threadnaught

First attempt at liveblogging! Why not? Don’t laugh.

My fingers are tangle-prone, so there’s some paraphrasing.

Daniel Burka Creative Dir, Digg/Pownce
Joshua Porter Founder, Bokardo Design
Chris Messina, Citizen Agency
Todd Sieling, Ma.gnolia

Porter: We’re talking about social design strategies. To give you a bit of background about where we see the space, we’re in a third phase right now. The first phase had static html sites you could just read, one-way communication. In the second phase, sites could save info and started 2-way communication (banking). In the last few years we’ve seen a huge rise in social communication, e.g. Facebook and Myspace social networks, object-based SNs where users upload content of some sort.

What we’re talking about are the design issues that come over time as you see more and more social interaction between your users. These are counterintuitive, and very hard to predict. Today we’re talking about the problems we’ve seen and our solutions to them.

We’ll have three quick presentations and then open up to broader discussion; Chris will moderate.

I’m going to talk about one of the challenges: how do you encourage good behavior? These are not really ethical matters, although those are interesting. Rather, when you have a site that promotes an activity, how do you encourage people to do that activity?

Tie behavior to identity. If you don’t do this, people don’t feel responsible for what they do on your site. e.g. on Amazon, people write customer reviews under ‘real name’. Real names give more authority to user comments; anonymous comments appear less valuable . People are much more likely to read reviews from people with recognizable names.

 

Also, eBay feedback profiles: millions are exchanged every year between people who’ve never met. eBay has a sophisticated design to their feedback profiles that allow people to assess sellers based on the history of the seller’s behavior. Each transaction is described in depth; eBay is effectively saying that the seller’s authority in the system is tied to their previous behavior. eBay doesn’t use ‘real names’, so identities are system identities – it’s their previous behaviour that matters.

Second way to promote good behavior is to give recognition. Feels good to be recognized. Digg uses this: ‘top diggers’; last year digg took the feature down…

Burka: Top diggers made a lot of sense with a small audience when we were trying to teach people what’s good about digg; digg is largely about games [one-upmanship] and once we reached a certain threshold, a small group of users got aggressive and it got less interesting to the rest. We debated whether to make it more transient or to axe it altogether; we ultimately killed it. I don’t think it was a bad idea to put it in place to begin with; best to adapt as you grow. Some features don’t survive a certain threshold of use.

Porter: Works best when recognition comes from the group. On digg, because top diggers was a result of cumulative activity over time, it was easy for long-time users to get on list than new ones.

Burka: At the same time you want it to be meaningful; there’s certainly a balance to achieve there.

Porter: threadless allows designers to get recognition, but that tapers off relatively quickly until they enter another contest.

Another way to encourage good behaviour is to show causation. A great example of showing causation is the netflix website. Most of us know how it works, and one of the reasons is that netflix tells you how on every screen; they show causation all over the place. They tell you no less than 4 times on ‘Movies You’ll Love’ page alone that rating games gives you better recommendations.

Burka: They’re not even telling you, they’re showing you.

Messina: Pandora also does that…

Porter: Leverage reciprocity. It’s incredibly core to social psych research: it’s the feeling you get when someone provides value to you; you feel somewhat obligated to provide value back to them. e.g. someone offers you something, a Hershey kiss or something, and that action triggers a feeling of reciprocity. LinkedIn has a great example of reciprocity on their site: a feature called Recommendations. What LI has seen is that when someone gives someone else a professional recommendation, the probability that they’ll get one back is extremely high.

Sieling: When you’re able to show the source of the causation it facilitates reciprocity.

Messina: .. it also can lead to people spamming to get the reciprocity.

Porter: There’s a notion of indivdual usability, how useful is the site to me, but also group usability, how useful it is to the group. Top diggers was axed in favour of usability to group..

Burka: I want to talk about privacy and community. When I first signed up to Facebook, the first thing i did was find all my friends. A few of my friends wrote on my wall. I’ve never lived in a dorm room, and I’m not even sure if we have dorm walls where I come from; I didn’t get the reference. I deleted all the messages on my wall cos I’d read them already, which was really antisocial cos the whole point is to use the wall to serve the group, not the individual. Community is such an extroverted idea that in order to develop a real community that operates on trust, you have to understand the privacy and operate within it in a trusting way.

It’s interesting that at one end of the spectrum you’ve got very privacy-oriented sites (basecamp, pownce) where privacy levels are clear; on the other you have public sites where everything is seen by everyone (digg, Metafilter). Everything you do on digg is public and that’s intentional. Facebook falls right in the middle; some aspects like wall are very public and once you learn that, you know that friends/fof can see it, but then there’s news feeds, PMs etc… At either end you have lots of clarity, but the closer you get to the middle you have to be more careful and explicit about who can see what, cos there’s much less clarity.

One of the most important things to understand about privacy is to understand your users. Online identity: which parts are important to them? Images, names are very loaded. Originally on digg we didn’t even require names; we left the field wide open. pownce is more about people than digg, and this has been a contentious issue on pownce – we swayed towards the private, and unless you have a mutual relationship with someone you have very limited access to their details. This is restrictive from a community perspective; it’s hard to identify people.

Another hotpoint is SNs themselves. on most SNs we allow everyone to see friends’ friends lists (not like real life – you can’t see social clouds of random people in the street). On pownce we have ‘first name last initial’ rule to provide some privacy. ‘Shout’ feature has also been major area of contention, cos it’s public and everyone can see shouts (it’s shouting, you’d think this would be clear, but..)

When you digg a story, we’re careful to show who dugg which story. Most activities you can do on digg can be seen publicly, but tracking site activity, especially on external websites like Facebook (Beacon), is becoming more of a hot topic. Beacon is a useful feature but that’s consolidating public activity in one place.

