Join 3,499 readers in helping fund MetaFilter (Hide)


Rethinking Software Look&Feel
April 7, 2007 11:14 PM   Subscribe

Magic Ink - Information Software and the Graphical Interface
posted by Gyan (29 comments total) 21 users marked this as a favorite

 
I looked at some of this and was impressed. Most times when I see a paper/article/blog post by a designer claiming that all previous design has been a mistake I get a bit upset. There are actually some good bits in the section where the author redesigns parts of Amazon and suggests alternate ways to showing movie times.

Still, I feel like I should complain about the 1 link FPP. This is metafilter after all.
posted by pkingdesign at 12:16 AM on April 8, 2007


Er, a summary would be nice, at least.
posted by flotson at 12:41 AM on April 8, 2007


I like the use of the ticks under the amazon star ratings to indicate distribution. I'm not sure how useful it would be to most, but I like knowing whether something gets 3 stars for contentiousness or mediocrity.

He also nails the movie time site redesign.

Curse you Gyan, I'm going to be reading this site for hours.
posted by BrotherCaine at 12:46 AM on April 8, 2007


Hey, add a Tufte / EdwardTufte tag? It looks like this guy references him a bit.
posted by BrotherCaine at 12:48 AM on April 8, 2007


Sorry, I call bull. Interface based 100% on visuals does not automatically equal better interface. I'd much rather choose a city from a drop-down list than click it on a map. Also, what about blind/visually impaired/geographically impaired users? Which is easier, finding St. Louis from a drop-down, or trying to find it among a cloud of cities in the middle of a map?

Just because this guy processes things visually doesn't mean that everyone does. I hate being required to choose a date from a calendar. Just let me type it in, okay?

That being said, it's true that so many websites have lousy, lousy interfaces. But in general I think that airlines do a pretty good job of making an intuitive interface. I personally think Kayak is fantastic.
posted by Deathalicious at 12:51 AM on April 8, 2007 [1 favorite]


Oh, and I love how the solution to current bad design (Amazon) is to put in even more text and make it really, really tiny. Oh, and apparently microcontent can magically come from nowhere...

These drop-down menus are awkward and uninformative. Geographical locations belong on maps, and dates belong on calendars.

Look man, just cuz you're too retarded to read doesn't mean I have to suffer because of it. If I know where Berkely is on a frikken map, then I don't need to see a map to know where I'm moving. But let's say I don't (I'm really, really bad at geography). I know that I have to move to Berkely. I will be driving there on a highway, which has helpful signs in English telling me which way to turn. The moving company also knows vaguely where in California Berkeley is. Neither of us actually has to know its precise location on some kind of map.

How is scanning vertically through an alphabetical list of place names more difficult than scanning vertically AND horizontally through an arbitrarily placed scattering of place names?!? I mean, you have to read the city name one way or another...
posted by Deathalicious at 1:17 AM on April 8, 2007 [1 favorite]


He also nails the movie time site redesign.

No, he doesn't.

Missing from new, improved design:The only aspect of this I agree with is re-ordering the display by movie rather than theatre.

Google does all this and more in its movie listings. Yes, tragically you must scan across a variety of theaters to find a starting time you like. One improvement would be to slightly fade starting times that have already passed.
posted by Deathalicious at 1:25 AM on April 8, 2007 [1 favorite]


Redesigning existing sites/applications is always kind of a cheap shot. There are many, many reasons for a site to stick to a design which, from the point of view of some usability principles appears suboptimal. This is because those principles often compete with each other requiring a balance to be struck.

I still liked reading it though.
posted by juv3nal at 2:22 AM on April 8, 2007


This guy: "I suggest that the design of information software should be approached initially and primarily as a graphic design project."

If his examples are any indication, his idea of "graphic design project" consists of "use unreadably small fonts, because hey, nobody likes text."

The guy's train timetable example is classic. He's taken possibly one of the more readable formats of timetable -- many are not, sometimes to the point of utter bafflement, but his example text is pretty good -- and turned it into some kind of biorhythm chart that not only fails to map to my mental model of train scheduling but requires I distinguish angle, line, and color and to break out a ruler in order to determine something simple like what time to be at a train station.

