Yahoo recently updated their mobile application Yahoo! Go 2.0 (beta) to Yahoo! Go 3.0 (beta).
I’ll admit I have always had great sympathy but at the same time mixed feelings about Go. It holds a lot of promise: everything you need, conveniently in one place, optimized for mobile. But Go has failed to deliver in my opinion… Not that there is all that much wrong with the app itself, it is based on some sound principles, it looks good, and I bet it demoes well!
The Go philosophy
The Yahoo approach is to deliver optimized content for your mobile existence. There is no “one web” thinking going on here! Yahoo massages content into a form suitable for constrained bandwidth, small screen devices. The idea is to deliver small relevant bits of information. The app uses contextual information (like location) to deliver a generous variety of relevant results and require as little text input as possible from the user.
Search for “pizza” and Go will return Pizza restaurants in your area, pictures of pizza, news about pizza, web search results for pizza, mobile web search results from pizza. To be mobile optimized, the results should not be a bunch of URLs but contain the actual information, sized for mobile. Restaurant names along with the physical distance from your current location, thumbnail images, news headlines etc.
This approach has been picked up by both Google Mobile and Windows Live Mobile.
One of the main usability issues with Go is the apparent slowness. The application invites you to explore, but its slowness punishes exploration. There is a delay every time you click anything that really takes away from the experience. The application feels sluggish, the huge download, and the occasional crashes adds to the feeling that this is a house built on sand, even if it’s not.
The early version 2.0 tried to pre-fetch feeds from all content sources when the app started. I don’t know if it was due to platform problems or poor programming, but the app was pretty sluggish for a couple of minutes after start. With the later version 2.0, this improved and the application felt a faster. Version 3.0 feels like a step back again in this regard.
It is of course more complex than just optimizing the application for speed. There is a balance between cool features/interactivity vs. speed. As history has shown, it is often a better long term strategy to spend your time creating cool features and leave it to the hardware manufacturers to fix the speed problem.
I’m sure that if you measure it, Go retrieves data faster than a browser would. But it feels slower. The browser builds up the page gradually, you see that its working. It may be slow but it feels like progress. Go displays a pendulum and the text “Please wait”. Simply changing the pendulum to a progress bar would help.
A new architecture
Yahoo has done some fairly substantial changes to the architecture in version 3.0. All sections of the application (finance, news, sports, etc) are now “Widgets”. 3rd party developers can create additional widgets that can run inside Go and on Yahoo mobile web sites. A widget gets a spot in the carousel. In this version 3rd party widgets lives on the server. Offline storage (ala Google Gears) will be added in the next version, according to Yahoo.
Snippets are mini-widgets with a standard layout that are designed to display a little bit of text or an image and be a link to a widget or a web site. Widgets and snippets must be approved by Yahoo and there are no 3rd party widgets or snippets available yet.
I have removed everything but Mail, Maps and Flickr and the application now makes a lot more sense! It also starts faster. I mean, now the app only contains stuff I need, while earlier it mainly contained stuff I did not need.
The main reason 3.0 feels slower is probably increased use of graphic effects. The UI has received major beautification. No one avoids the iPhone effect. As you can see from the above screen dumps, the fonts are now anti-aliased, there is a lot of gradients and shadow effects. A lot of this probably done with increased use of SVG (Scalable Vector Graphics). SVG is of course the way ahead, but even the Nokia N95 I tested on (which has its own dedicated graphics processor) does not render SVG that fast.
Btw, the next version of Symbian will have a new graphics architecture called ScreenPlay. It is optimized for vector graphics rendering and image composition. But my guess is that come March or April, we will see an blazingly fast iPhone version of Go.
User interface issues
Yahoo has had problems with figuring out softkey labeling. It may seem like buttons are preferred in the US while softkey labels are preferred in the rest of the world. Any way, version 3.0 is a mess in this regard. There is way too many message boxes and dialogs in Go and no consistent handling of them.
UX decisions I think is questionable:
- Lists don’t loop (they don’t jump to the first item if you scroll past the last item in the list).
- Version 2.0 had a great feature where you could search by just starting to type from anywhere in the application. This is now gone.
- Version 2.0 had a nice demo mode that ran the first time you started the application. Version 3.0 has dropped this. Worse is that it introduces a number of new concepts, features and settings but there is absolutely no explanation even for the interested user.
I just hate the use of the Macintosh pop-up menu, a selection control as a command button. The pop-up menu has been around since the early Macs back in the ’80s and its behavior is pretty well established. Also, don’t use the standby power symbol for an Exit command.
Messing with the popup list and the standby symbol gives the impression of poor attention to getting a clue.
In previous versions of Go, the Back command took you back to the Carousel. The effect was that the Carousel, the main menu, acted as the center of gravity in the application. If you feel lost, back-step until you are in familiar territory. With version 3.0, back is now an application wide stack like the web browser history. The same change was done from Windows Mobile version 5.0 to 6.0. I think this is an improvement.
Mobile web content
Go displays web content in an internal browser that only renders text, images, and links. Web pages are run through a transcoder that strips everything else including forms. Pretty much all phones that runs Go has a much better internal browser and I would prefer Go to display web content in the browser. Even mobile web pages are transcoded.
A random mobile web site (from BBC). The transcoding removes the search feature and splits the site up, leaving it pretty much useless. You might of course say that the Yahoo version is less ugly and you would have a point there.
Who is version 3.0 for?
With the previous versions of Go, Yahoo obviously did a lot of thinking about what a good mobile application should be like. The new version 3.0 is all about solving problems for Yahoo, not for the users of the application. I get the impression Yahoo needs an “answer” to Google Android, an “answer” to Nokia Widgets, an “answer” to the iPhone.
Based on a few hours of fiddling with the application and I’d say that there are some visual improvements and some user experience decay. The primary personae for version 3.0 may well be “a Yahoo shareholder”. So, if you are one, get this version!
Posts like this tends to focus on problem areas, after all this is a blog about Interaction Design issues, not a product review. I’m going to keep using Yahoo Go 3.0 for a while. The new widget concept holds some promise. I want to stress that Mail, the Map and the Flickr widgets are pretty decent apps! The maps especially is very impressive and quite useful. If you haven’t yet done so, go check it out here.