Back in December of last year, I did an HCI project around designing an alarm clock. I wrote about it a little bit here, and - in an attempt to duplicate a design-crit process I've seen elsewhere - solicited some really helpful feedback from some of the readers of this blog.

For one reason or another the project took a little while to get marked, but we had it handed back this week. It's unlikely to be of interest to anyone, but given that quite a few of you helped, I've put the report (all 50mb of it!) here. It's been given a provisional mark of 83%: not shabby, but not amazing either.

Overall, I felt a bit "meh" about what I'd produced, for a few reasons:

  1. It didn't really conclude with what I felt was much of a design. I'd tested a few ideas but with 3 test subjects, didn't feel I drew many conclusions;

  2. As noted in the report itself, I didn't get much value from doing a mid-fi prototype (which was mandated as part of the project). I think I would've done better to explore more ideas as lo-fi sketches, and move to hi-fidelity when getting them onto a real device. I felt that some issues test participants observed could be explained by the deceptive fidelity of the prototype;

  3. I referenced, but didn't feel I ever tested, "stroking not poking" (Warning: "bad workman" ahead). Throughout the project I struggled to find tools which would let me prototype the fine-grained touch interactions (e.g. drag a hand around a clock) on a real mobile device. At the time I presumed that they exist somewhere but are proprietary. Since then I've heard rumour that even within Apple, iOS app prototyping is done with Keynote. I evaluated Keynote and found it wanting - so I missed a trick there, I think;

  4. I think I'd made my life harder for myself by doing some research up-front in the form of a survey (reasoning that I know nothing about sleep habits, and ought to). There was a 2500 word limit for the project, and reporting the survey ate into this. One very reasonable criticism from the marker's feedback was that I could've gone into more detail on the literature; I'd originally written much more, but cut it down for the final version. A lesson learned: this project was supposed to be either research or design; I shouldn't have tried to do both.

Some good stuff did come out, though: I was surprised to observe that analogue clock-faces are easier to read from a distance, compared to digital displays. The survey validated my original suspicion that the high-cost-of-failure associated with oversleeping, combined with the high frequency of oversleeping, indicates there's a worthwhile problem to be solved here. And in future I would use the combination of "Omnigraffle exporting to HTML" and LiveView to quickly test UIs which needed fewer unusual touch interactions.

Thanks to everyone who lent a hand - the 186 people who filled out my survey, and everyone who commented on the blog or sent me emails. In particular Nick Richards and Dave Whiteland were awesomely helpful :)