SXSW 2011: A Peek at My Notes

At this year’s SXSW, I didn’t see any break out hits or major controversies. Foursquare continued to bask in the press’s glow, group-texting services battled over who could give away the most free tacos, and Chris Brogan wasn’t around to pitch a fit over douchebag marketers. The keynote interviews were fine, if not exactly revolutionary. There weren’t any outrageously scandalous parties, although, this happened. Even though Southby has the reputation of Nerd Spring Break, there are some awesome panels that take place during the day. It’s a shame more arts organizations can’t afford to take part in this great learning opportunity. So I wanted to share some of my notes with you.

Or check out the Samsung hub, or on Slideshare for more Southby presentations.

Using Twitter to Improve College Student Engagement (#twesearch)

Leximancer is a content analysis tool with a heat map and a frequent words map, linked by key concepts. It’s expensive, but it’s the prettiest maps I’ve seen from one of these tools. Studies show that in college, academic engagement → critical thinking, social engagement → psych-social adjustment, extracurricular activities → student retention, degree attainment, pursuit of non-traditional careers. In order for Twitter to be useful, instructors & students need training & encouragement, tool should be integrated with course content, and collaborative learning should also be encouraged off-line.

Colleges can use Twitter for

• Continuity of class discussions (more time on task is better)
• Low stress way to ask questions (leads introverted students to be more extroverted in class when they can engage online first)
• Class/event reminders
• Help students self organize into study groups

The Game Layer

Seth Priebatsch, founder of SCVNGR, gave examples of how game mechanics are or could be used to solve 5 different problems: education, customer acquisition, customer loyalty, location based services, and global warming. I thought the most interesting one was about school. School already looks like a game — you have friends and foes, you advance through levels, you earn points for completing tasks on a deadline, there are cheat codes, etc. With this understanding, there are at least 2 ways we could make the education system “work better” (this coming from someone who dropped out of Princeton in his first semester):

• Decrease the correlation between grades and success: earn enough cumulative points, you move on to the next level, always at your own pace.
• Decouple subjects from grade levels: you could be a level 6 math wizard and a level 2 writing warrior, and in that case you should spend more of your time that year working on your writing skills

Provocatively new ideas, no? Seth also gave a shout out to the book The Social Fabric of Gameplay by Jane McDonegal, whose name I heard thrown around by more people more times than anyone else at the conference.

Real World Moderation: Lessons from 11 Years at Metafilter (#RWModeration)

Keys to effective community management

• Be the best participant in your community
• Don’t be overprotective by setting too many rules
• Give users a place to vent against you
• Involve users in the process of community management
• Good moderators notice patterns
• Really important to share information between moderators
• Put sketchy users on a pre-emptive watch-list and then watch them carefully
• Spammers are getting much smarter
• Commenters have a lifecycle-those that are highly active tend to burn out quicker

And a note mainly for Ian to say, I was wrong about community managers needing to scale with networked communities. Metafilter has 12,000 active users per week (including 300 heavy users), 200 posts/day and 3,000 comments per day, and employs just 4 community managers. But they’ve spent an extraordinary amount of time & effort building their own tools to make community management more efficient.

StackExchange uses good community game mechanics; if this is the kind of thing you’re into, Matt said has a ton of papers on community management, just ping him on twitter for access.

TED: Radical Openness How We Grew TED by Giving it Away

Online views of TED talks breakdown to:

• 20% on TED.com
• 20% downloaded via iTunes
• 15% YouTube
• 5% embedded on other sites
• 5% other

YouTube told TED that embedded plays are the tipping points for virality. TEDx came about because people kept asking, not as a brilliant idea from someone on staff — now there’s been 1500 Tedx events in 95 countries in 35 languages, TED has begun crowdsourcing speakers from these events.

What makes an open strategy work:

• Put forward a clear goal that inspires
• Draw a passionate user base
• Provide clear guidelines with rewards & consequences
• Allow the community to police itself
• Make the community rockstars

The Science of Influence

TweetMap is a mapping tool showing how content spreads — the maps often look like an inverted pyramid, meaning that the quantity of followers you have matters more than quality of content you put out. Needlebase is a company I’ve been excited about since hearing about it from the panel’s moderator, who promoted it again: it’s a really smart, intuitive web scraping tool. Social proof is a myth. Dan Zarrella, whose work at Hubspot I’ve been a fan of for awhile, did an experiment: A/B test the same blog post, the only difference being the TweetMeme button showing how many times this article had been tweeted. CTR rate was highest on the button showing the least number of tweets.

