“That’s how it is with people. Nobody cares how it works, as long as it works.”
- Councillor Hamann to Neo
On Usability, Technology and Photography
“That’s how it is with people. Nobody cares how it works, as long as it works.”
- Councillor Hamann to Neo
Netflix might have a well-publicized recommendation system although they never cease to make puzzling recommendations (also well publicized).
For the uninitiated, Netflix takes a guess at whether or not you’ll like a movie and indicates that guess by the number of red stars present below a movie (yellow stars represent your ratings, none of which appear in the image that follows).
Here’s an example (note: I have no idea what these movies are, or what they’re about).
Let’s imagine Netflix as a helpful assistant, and how that conversation would go.
Rob: “I have a few hours to kill for a layover and free wifi. Computer, I’ve told you what I think of 1,004 movies and I’m not sure what’s been recently released. Can you suggest some?”
Assistant: “Sure Rob, I’d be happy to. Here are 5 movies.”
Rob: “Thank you computer. Actually after looking at these movies, I’m not sure I’d like to watch any of them. Gallipoli isn’t quite the war movie I enjoy. The others aren’t even close to the types of movies I would be interested in.”
Assistant: “That’s what I suspected before I showed them to you.”
An awkward interaction.
Rather than display 5 movies it thinks I’ll rate “Hated It” or “Didn’t Like It”, is there another alternative?
According to Wikipedia, in November 2008 there were over 17,000 titles in the Netflix Instant catalog. Netflix is certainly willing to show me older movies that have just become available instantly (e.g. 1981’s Gallipoli). How about continuing backwards in reverse chronological order until it finds things five items (to fill one page width) it thinks I’d rate at least three stars?
Put another way, I’d like to say “Computer, when you think I’ll rate something one or two stars you’ve been correct so often I wish you wouldn’t second guess yourself. I’ve already hired you and put in the effort to tell you over 1000 of my preferences. That’s a vote of confidence that in your ability to recommend movies. Have the courage of your convictions to avoid showing me movies I won’t enjoy.”
I’ve taken the time to rate over 1000 movies. If I didn’t at least somewhat find the Netflix recommendation system helpful, I wouldn’t have done that. I’d like Netflix to respect that time by choosing to avoid showing me movies it guesses I will either “Hate” or “Not Like”.
This past weekend, my wife and I flew to Valparaiso, Indiana by way of Chicago O’Hare. We had some terrific eats around the city, most notably The Bongo Room and snacks from Sensational Bites. Oh, we also went to a wedding. Congrats Wendi and Justin! ;-)
Back to Valparaiso and the Hampton Inn: one thing every hotel guest does is figure out how to take a shower or bath with a specific set of controls they likely haven’t used before. Here’s a picture of the faucet from our room.
Would you think that the shower is dispensing hot or cold? I say “cold” as the larger hand closer to me is pointing towards the blue notches on the dial with the blue “C” prominently displayed.
Let’s take another look at the dial, this time head on:
From this angle, the dial forms an arrow with the tip narrowing towards the red “H”. Looking at the dial head on, I would have correctly guessed that hot water was on tap.
When you turn on the shower, do you stare straight at the faucet? If so, you are also the kind of person who likes to be sprayed in the face with ice cold water. You probably aren’t that kind of person :-)
Why do we have faucets whose behaviour is not predictable from both in and outside the shower stall?
What’s strange is that I doubt faucet designers haven’t bathed before. Is it possible that our own experiences as designers aren’t sufficient for the task of design?
In addition to not blogging about design, I also have recently not been blogging about photography. I’ve managed to work out Lightroom 2 and the amazing Flickr Export plugin to my liking, using some images taken during a 2-day stopover in Hong Kong earlier this year. Still working out the difference between Sets, Collections and how to keep things organized.
In the meantime, my Hong Kong, May 2009 set!
MozyHome offsite backup goes above and beyond.
Off to Toronto for the Labor Day weekend!
Halfway through our vacation in Asia, my wife realized we hadn’t reset the time zone in either of our cameras. We caught it in time for Vietnam and all of our pictures were stamped correctly. However we were left with hundreds of images taken in Hong Kong stamped with San Francisco’s time! Fortunately, this is a snap in Lightroom.
First unfold the Metadata panel in the Library module, with the Default metadata set.
Second, locate the Capture Time (it’s about halfway down) and click the Edit Capture Time button. The capture time is shown as <mixed> because I have multiple images selected.
The Edit Capture Time window appears:
Select the second option and select the appropriate time zone modifier.
