Liveblogging the Kathy Sierra Keynote at SXSW

First attempt at liveblogging. Here goes:

Kathy opens by proposing a dilemma for those of us in attendance: We’re responsible for evangelizing the products that make this interaction unnecessary. That’s the question. So are we saying, “We don’t need real humans”?

Real logical reason we’re here? Shows the slide: “I’m here to get laid.”

We need to put “human-ness” into our software. That’s not the same as having a “humane” interface.

Asks 3 groups to stand up: Designers, Coders, Money people. Some of us may fit into more than one category. Introduce yourself to two people from the other group.

We should do 2 things:

1: Help our users get together offline.

Begin by accepting that face-to-face matters. The more people get together, the more interaction takes place.
- Start / help a user group
- Sponsor a camp

2. Make our applications feel more human.

What can a human do to another human that they can’t do to a computer? Show emotion. (Not what you were thinking). Ask a question.

Being able to “look confused” is crucial to interaction.

Neurological quiz with Asperger’s symptoms. Why? All our applications have Asperger’s. How can we compensate for that?

• Give it a way to know that the user is confused.
• How about a WTF? button
• FAQs and Online Help don’t solve this - there’s a big difference between what Help thinks you feel like and what you actually feel like. This is called the “Canyon of Pain”.
• If Online Help was the right solution, the first line of the help file would be: “Don’t Panic!”

It’s not about the tools you build … it’s about what your users do with them.

Start by thinking like a human. In real life, when someone looks confused when you’re talking to them, you ask them a question. So we need a interactive dialog with the user. It turns out that the same things are said over and over. We can capture 80-90% of what the user is looking for.

The goals:

• Get the user to the right context as soon as possible
• Then give him a understandable set of questions

You can do it be letting the user choose a high-level statement (”I’m lost”, “Why did this happen?”), then narrow the context. It’s like way better context-sensitive help.

Bottom line: Give the user a way to express herself to the system in a more human-interactive way.

What other emotions could the computer “recognize”? Could it ask “what does your face look like right now?”

You’ll know when you’ve succeeded with your users when they’re a little creeped out. That can be accomplished by grouping things in the right context. (tips, pitfalls, troubleshooting) Most importantly — talk like a HUMAN.

Talk like (Robert) Scoble, or Jason (Fried). Don’t talk like a Non-Human. Like Data. Or Spock. Or Ze Frank. (kidding)

How we treat customers before and after they buy our product. Before: flashy brochures. After: Crappy manuals.

Help, FAQs, and user documentation might not sound sexy, but it’s the key to passionate users

You can outspend your competition, or out-teach your users. Get past the suck threshold and to the passion threshold.