Recommendation Engines (#DiscoTalk)

This is one of the best panels I went to. Engineers & founders from YouTube, Pandora, StumbleUpon, and Google HotPot all described how their service uses algorithms to recommend new content to users. It was a great mix of practical advice, debate between panelists, and truly interesting questions from the audience.

Most recommendation algorithms combines Who you follow/friend, What’s trending, and Your historical preferences. For example, only 50% of searches on YouTube are looking for a specific video, the rest want to explore. So YouTube uses “related videos” to keep users engaged with the site — these can’t be 10 videos of the exact same topic, you actually need variance within the cluster to keep people interested. YouTube also gave a great example of sometimes we want to find what’s least popular — what’s that really awesome video that none of your friends have seen yet, that allows you to look like a tastemaker rock star by emailing them about. Telling users why you’re recommending these videos to them increased click through rates.

In order to make a good recommendation engine, you need metadata about your content, canonical metadata is even more useful. In the beginning, brainstorm what “perfect metadata” would look like, even if you know you can’t get it yet. This will help you build towards that direction. People tend to overestimate the burden of manual labor in collecting this information — Google has driven cars around nearly every street in the country, has people scanning books by hand, etc. Sometimes the best way to collect the metadata you need is to put in some legwork.

Off the shelf recommendation engines exist — Hunch & Gravity were both mentioned, but it turns out nearly all recommendations are vertically oriented. Just because I know what books you like to read, doesn’t mean I know what kind of music you like to listen to. Beware that better recommendation engines hinder some traditional web stats — page views may decrease because people spend less time flipping through non-useful pages. Pay attention to metrics like time on page.

The Thank You Economy

I’ve read Crush it!, I follow @garyvee, I’ve heard stories about how engaging he is in person, but it turns out Gary Vaynerchuk is even wilder, funnier and swearier in person. Here’s a bunch of paraphrased brilliant one-liners:

• Don’t be a 19 year old boy, don’t try to close the deal in your first minute.
• All you have to do is outcare the competition. People are going to start battling out on the care front — it’s too practical not to.
• I spend 20 minutes taking an episode of WineTV, and 20 hours building a community online
• Don’t retweet the nice things people say about you, that’s being a braggy douchebag. It’s a dickmove to do off line, don’t do it online.
• Content calendars suck. Do you take a script that you wrote last month to chitchat at a cocktail party?
• Saying hello when a customer walks through your door doesn’t have an ROI. Being online is about building a relationship.
• If you’re not a CEO, it’s really hard to change your organization from the bottom. But Egypt did it.

And now for something a little different. For 3 sessions, I tried new playtoy Storify. Click through each page to see the live digital notetaking I did…

Networked Nonprofits (#netnon)

I got my Storify invite a month ago, but hadn’t had time to use it. In Clay Shirky’s speech, the guy next to me took notes via Storify and I was truly inspired. I thought Beth’s panel would be a good candidate for some experimentation with Storify because she invariably involves lots of different people, examples, and multimedia. Unfortunately, Storify wasn’t playing nice with Twitter, so you’ll notice a bit of the formatting is off, but it was still an interesting test case.

[View the story “Nonprofits and Free Agents in A Networked World” on Storify] Building an API for Public Media This time, Storify was on fire, and I was able to collect this: [View the story “Building an API for Public Media “ on Storify]

Creating a Social Hackathon (#goodengine)

Talk about an all star panel of folks I twitter stalk, and an idea I’d love to implement for the arts…

[View the story “Creating a Social Hackathon for the Good — Justice League Style” on Storify]

Today begins the Nonprofit Technology Conference in DC. I’m giving an Ignite talk tonight on 13 Guerrilla Research and Analysis Tools to Rock Your Social Media World. Hopefully there will be another set of “conference learning notes” coming out next week from all the great panels I’ll be attending the next few days.