Sieling: People are developing a gut-level instinct on what kinds of behaviours are being shared; things like beacon take private acts and make them public acts; this makes people less comfortable with the system.

Burka: Build in tools for people to control access to their info. Any time you come up with a crutch of making something a ‘preference’, it’s dangerous. (firefox v seamonkey.) At digg we’ve already gone down the road of providing too many options; at some point people stop understanding all the options. If you’re building a site, it’s harder to take away preferences you’ve added so add as you see a need for them.

Messina: You need to add value, new features, to keep users from getting bored… but that [complexity] might cost you new users.

Burka: Hopefully users wouldn’t get bored on digg [laughter]

Messina: Isn’t that what digg is all about? [laughter]) WordPress has been a great platform for blogging, but as it moves towards SNs its focus is starting to blur. How do you balance tradeoffs of adding new features without diluting your purpose?

Burka: Hide some of the features. Put some of it under the hood so it’s not immediately appreant to new users, but they’ll realize over time there’s more good stuff there. As users become more confident they get interested in learning more.

Porter: Facebook is suffering from feature overload now, so they recently launched tabbed profiles to alleviate the strain.

Burka: The last thing I want to talk about is the importance of transparency. One of the things Josh was alluding to is that hopefully you show and don’t tell. On digg it’s clear at the point of posting who can see what you’re about to publish. Particluarly in the context of where you’re contributing to the site, it’s important to understand who can see that. The more transparency the better.

Summing up: When you’re building a site, if you lean toward either being very public or very private, things are simpler. Be nuanced and sensitive to people’s personal info. Give control to users about what info they share, privacy levels, but limit yourself – don’t use preferences as a crutch. Be transparent; show it, don’t just talk about it.

[Trouble with slides. "Someone's twittering this - and if they're not, they should be!"]

Sieling: Something I’m gonna be talking about is spam. With Ma.gnolia we didn’t give too much thought to what would happen if we made it available to everyone. People came with intentions we didn’t anticipate, but we knew them when we saw them. We found what we think are a fair balance of solutions.

Spam isn’t just a drag generally, it’s a drag on the community. Spammers love social software, and the tools we use for good are the tools spammers use for bad. They love tracing their actions and seeing the effects, cos ithat teaches them how to be more effective. 75-80% of new accounts made on Ma.gnolia are made by spammers – which seems a shocking number – and they manage their work in lots of ways.

The ways you set up a social system will be more or less friendly to spammers, but you don’t want to end up cutting off your nose to spite your face. How do spammers work?

• One site, many accounts. Spammers want to get into the system and get the most eyeballs
• They take one or two legit links, bookmark them and fil the rest with spam to achieve (minimum) credibility
• Joe SEO – get rich quick! They’re ‘helping people’ by sharing info, morally neutral. They don’t realise they’re being annoying
• You Can’t Fool Me – put a lot of effort into impersonating community members. Hope to leverage credibility within community to get eyeballs
• Importing volume links: long lists of bookmarks slammed into every account a spammer has

These are wily problems to get around – you don’t want to axe good features just to contain spam. We realized we weren’t going to win the war on spam; we had to accept that it’s part of the ecology and we just need to minimize its damaging influence and protect our users.

What didn’t work?

• No-follow: Google to ignore destination of a bookmark; this removes incentive to use social bookmarking. Doesn’t seem to have any effect. [Burka: digg did focus group and had a couple guys define themselves as SEOs and they asked immediately about no-follow.] If you get the question, it provides an opportunity to educate SEOs [laughter].
• Akismet: fantastic for WordPress. Applied this at account level to flag possible spammers; legit links would throw it off, it required more post-processing of A’s work and the situation got cloudy. Might have worked with more effort, but we decided we needed human intelligence
• Weed-on-sight: If person sees spam, click to hide
• Recaptcha: No effect. People paid people in lower economies to do human work of adding accounts

What did work?

• Accept there’s no 100% win against spam
• Gardeners: humans – enabled trusted members who move accounts on and off whitelist; not job, contest or vendetta. Gardeners make new ones, use network for good. They’re altruistic volunteers [Messina: I haven't met any altruistic users; the expectation of getting nothing in return is incredibly rare. Porter: Why do people submit reviews (restaurants, for instance), to help other people? Users often mention not wanting people to make the same mistakes, or wanting to share good experiences. If you don't ask a follow-up to that and they keep talking, they talk about how many people they're helping, how many eyeballs they have. Are there other draws of being a gardener?] The people using Ma.gnolia aren’t there to waste time; true altruism isn’t really it. They want to make sure the system is clean and useful to them.
• Whitelist, with shade of grey (similar to nipsa on flickr)

 

 

 

Q: I’m a gardener [laughter]. Altruism is part of it, but there’s a sense of ownership too. I discovered it really early on, and I had a feeling of wanting to keep the group clean and I wanted to retain high quality. Ma.gnolia makes it easy to give back to something I get good service from. I can rely on that service and feel responsibility to give back.

Sieling: Don’t only look to tech solutions to solve these problems; sometimes you need human intelligence.

Q: There’s an effort of companies to try to monetize SNs. Is there a legit way to do this or is it all spam?

Sieling: Ma.gnolia has ads and it’s largely about how you place them and what they’re selling – do they fit the community and the site’s experience. People are looking how to leverage SNs but nobody’s really sure how right now.

Porter: Monetization needs to be indirect. Passionate users need to get more recognition and bring others into the fold. You can’t set up a system so that every activity generates money; it’s about building and supporting culture.

Burka: Community will grow things for you.

Porter: More money will be the fallout of increased passion.

[Have to wrap it up; will take questions informally in the back.]