« November 2010 | Main | January 2011 »

December 28, 2010

Pandora: iPad HCI fail

"Oooh! Pandora for my iPad, cool! I've been getting into Pandora lately, bought a premium subscription, using it a lot more & becoming a fan...."

Download & install -- fairly simple. Could be more sexy of an app.

"Oh, I like this song and I haven't trained my tango channel very much. I want to thumbs-up this.... Okay, there's the thumbs up button, I'll click it, seems clear, although you'd think the button would be bigger since there's so much wasted space on this screen and this is one of the primary actions. Would be even worse at a party when I've been drinking more than I already have been."

I tapped the button. Nothing happened. (The icon / button under my finger didn't animate or change in any way.) I hit it a few more times. Nothing happened. I felt frustrated and annoyed -- I really liked that flamenco song! Remember it! -- and looked at the screen. I saw the song title and metadata, a pale gray disabled thumbs up symbol, and a "menu" button. I clicked "menu" and it didn't give me an option to like the song. I felt defeated. Can you guess why?

Step 1: Aiming for the button-like Thumbs Up button.

Step 2: Frustrated, looking at the whole screen.

I'm a right-handed user. My hand blocks the feedback that your thumbs-up button has worked.

  1. The feedback should be in the button itself. Thumbs up / down is a mutually exclusive toggle-type state; the iPod-style controls don't convey that relationship. Nor do they visually convey that my preference for that song has stuck.
  2. The feedback that you have is a gray, disabled-looking flat graphic. You have plenty of screen real estate to tell me "You liked this track". Why on earth would I think that a gray, flat, disabled-looking silhouette icon means that I have taken action on this item? Seriously. I'm a designer. I'm more attuned to visual subtleties than the average bear. I didn't get it until I systematically parsed my way through the experience -- after 3 failed attempts. You've probably got a usability issue.

December 21, 2010

Should you take a pay cut to work at a startup?

Elad Gil: Bootstrapping: Raise Debt From Your Employees

Elad has written several interesting posts about getting a startup running. I worked with Elad and Othman, cofounders of Mixer Labs, doing a little bit of design consulting when they were just starting out. They also ended up offering me a job -- so I had some direct experience with the types of practices Elad mentions.

In my case, E&O talked me down to about half of my normal consulting fee. This was fine because I was still working fulltime and it was a small project -- maybe 40 hours over a 6-week period. I elected not to take the fulltime job for a couple of reasons, one of which was the cashflow issue. I wasn't sure if I wanted to basically invest my own money by taking a 50% - 70% salary cut, and didn't think I'd get enough equity to be worth it.

And that's how I looked at it: The startup was raising money from me, but I probably wasn't going to get a comprable equity cut. Suppose a startup takes $2 million in VC funding. The VCs get a 40% stake and probably preferential treatment on an exit event (i.e. if you get bought out for $2.1m, the VCs take the first $2m and the rest gets divided amongst the remaining equity holders). The founders get in the realm of 10% - 20% each. Senior engineers / VPs / early directors may get somewhere between 50 and 150 basis points (0.5 - 1.5%).

Suppose I worked there at a salary cut of $50k per year for the 2 years until the startup has an exit event. I've invested $100,000 in the startup via lost earnings. (Remember, if the startup tanks, I lose my money too.) As a design-type person, let's assume I was one of the first 10 employees and got 50 bp of equity. That means I've paid $2,000 per bp. On the other hand, the VCs who contributed $2m for 40% (4,000 bp) are paying $500 per bp. If the founders take, say, a $100k / year salary cut, they're each investing $200k for, say, 10% -- or $200 / bp.

If I'm going to get bupkes for bp as a non-founding employee, then I want a full salary. If I'm investing, I should get a comprable equity cut to other investors.

December 16, 2010

Statistical linguistics made possible by Google Books


Already, researchers have traced the accelerating evolution of the English language, mapped the rise and fall of various people, and uncovered patterns of censorship and suppression in Soviet Russia, modern China, and 1950s America - and that's only a beginning.

"This dataset is going to underwrite a field which is far, far more interesting than anything we could talk about in a single paper," says Erez Lieberman Aiden, an applied mathematician and bioengineer at Harvard University who led the research along with Jean-Baptiste Michel, also of Harvard.

This is really, really cool.

I used a similar, more primitive type of linguistic analysis when I wrote my undergrad thesis (The Rhetoric of Privacy) at CMU. I used Lexis-Nexis to track mentions of the word "privacy" in U.S. periodicals and the L-N corpus of transcribed speech, from ~1980 - 2000. I just sat there in the library doing individual queries for each month and tracking them in a spreadsheet...

December 13, 2010

Thoughts on Interface Consistency

We had a big discussion today at work about consistency. When do you force it? When do you challenge it?

On one hand, I understand the point made by a PM: "But I just think it looks bad. I don't want that on my product. What should I do?" I've got issues with the current styleguide I'm working with, not least of which is that it only has partial coverage of design patters: hits the visual, but doesn't address interaction very broadly. (This is largely due to resourcing as far as I can tell.)

On the other, if I have to have one more fucking discussion about how to design a standard element like a menu, a combo box, a wizard, etc., I feel like I'll want to stab myself in the ear. I don't want to spend time on commodity web design. We know the web -- or we should, anyway. What we don't know are new technologies and business processes. Do you really want to be arguing about link colors when you could be figuring out how to get more people to sign up? (Actually, link colors may help in that situation -- the big blue button will almost always get more action than a tiny gray-text link.) Should you be spending time re-exploring sitewide navigation (two choices: top nav or left nav) or re-exploring how to make your users smile?

And yet: as Kevin Fox tweeted recently, "If we weren't always reinventing the wheel, wheels would still be made of stone."*

Consistency is really really important, but consistency with the rest of the sites that users spend their time on (most) is much more important than consistency within your own site. Consistency within your own site does add a certain j'ne sais quoi (very important!), but it's a much smaller refinement than behaving in a manner that conforms to other people's experiences on the other sites they frequent.

Nielsen on surface design vs. deep design (task-centricness).