Archive for category data visualization

Google’s Flight Information: Wild Guesses, not Actual Data

Posted by jasonwong in bad design, data visualization on December 28th, 2009

Google is useful for all kinds of things: certain information typed in a search box will bring up the information automatically. For example, “weather 02841″ will bring up the weather for the Newport, RI zip code. Additionally, typing in “UA 628″ or “DL 1064″ will bring up flight information for the United Airlines or Delta Airlines flight. One issue with this, though. The site that Google uses to pull up flight times in flightstats.com, which is not ideal for this task.

Let me explain with a concrete example. I was flying back from San Francisco to Providence on Saturday, December 19: the day of record snow (20 inches) in Washington, DC due to a blizzard all up and down the East Coast. Thankfully, I was routed through Atlanta (no snow), but things were messy enough that day that I was worried about being late.

My flight was to leave at 6 AM from San Francisco, but we hadn’t boarded at 5:45 AM yet. The gate said no change, but putting “DL 1064″ into Google brought up these results:

Delayed by 3 hours?! Instead of leaving at 6 AM, we’re leaving at 9 AM?! What? And then I noticed the words “ESTIMATED DEPARTURE” and decided to pull up the Delta site. Sorry for the poor resolution:

However, you can kind of squint and see that the flight was scheduled for an “ON TIME” departure. Turns out that we left maybe 15 minutes late.

The lesson here? The information that Google usually pulls up can be trusted. It’s flight information cannot, because it pulls Estimated Times from flightstats.com instead of current data. In this case, Google has failed to anticipate the information users want (actual data, not wild guesses), and that makes this a human factors error.

,

No Comments

Movie User Interfaces

Posted by jasonwong in bad design, data visualization, good design on December 3rd, 2009

Gizmodo says it perfectly:

What do The Bourne Identity, Mission Impossible 3, Mr & Mrs Smith, Children of Men, and Agent Cody Banks 2 have in common? Absurd, futuristic, and totally fake software interfaces, designed in part by one man: Mark Coleran.

Designing a fake dashboard for an imagined supercomputer or a hovering control panel for a worldwide surveillance system is a different process than creating a genuinely usable UI. Your goal is to imply things: that a machine is powerful; that a villain is formidable; that the software is intuitive, but that the breadth of its powers borders on unknowable. At no point does real-world usability factor in, and nor should it—this is pure fantasy, for an audience raised on Start Buttons, desktop icons and tree menus. Here’s a gallery of some of the most famous interfaces; see how many you recognize.

Coleran Reel 2008.06 HD from Mark Coleran on Vimeo.

No Comments

Instructions Fail

Posted by jasonwong in bad design, data visualization, marketing on September 27th, 2009

I understand that some people do not know how to floss. However, I do not think the diagram on the lower right will help anyone floss unless they can split their head open like a ripe melon.

floss

Flip-flop head idea courtesy of Reach toothbrushes:

No Comments

Sony Infinite Zoom Adds Spatial Relationships to Information

Posted by jasonwong in data visualization, good design on September 9th, 2009

My friend Andrew sent me a link to a video of an interesting Sony interface concept. Ignore the title and description – this is not an interface that will load up any image and allow for infinite zooming in. Only computers on television can do that (“Can you enhance that parking lot camera and magnify by 100 to get the
license plate?”)

Instead, this can be thought of as an arbitrarily large image, and the interface lets you smoothly zoom into a specific portion. Therefore, in the calendar example, you have a huge calendar, and each day (represented by a grid square) contains a smaller image that you can zoom into.

Link to the Engadget article

This is a clever interface because it is similar to hyperlinking, but it is more visually intuitive. Imagine the display of hierarchical information – for example, January 5th contains a meeting with a specific location and meeting agenda. You could use this interface to choose your month by zooming in to January, then continuing to zoom in to the 5th, and then you could zoom further into the meeting location, zoom back out to the 5th, then zoom back in to the meeting agenda.

In a standard interface, you would just click on the links that would take you to a different web page, a different view, or some likely different display. However, zooming in and out allows for consistency between the different levels of information. This is especially useful when browsing around for different pieces of information – a uniform browsing experience makes the information feel connected and holistic, instead of clicking in multiple places and using the Back/Forward buttons in the browser, which gives the information no structure. An interesting concept, and it will be interesting to see how Sony uses this technology.

No Comments

Safari 4: Top Sites vs. History CoverFlow view

Posted by jasonwong in data visualization, visual search on February 28th, 2009

Safari is a web browser developed by Apple and is primarily used on Macintosh systems, though there is a Windows version available. They just released Safari 4 Beta, so it’s not quite ready for primetime, but it’s close.

There are several great new features. One of which is called Top Sites. In place of a home page (though you can still set a home page), you get a panoramic display of 9, 16, or 25 of the sites that you visit the most. It’s displayed dramatically, arranged so you can see all your sites at a single glance. You can move sites, get rid of them, and pin them to a certain location if you want. The thumbnails are updated periodically and starred if they’re changed. Handy.

safari_topsites

Another feature is a visual browsing history. This is typically shown as a list of sites you’ve visited, going backwards in time. The list isn’t especially helpful, since all you have to go on are the URLs (because http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&ssPageName=STRK:MESELX:IT&item=320344855221 is not helpful) and the page titles, which are sometimes helpful. Having a visual for each page in your history, though, is great – it’s way easier to recognize visual things visually. Apple implements this feature using CoverFlow, which looks like this.

safari_coverflow

Now, it’s visual – the item you’re focused on is presented fully, and the items before and after are pretty visible. Trying to look at sites three or four items away, however, is unhelpful – you can’t see much detail in the site. All you can see if the left or right margin of the page, which tells you very little. Instead, if history was implemented like Top Sites was, where you could see all (or at least more) of each site, the feature would be more useful because you could more easily identify the site you wanted to go back to.

It turns out that this issue has already been discussed here before. Window switching in OS X using Expose (where you can see all the windows) was compared to doing it in Windows (where you can only see part of the window), and it was the same argument. Interfaces where you can see more of each window allows you to pick out the object you want more quickly because you can search in parallel instead of in serial. I think Apple should take this advice and apply it to more of its interfaces.

,

No Comments