, and yup, the heater is on. On a 76° day. One other hint, the temp display has a tendency to say "LO" instead of a number.

So, I finally give up with shorting leads and beating it with a screwdriver, and call the tech support line. Turns out LO means off scale low. The tiny, stupid computer in the thermostat thinks that its freezing cold. But its not, its just a bad temp sensor. Apparently, it feeds signal directly as a resistive signal, and 0° is about 0 volts.

Okay, its not the computer's fault. Its the designers. While its nice it says (cryptically) off-scale low, instead of a falsely low temperature when it gets a zero voltage response, who decided that the thermostat is smarter than the user? I am requesting "cooling" mode, so why switch to heat mode all by itself? Or, how about a spurious data sensor? It seems to disregard the 95% of the time it gets valid temperatures in order to fire some emergency recovery mode if it gets ANY zero-voltage temp gauge readings. This is poor design, making me sweat and wield a screwdriver needlessly.

Saturday, February 23, 2008

Terrible Interface of the Week: Redbox

The wife, a friend and I are hanging out at our house this evening, and after watching all the Dirty Jobs on the PVR, and everyone else decided that they hated the Netflix I have, and didn't want to watch my Thai homage to 50s Technicolor. Plus, we needed ice cream. So, off to the Hy-Vee. And the video department is being cleared out. Now there's a Redbox off to the side: Besides being the only rental solution at the store, its something I had never actually used before so I couldn't resist. Home deck is not bad. Big, obvious choices with no gimmickry. So I click the obvious one, the "Rent a DVD" option. The expected screen comes up, with a list of movies... ...and that's it. There's no overall count of movies available, location in the list, search box... wait. There's no search box? Really, they seem to think that the limited selection (I didn't count, but like 50 movies) means they think browsing is enough. But its really no good. And the buttons are so generic, like any bad PINpad. It took me a good 15 seconds to figure out there was a 'next' button. If it was on the side, or had a big arrow maybe it would work. There is a minor ability to sort, but its not that great. There are two small tabs at the top to sort by release date or title (I guess, alpha, the default). They are small, the color highlight is vague (I got it, but a friend read it backwards) and they are hard to hit. You can get above it, so the cursor appears to be on it, but its not activated; that's a serious failure of basic design principles. The tabs are hard to hit at that target size due to the bezel getting in the way and casting a shadow, and the parallax error. Touch alignment isn't perfect. The scrollbar (used a few other places) is just as bad in the same ways. Perhaps my favorite is when you press the "Online Rental Pickup" option: There are clear, animated instructions. Unless you don't want that option. At which point you have to wait for timeout. There is NO back button of any sort. Nothing to get you anywhere else, to a help screen or anything that is available in the rest of the application. You'll note I don't comment on the payment, delivery and return processes. That's because we didn't rent anything. As I said, we were somewhat in the mood to get something to watch, but the inability to actually find anything but top-40 playlist items from the Redbox meant we didn't begin to consider getting anything from there. We actually spent time looking thru the clearance bin (from closing out the store video department) instead. Even not finding something worth buying, that was a more satisfying browsing experience. Aside from the selection, there may be something satisfyingly social about browsing movies on a scale larger than touchscreen. Have to think about that some more.

Wednesday, February 13, 2008

On washing machine hacks

This is our washing rock: I have come to realize that, in my time living on my own, I only learned fairly perfunctory washing procedures. There is much more to it if you really care. And like practically all systems used by folks who have to accomplish a specific task with their system, hack emerge. A favorite of mine is the washing rock. I have seen similar items other places, and not always from people who actually talked to each other previously. This particular hack involves the spin cycle. For reasons that elude me, its nice to be able to soak and agitate clothes twice. But its either unimportant or actually bad (I don't know) to spin between the two soaking cycles. The machine doesn't mind soaking and agitating with the lid open, but the spin cycle is apparently dangerous, so it will not start -- and if running will immediately stop -- if the lid is opened. So, the rock goes into a hole where some lame dispenser goes, and it keeps the lid closed enough bugs don't fall in, but open enough it won't start the spin cycle. While "researching" for this post, I asked about a few other's habits, and found something else interesting. Which, again, the wife with her great washing knowledge, already knew. Washing machines by no means always have the same UI. This struck me as quite pointless; they all do almost the same thing (I guess some of the more electronic types have additional smarty-pants features) so why have they not settled into one closely-related set of behaviors and interfaces. Here's my favorite example of how divergent they can be. In the spin cycle mode, one machine locks the lid. Like a stove on the cleaning cycle, you cannot open the lid at all. There is in fact no way to cancel the spin cycle. On a power outage once, their clothes were stuck in the machine until the power came back on. How does this make sense, and why would you expect this behavior based on use of other washing machines?

