Hello! It's been a little while, but a ton of work has come together. I passed a big milestone last night: 30,000 words. But more important is this here Kickstarter project, so the book can be published in the first place. Again, I want to thank everybody for putting their faith in me over the past two months; it pretty much rules, and I can't begin to express my gratitude.
We have one month left as of today, and so far we haven't yet broken $4,000. (We're very, very close, though!) If you want to see Cadence & Slang find its way into print, please, please ask people to pledge. Even if they promise to later in the project, it makes no difference whether they pledge now or later, because nobody's cards are charged until the project ends (and if it's fully funded). I'd be able to sleep better at night knowing that the support is actually out there, and that folks are willing to back their words up. If you're reading this and on the fence, please donate today! Pre-orders are only $40, and you get a still-ridiculously-useful handmade outline of the book for just $20. (And if you want to change your pledge to something higher, click "Manage your pledge" on the project's front page.)

I've spent the past week working on edits to chapter 1, so I can strengthen the initial arguments as much as humanly possible. I've also sent a draft of the first 15 pages' revisions off to some friends and colleagues for review. I know that's not a whole lot to write about the book, but just those revisions have taken a considerable amount of time. When I'm not writing at night or on the weekend, I cram editing in on my trips on the El and bus, and during my lunch breaks. It's a lot of editing.

Finally, I spoke about Robert Hoekman's A List Apart article about usability testing and Lukas Mathis' rebuttal last time, and promised to offer some of my opinions.

There are a few things I love about Hoekman's article. He suggests tracking every single user click on a web site. Click statistics are relatively easy to implement, and can offer some extremely useful data about what navigation is (or isn't) being used, what links are (or aren't) being followed inside main content, etc. While some kinds of users could befuddle the data (for example, I'm one of those people who compulsively highlights blocks of text while reading articles on the Web - so I probably generate lots of click "noise"), and while click data may not be useful for software applications, it represents a good step for the Web, especially on sites that want to direct users towards some specific content or action.

Five-second tests sound like a great idea. I've never given one, but for sites that need as much clarity as humanly possible, which tend to perform only a couple of functions, five seconds lies at the high end of how long it should take users to recognize who you are, what you do, and how they can get things done. Think about how fast you browse the web: you likely skim text, gloss over things that doesn't clearly relate to the site itself (like advertising or "share this" links), and you spend as little time as possible on search engines. Five seconds is more patience than we give the hours of most peoples' work to get content online. I was inspired by this part: to be honest, I can't believe I hadn't thought of it previously.

So Hoekman's opinions about how to improve usability testing are pretty good. But those methods are a testament to why usability testing should be improved, and that it's potentially useful - not that it's a myth. The first half of that article attempts to debunk heuristic evaluation as unfocused, fraught with noise and conflicting opinions between ostensible "experts." Hoekman hits on some valid criticisms of usability testing, but they're not catastrophic enough to write off the entire field.

In two different examples, Hoekman cites tests that were run with the wrong set of users that the client wanted to target: existing ones, rather than new ones. And with the comparative usability tests among different teams, the goal was to find the right practices for usability testing in the first place. Of course each team would come up with their own heuristics. If each were testing according to the same heuristic, then it would be scientifically accurate. But each team was trying to do different things, measured against different standards. Heuristics should be developed with as much sensitivity to the user's and client's needs as possible, and they should be kept consistent among each test.

So, as Mathis argued, usability testing is apparently a myth because... bad testing yields bad results. This only gives me more conviction that we need to find ways to rigorize and strengthen testing methods.

If we assume that Hoekman's criticisms are valid, however, that means external usability testing services - which he recommends as an alternative - are useless as well. In fact, I disagree that external services are a preferable way to test in the first place. You gain the benefit of impartiality, but with many projects, you would have to be very precise in the demographic requirements. And it may take too long to get results back if you refine too strongly; UserTesting.com says one hour, but that's for people less fussy about who they get. And third-party testing is impossible if you're testing against existing users.

All this considered, I think there are two important takeaways: 1) usability testing is still good, but it needs to be refined a lot, and more conversations about big-picture stuff like this need to take place; and 2) if we're talking about big-picture stuff, it's a clear sign that our field is still pretty young, and trying to figure out how to take its tentative first steps. I'm really glad we're all talking about this, because it's important, and way too many organizations gloss over it.
Next: book excerpts‽