Thursday, January 4, 2007
Just like other Windows OSs we support, Google Talk can be installed for all users or just a single user on a Windows Vista computer. In the past, Google Talk installed for all users on a computer by default. However, because of its new security features, the default install on a Vista computer will be for a single user.
One of the nice things about working for Google is that even though we engineers have a lot to do in our day jobs, we're encouraged to learn new things. We attend "tech talks" (which aren't always technical, despite their name), tackle 20% projects, take training courses, and so on. Among the things I do is try to read one technology book each quarter. That can be a lot of effort, given the complexity (not to mention the sheer mass) of some of these books.
Last quarter, I read User Interface Design for Programmers by Joel Spolsky, host of the popular Joel On Software. It was more interesting — and more useful — than I expected.
It's also short (about 140 pages paperback) and fairly light reading, a rare pleasure in a technology book these days. I recommend it. My only complaints are that it's five years old and mostly covers Windows (so it's really 15 years old in Mac years). Most of the book revisits things we (should) already know: user-centered design, the user's model, the importance of usability testing. Spolsky is great at explaining these things succinctly and memorably.
Here are some of his thoughts, which I think are worth keeping in mind, even though I don't agree with all of them:
- Usability testing of five or six users tells you all you need. You're looking for big problems, not statistical significance on the small ones.
- Users want to accomplish their tasks, not use your features ("users care about a lot fewer things than you might think").
- Usability isn't the same as learnability, and "usability testing" usually tests the latter.
- So what if 30% of users fail to accomplish the task in testing? Those people will ask for help or read the doc, or aren't your target users, anyway.
- Karen Fries at Microsoft invented the Wizard as a way to teach you how to use the usual UI, not to do the task for you.
- Users don't read (documentation or dialog instructions).
- Users can't use the mouse as easily as you think.
- Invent imaginary users, name them, and talk about them to make them seem like real people, which they do represent.
- Icons work well to represent nouns, but not verbs.
If you have time to read only one chapter, read "The Process of Designing a Product," which describes Activity-Based Planning. It's one of several chapters you can read for free on Spolsky's site, but as he points out, the published book is more polished, and you might enjoy it enough to read the whole thing.
Wednesday, January 3, 2007
I've always been a big fan of charts, tables and other ways of analyzing and visualizing data. On my own blog I will often plot things just to get a handle on them. Even here I've posted some analyses that I've done of the data that the Reader team has on hand. It's therefore no surprise that my favorite blog of 2006 is Data Mining.
With this in mind, I've always wondered how I could apply this interest to Reader's feature set. It then occurred to me that there were plenty of things that could be dug up about the items that a person reads and the feeds they are subscribed to. Perhaps some feeds post very often but don't actually get read, so they may as well be removed. Conversely, other sites may not have updated in ages, so they're just taking up room in the list of subscriptions.
Playing around with these ideas, I created a simple Reader "trends" page, inspired by Google Trends as well as the Search History Trends page. After showing it to the rest of the Reader team and a few other interested Googlers, I began to gather feedback and implemented it. Most helpful was when Jeff and Doug rolled up their sleeves and applied some of the same thinking that went into Measure Map to this new Reader page. This went beyond just making things pretty; they helped make things easier to digest and understand.
The result is the Reader Trends page, accessed via a link from the "Home" page. For example, it lets me see the spike in read items two days ago (the result of my New Year's resolution of staying on top of my 322 subscriptions). There are also my subscriptions sorted by various criteria, so I can see which I'm having trouble keeping up with. Each subscription has a unsubscribe icon next to it, which I've used for those that I decided were not worth keeping around.
If you have any New Year's resolutions about time management or are a chart geek like me, trends should be useful and fun. You may discover things about your reading habits that you didn't know.