I agree with many of the points he brings up, but he illustrates them very, very, very, very badly to the point of making a mockery of his underlying ideas. This kind of result is why people like me object to the tyranny of the visual thinker.
posted by majick at 6:41 AM on April 8, 2007 [1 favorite]


The flight-planner thing hurt my brain. So much.
posted by billybunny at 7:03 AM on April 8, 2007


He really could have laid out his ideas more clearly.
posted by The Deej at 8:04 AM on April 8, 2007 [2 favorites]


I could just see an average user struggling with any real world implementation of his ideas in the "Engineering inference from the environment" section.

"There, finally done. 17 pages on why strip clubs should be outlawed."

"Honey, can I use the computer to map out our trip to the park?"

"Sure."

"What the HELL?!"

"It's not what you think! The software is making inferences on my behavior based on what I type!"
posted by saraswati at 8:08 AM on April 8, 2007


The Train Schedule case study has some very nice elements in it that are quite suitable to those who know the area and the map (in this case a metro map, which anyone who uses the metro regularly will already have a good spatial understanding of).

The complaints of dropdown lists are somewhat valid but even then I'd rather have an autocomplete search/input box and have a alphabetical index available if need be. Autocomplete with a bit of AI such as the one found in Quicksilver is the shit when it comes to simple interfaces and efficient lookup/entry of information. I hope it catches on as a trend.
posted by furtive at 8:25 AM on April 8, 2007 [1 favorite]


Sorry, here's the link to Quicksilver.
posted by furtive at 8:26 AM on April 8, 2007


I really like his main idea---distinguishing "information" and "manipulation" software, and avoiding interactivity in the former. On the other hand, the site mixes his idiosyncratic ideas of what constitutes "good design" in with this other really good idea of his that designing information software is really just answering the user's questions as quickly and intuitively as possible. I also hate, just hate the idea of having software "intuit" what you want to see based on context: I like my software to be predictable, dammit! If I want any but the most basic context, I'll supply it.

But he has a point, and I'll certainly keep it in mind when designing software in the future. If an application's job is presenting information, that's what it should be doing. Ask the user to fiddle with knobs only when necessary.
posted by goingonit at 9:54 AM on April 8, 2007


this is cool - thanks. without the FPP i would have missed this article ...
posted by specialk420 at 11:04 AM on April 8, 2007


I immediately see three things about the base assumptions which are not necessarily true:
  1. People intrinsically understand visual displays better/ faster/ etc. than written words. On the BART example, an orange bar with "Richmond" printed across it didn't scream "take the Orange Line towards Richmond" until I saw that Richmond was on the end of the Orange Line.
  2. People have enough spatial knowledge to easily find where they're coming from/going to on a map. If I'm presented with a map of the London Underground (which has more than 300 stations displayed on it,) and all I know is the name of my destination station, that'll take a while. And how does a global airline handle putting every destination on a readable map when you're operating both 45-minute short-haul flights and New York to Bombay?
  3. Point-and-select is faster than the keyboard. I go back to the days when you had to use the command line for everything useful, and still prefer the keyboard shortcuts on graphical programs. One of the few specific pieces of knowledge I retain from my undergrad human factors class is that a good typist is an order of magnitude faster switching between form inputs via the tab key than (s)he is using the mouse.
There are some good points (filling in the blanks in sentences is easier to understand than simply listing the variables, for instance,) but I see so much that I strongly disagree with that it gets in the way.
posted by djlynch at 11:04 AM on April 8, 2007


he should put some of this on that
posted by arialblack at 11:09 AM on April 8, 2007


There are some good points (filling in the blanks in sentences is easier to understand than simply listing the variables, for instance,) but I see so much that I strongly disagree with that it gets in the way.

I do wonder if the Metafilter crowd is really the target audience for these sorts of improvements, though. The audience for this site has grown considerably in the past eight years so clearly it's not so true now, but arguably Metafilter readers are far more likely to be power users than not.

There's a fair amount of HCI material about the difference between power users and novice users, but the key point is that the two groups use software in fundamentally different ways. I used to think everyone used a computer much like I did until I went to university and lived with a bunch of people who spent about as much time on the computer as I did, but didn't know them nearly as well. They did everything with the mouse, for example, even cutting and pasting. This seems horribly inefficient on the surface, but at the same time it's not necessarily ideal to force people to remember lots of hotkeys for all the software they use. Speed isn't everything; people would rather be slow than confused.

That doesn't mean Victor's examples are all good; I get that the constant use of maps may not work for the geographically impaired, and he does make some odd assumptions sometimes (like not telling you where a movie theatre is because "you'd probably just know where it is"—one of my friends, who is perpetually lost in a city she's lived in for years, would beg to differ). But there is a lot of interesting stuff in the article, and certainly his insight that adaptive systems are crucial to good user experiences is one the HCI community has latched onto as well.

