selenamarie

Interviewing

• Choosing Participants

Representative of target users or may be users of the current or similar system. May be non-users. Might need to pay people — in the bay area, $50-100/interview. Could maybe offer a token of appreciation in exchange instead if you don’t have cash.

Approximate the participant profile you need if necessary.

The importance of being curious

• Convince yourself that every subject has some interesting aspect.
• Figure out the difference between power and knowledge. Start in the middle, not the top, to find a story. The people in the middle are the ones that do the work in the world.
• Self-consciousness is the enemy of interestingness

What makes a good question when you’re interviewing

Some bad questions:

• “Is the daily update an important feature for you?” (LEADING QUESTION)
• “Would you like stores with less clutter?” (LEADING QUESTION)
• “What would you like in a tool?” (Users are experts in their own lives, not experts in design)

Good questions: 

1. Open ended questions
2. Compare to another product, rather than an absolute scale

Allow a little bit of silence! So that people can answer your questions.

The birth of HCI

Vannevar Bush - instrumental in setting up large-scale government research. Instrumental in creating: National Science Foundation and ARPA. Wrote about Memex in 1945. Invents hypertext as well (“trails” through the microfiche!).

Knowledge workers need to create as well as consume. And you wear a camera in the center of your head to capture things. :D

Grace Hopper created the first compiler. This provided a much better interface to computers for programmers.

Jumping now to the Graphical User Interface

MIT’s Lincoln Labs — key innovation is that the user’s input is overlaid directly on top of the system’s output. A light pen!

Mouse and hypertext. Doug Englebart navy radar technician. Found a copy of Life, which had reprinted Bush’s atlantic monthly article. 1968 demo video of a mouse!

Alan Kay moves to the Stanford AI lab, but saw Dough Engelbart’s demo. 1970s prototype of a laptop computer out of cardboard called the Dynabook.  In 1981 - Star computing system from Xerox was introduced.

The “Long Nose” of innovation: Early ideas come out decades before major adoption! :)

Evaluating design: empirical methods

How do we measure success? Simple question, complicated answer!

It’s difficult to tell how good a design is, until we evaluate designs with people. Hard to predict what real users actually do. Start with defining what some of the things you’d like to know. 

Need lots of different methods to achieve different goals.

• Usability Studies (on a spectrum of informal/formal)

Advantages: learn a ton in a short period of time

Problems: not an “ecologically valid” location, “please the experimenter” bias, no comparison points, requires bringing people to your physical location 

• Controlled experiments (video was problematic)
• Surveys & focus groups

Advantages: Automatically tally results, quick/large group gives feedback

Problem: what they “say” what they actually do, groups: tease out of colleagues what they might not say otherwise but people are more likely to say polite things or say things that are not correlated with what they actually do. Very difficult in bias or culturally sensitive research projects.

• Feedback from Experts

Peer Critique

Dogfooding

Heuristic Evaluation

• Comparative Experiments

A/B tests! :) Learn which option is the more effective. And which variables matter.

Comparative feedback is often more *actionable*.

• Participant observation

Need realistic, longitudinal data? Do this.

• Simulation & formal models

See Monte Carlo simulation. Examples: development of shorthand, models of web surfing behavior. (Can estimate click behavior based on link text! :)

Issues to consider when picking a method:

• Reliability/Precision
• Generalizability
• Realism
• Comparison
• Work Involved

Evaluating design with people is easier and more valuable than you’d expect.

Prototypes are questions. Ask lots of them!

Notes form the second video:

How to think about Prototyping: “A question rendered as an artifact”

The goal in prototyping is *feedback*, not the prototype itself. 

example: Kodak DC210 Plus Digital Camera

Prototypes should nearly always be incomplete.

“Prototyping is an efficient way for dealing with things that are hard to predict.”

Our intuition is often wrong.

Focus on the GOALS of the design, and evolve the design.

Book mentioned: Sketching User Experiences: Getting the Design Right and the Right Design (Interactive Technologies) by Bill Buxton

What do prototpyes prototype? 

• Feel
• Implementation
• Role

Think of prototypes on a Learning/Communication vs. Time axis. Minimize time and maximize learning/communication.

Walter Dorwin Teague’s rough prototypes of airplanes. Famous industrial designer who prototyped large designs in warehouses. Apple did a similar thing with their first apple stores. They decided to organize around activities instead of products based on interaction with prototypes!

Linus Pauling said “The best way to have a good idea is to have a lot of ideas.”

Serial/iterative design —> local improvements. Simulated Annealing —> helps you find global maximums, instead of local maximums. Also useful in database planner optimization!

starting with the HCI course from coursera today

Just got through the first video. The idea here is to teach people “pretty good” design, so that we can save 1 billion people an extra 10 minutes on all these interfaces they try to use every day. 

Let’s just stop wasting people’s time with horrific design, ‘k? :)

Liking the tone of the course so far!

And Users are just one stakeholder — also the developers, people funding development and others. He recommends The Design of Everyday Things for further reading. Pretty sure I have that around the house somewhere.