Friday, January 4, 2008

A different GUI evaluation

This training organization I work with has, at astonishing expense, acquired forty sets of MILES. For those too bored to follow the link, that's the laser field training system the U.S. military (and increasingly, others) use so they can shoot each other without bullets. Its a complex, multi-part system, with control units and setup screens. We don't even have all the components (there are desktop computers that can track and record everyone's movement). The laser is emitted from this device that is strapped to the barrel of the rifle: It has a pretty transparent user interface (it shoots when you pull the trigger and a blank fires from the normal gun, in the normal way), and no GUI, so we'll ignore that, but there are three things that do have buttons and screens. Some are quite odd, so I thought I'd review them briefly. Starting with the simple one, everyone with a gun (and anyone else like a civilian or hostage) has a harness, worn over your uniform and other gear, and a "halo," strapped over your hat or helmet. The black dome things are the laser receivers. there are some on the chest harness also. The halo is dumb, and simply sends signal to the chest harness with an induction loop, so there are no cables. The chest harness has a little computer that interprets the signals, and does the appropriate thing based on if it got hit or near-missed. It beeps when you are dead, and kills the weapon so you can't shoot others. The computer has a little control interface on it. I only have this one terrible shot, and its boring, so I am not gonna review it much. It works like most of the simple menu driven interfaces, like your basic printer or something. Scroll arrows, enter key to execute functions. The "i" key is supposed to be pressed to turn it on, but its not needed; any scroll turns on the display also (to preserve battery, all these items are very inclined to go to sleep). It has lots of unlabeled functions. Tapping any key will silence an alarm. Holding enter down for 2 seconds toggles backlight. Holding both arrows at once initiates a count to turn on an alarm so controllers can find you. Like if you fall off a cliff. The CD/TDTD (don't worry about it, just call it a "god gun." Everyone else does.) is a free-standing device used by administrators when setting up the devices, and by observer/controllers (O/Cs) during the exercise. The system is modal, and softkey driven. There are a few obvious keys, like power and light (backlight), and the second column from the left are the three modes of operation. These are pretty clear, especially the Run mode, which is that used during exercises. Which is good, because it needs to be used rapidly, outdoors, in all weather, etc. The right side buttons are more or less normal softkeys. They vary between modal switches (just selection) and drilling down into sub-menus. In the Run mode, they are just selections (highlighted to indicate). Then you press the trigger to activate the function selected. Except they have one weird behavior, even in Run mode. See the parenthetical phrase (Cntrl Mode) after the NEAR MISS label in the image above? If you press the softkey again, you get that other function: Not sure this is a good way to do things. Its a way to multiply your functions, and they are often related to each other, but its got poor affordance. And I have never seen it anywhere else. I assumed that the parenthetical was another name for it, or something. A slash on the button, with a paren under it might be able to work: /() There are some other nice things tossed in there subtly. The Run mode items are all flush left. All other items are indented a little bit. At a glance the O/C can tell if they are in Run mode or accidentally in some administrative mode, without reading the actual words. The trigger is more or less the OK/Enter key. This makes sense, as usually a function execution involves sending or receiving codes through the front of the gun; its gun-like, so the trigger works great for this. When fired, lights come on the back to indicate it, and warn that you are emitting a laser (its marginally dangerous to eyes at under 10 meters). Unfortunately, the labelling is poor; the bottom line where it says "TR= whatever" is the trigger function. Sometimes its in code due to space and is hard to understand unless you are trained a lot. Oddly, the "OK" key is essentially a Back button. In most situations, it will send you back up one menu. The softkeys occasionally change to be other things, like scroll keys, as here when used to edit a number. Reasonably self-explanatory, but since its a behavioral change its a bit jarring. The main laser emitter is not the prod in the front, but a large, flush area below it. This is a problem for some testing; people who use these know that the frontmost part of a gun is the muzzle. In this case, that prod is a less-used IR data port. The last device we'll look at today is the ASAAF. This one we call "the ASAAF." No nickname, sadly. Its a massively fragile, fiddly device, and expensive as hell to boot. The Alignment Head clamps to the laser emitter on the rifle, and the Display Assembly is the control unit. Permanent cables tie the two together in a sorta clunky manner. To zero the lasers to the gun you are telling the unit what gun you have, and then a red aiming target appears. You press buttons to move it till its lined up with the sights on the rifle. Speaking of buttons, here they are: It seems that the left side array is a typical 5-way pad. I love seeing these made out of membrane switches. Then I try to use them. Here is as bad as it can be. You have to use it without really looking, as your eyes are occupied with the sights. Its quite hard to use. For other functions it's used a lot like a normal scrolling menu, like to pick the weapon system from a list. But its not a 5-way pad. The center Aim button is in NO way OK/enter. Its one dedicated function. Just like Align is a dedicated function. Pressing either of these makes actions happen to the mounted laser device, immediately, regardless of what else was happening. As far as the other two, Weapon Type pulls up that scrolling menu, and User is a setup screen, a heirarchical menu of options. There is no enter button, in fact. Selecting something makes it true. You just go to the next function that needed the value. This is not just weird and hard to understand, but its worse because its different than the other devices. In fact, that's my biggest gripe about the whole system. Not that any one system is hard to use, or requires training to really understand, but that there is no one UI paradigm. Each device is totally different from any other device.