If you’re traveling to the family reunion in Ohio, it’s a straightforward conversion from Pacific to Eastern (Eastern is 3 hours ahead). When you’re traveling halfway around the world, it’s not quite as obvious.
So how did I arrive at +16? To get from PST to HKT, you need to fast-forward 16 hours.
I realize this dialog is doing double or triple duty, allowing you to modify the time in various ways although I don’t understand the inflection. How many times do you want to modify the capture time to a specific time, especially in batch selection?
If you’re selecting the “Time Zone Adjust” option, where are the time zones? Or even better, a map like when you choose your time zone in the Windows or OS X date/time settings? Really, that’s what time zone selection is – a poor man’s “Where I am” and “Where I was”.
It’s probably safe to assume the original time was captured in the (incorrect) Lightroom time zone. We have to make this assumption because the EXIF capture time is ambiguous, as it does not allow recording of the time zone. There is no reasonable default for the corrected time.
Fixing time zone offset is one of those things I do rarely although when I need to do it, it’s VERY important to me. Forcing me to do time zone math seems out of place in a streamlined environment like Lightroom.
When selecting paper on the copier at the office:
I know what’s in trays 1 and 2. Nobody knows what’s in the final tray and I have a strict don’t-ask-don’t-tell policy with the copier after an altercation years ago, that we’re both working hard to forget. A simple question mark wasn’t enough to convey the copier’s emotion, so two more were called in. It could be Gremlins? Or Zombies? More likely some advanced form of paper from the future.
I periodically visit Google Analytics as it has the interesting responsibility of presenting a massive amount of data in a consumable form. Given the numbers in the screen capture below, I derive more value out of interpreting their design decisions than I do analyzing my “readership” ;)
You’ll find much discussion on the relative merits of pie charts (or lack thereof). Keeping the discussion focused on this display, the pie chart and area required to display it consume nearly 50% of the available width. It’s the main draw here – large and full of bright, vivid colors. It’s labelled with the percentages (kinda what the wedges should already be tellin’ ya, amirite?), forcing you to map between the various colors and labels several inches away.
If relative comparison is important (it is important for platform support decisions), what if we embedded a bar chart inside the table?
This allows easy comparison between the browsers themselves though I’m unsure if, without first looking at a visit percentage, it would be clear the row color consumes 50% of the table cell. It may also not be as aesthetically pleasing than the tidy compartmentalization on the first image. There are benefits to this approach, some of which come simply from choosing to use a bar chart in lieu of a pie chart: better area utilization, avoids having to map swatches/browsers to slices, etc.
Here I am creating a new To Do item. In the “Date Due” field, watch what happens when I type “Monday” and press ENTER. Remember that today is Saturday, June 6, 2009.
M-o-n-d-a-y <enter> June 8, 2009 <enter> <committed>.
Would anyone have been surprised if instead, any one of the following occurred?
Now I’m not saying this free-form entry field for dates is a usability needle-mover, just that I was pleasantly surprised that when I said Monday, Apple Mail assumed I meant the following Monday. By the way, Outlook 2007 makes the same assumption in the Tasks view.
As designers and developers, we need to have this courage of our convictions to say “Yes, this is a reasonable interpretation of that input” (tested, backed by research, etc.!), giving our solutions the same social presence as that had by a helpful assistant. This means making reasonable assumptions when told what to do, and not harassing our users with needless confirmation. In 2009, with the availability and power of development frameworks, there’s no excuse for it.
In January of this year, Alan Cooper was interviewed over at Info Queue:
I wrote a book called “About Face” in 1995 and I wrote a book called “The Inmates Are Running the Asylum” in 1999 and both of these books were pluming this idea of how does software get constructed and how can it be designed well from the point of view of how it behaves in the world, as though it were a person with a social presence.
What I’m seeing more of these days is increasingly complex software with more verbose labeling and explanatory passages. It’s not surprising given Cooper’s impact on design and that others like Luke Wroblewski in Web Form Design suggest something similar:
Take the advanced administrative interface for Netgear’s ReadyNAS Duo network storage device.
Previous iterations of interfaces like these would have made do with the terse, less expository text adjancent the combo box. With the conversational tone and actionable quality of the additional text, boneheads like myself have all the information they need to know whether or not to enable this feature. For interfaces like this used once in a blue moon, learning is irrelevant. Infrequent use means that you re-learn the entire thing from scratch each time you use it and this text is a life saver.
What’s great about Netgear’s decision to bold the label is that advanced users can happily scan on by without reading the additional text. I’m pleasantly surprised by this level of sophistication from a network hardware company.