I got to the part about the BART widget, thought "okay, this must be his final proof of concept, this is great," and then saw I had more than half the article left to read. This will take some time to absorb fully.
posted by chrominance at 11:23 AM on April 8, 2007


(Also, the web standards person in me is slightly sketched out whenever a site says "this will not print in Firefox/IE! please use this PDF instead!" Like, why won't you print in Firefox/IE?)
posted by chrominance at 11:24 AM on April 8, 2007


(Also, the web standards person in me is slightly sketched out whenever a site says "this will not print in Firefox/IE! please use this PDF instead!" Like, why won't you print in Firefox/IE?)

Yeah, I wish they'd just suck it up and make sites that will print with the programs that are designed to display them, instead of pushing a pdf on the user because it preserves their precious spacing or whatever.

Although, on (print) preview, it seems that it will print with Opera, and without anything weird.
posted by Many bubbles at 1:28 PM on April 8, 2007


There are some good points (filling in the blanks in sentences is easier to understand than simply listing the variables, for instance,) but I see so much that I strongly disagree with that it gets in the way.

I do wonder if the Metafilter crowd is really the target audience.... arguably Metafilter readers are far more likely to be power users than not.

There's a fair amount of HCI material about the difference between power users and novice users, but the key point is that the two groups use software in fundamentally different ways.


I'll agree that power users interact differently from the rest of us; my objections are mostly that it doesn't take into account that different people have different preferences when viewing and/or entering information. That's not an easy task, and I'm guilty of forgetting to do it myself.

For instance, the web has evolved a fairly standard way of entering dates — a set of menu boxes, with a clickable calendar displayed adjacent to it. People who prefer the mouse can do it either through the menu boxes or the calendar, and I can zip through with the keyboard without having to look through a grid or list of numbers for the right day of the month.

It's that kind of thing — not making people search visually if they don't want/have to — that's missing from the designs I see on there.
posted by djlynch at 2:16 PM on April 8, 2007


For instance, the web has evolved a fairly standard way of entering dates.

"fairly standard" != (rational || useful || !sucky)

in fact, quite often it = sucky.
posted by signal at 7:00 PM on April 8, 2007


What I take issue with in the flight schedule redesign is that he has shown a graph with all the text data superimposed on the "graphic " element. You shouldn't need both if your graph is well designed and easy to read. Rearranging textual information on top of a bunch of colored bubbles is entirely missing the point of graphic communication.
posted by oneirodynia at 8:41 PM on April 8, 2007


Why can't I use something like Quicksilver within programs? There are plenty of functions that I know exist within MS Office, know the names of, but don't know exactly where to dig for.
posted by Tlogmer at 10:36 PM on April 8, 2007


You can use Quicksilver in programs. Indeed, this was discussed on either 43Folders or LifeHacker within the past two weeks. You can assign shortcuts to program commands.
posted by five fresh fish at 11:27 PM on April 8, 2007


Annoyance: That the perfectly sensible YYMMDD system is not a worldwide, long-held standard. It possesses the singular advantage of being date-ordered when sorted in ascending numerical order. In light of that advantage, no other arrangement makes the least amount of sense.

Yet it mostly isn't used. Instead we get a hodgepodge of MMDDYY and DDMMYY. Just plain dumb.
posted by five fresh fish at 11:37 PM on April 8, 2007


I like his BART schedule desktop widget because I'm visually oriented, and I think the nature of a widget and the nature of a commuter train schedule both lend themselves especially well to some of his assumptions about what users want and what their regular habits are. I would caution him against using that design strategy for every data-related problem... we don't live in a one size fits all world.

His model especially breaks down in a transactional situation -- like his southwest plane tickets example -- but he makes a strong point (unintentionally?) about how we, as an industry, have over-specialized and pigeon-holed the skillsets of interaction design (just make it work) and visual design (just make it pretty), and somehow lost the art of information design in the chaos.
posted by nadise at 12:53 AM on April 9, 2007 [1 favorite]


five fresh fish, I use YYYYMMDD a lot of the time.
posted by BrotherCaine at 3:33 AM on April 10, 2007


« Older The volcano Piton de la Fournaise on the island of...  |  The National Music Museum... Newer »


This thread has been archived and is closed to new comments