Seminar 2 - Ariel Blomkvist Rova

I think that this is where it get's interesting. This is where it gets design, testing ideas against the real world.

Chapter 13 introduces introduces guidelines to different evaluation tactics. 

Controlled settings involving users, fit for usability testing, for example in a lab environment. 

Natural settings involving users, fit for identifying opportunities and establishing requirements.

And finally, any setting not involving users, focusing on using predictive models to identify usability issues that users might have. More on that later.

I found both the case studies to be very interesting. Especially, the in the wild of skiers cause it made use of so many different technologies which were unobtrusive to the user being tested. It also reminded me of an kexjobb-study I took part in this spring. I got to watch a CS:GO stream with an eyetracker which logged how I responded to ads in the stream. Very cool! Makes you think about what an ideal way of testing our concept would be if we had unlimited $$$ and time?

Chapter 15 dives deeper into heuristic evaluation, walkthroughs and analytics. 

Analytics is quite self explanatory, the other two more interesting: 

Heuristic evaluation is when an expert roleplays as a user and inspects the service or tech from the view point of certain design principles. Of course, this can never generate as specific results as a test involving users, but it's a relatively cheap way to avoid common pitfalls. 

Especially interesting for us is that there apparently are many evolvements of the original heuristic framework, which caters to different technologies. In our case, since we're moving in some kind of realm resembling ambient devices/displays, we might wanna look into heuristic evaluation of ambient devices, as well as hedonic heuristics and similar. Another realm we are nearby is emotional design, maybe there is a framework for that too!

Another method from 15 is the walkthrough. The cognitive walkthrough is about analyzing whether or not an task is cognitively easy enough to carry out. This is also most often done by an expert.

Predictive models are mathematical approaches which can be used for example to find out the ideal placement of buttons.