awesome comments

Post on 13-Nov-2014

7.243 Views

Category:

Design

0 Downloads

Preview:

Click to see full reader

DESCRIPTION

 

TRANSCRIPT

Xianhang ZhangBumblebee Labs

1iSchool Research Conversation - October 24th 2008

2 person team

4 months old Goal is to build “provocative social

software” 2 product releases, 1 more in the

pipeline Happy to talk more after this about

collaboration.2iSchool Research Conversation - October 24th 2008

3iSchool Research Conversation - October 24th 2008

For many people, commenting is their first

and possibly only foray into the social web

4iSchool Research Conversation - October 24th 2008

“Communications tools don't get socially interesting until they get technologically

boring”– Clay Shirky

iSchool Research Conversation - October 24th 2008 5

There has not been much innovation in the comment space

Slashdot pioneered their “Karma” system in 1997

To this day, I’m not aware of any other system of that level of complexity

most major sites are still using the most basic commenting systems despite their obvious flaws

6iSchool Research Conversation - October 24th 2008

iSchool Research Conversation - October 24th 2008 7

8iSchool Research Conversation - October 24th 2008

Signal/noise ratio is too low

Insightful comments get lost

Lack of readership due to low quality

Hard time keeping track of the narrative thread of the conversation

Despite this, people yearn for conversation (Herring)

9iSchool Research Conversation - October 24th 2008

“When all you have is a hammer, all you can see are nails”

Sociability problems CAN be fixed through design. But…

They cannot be fixed through Interaction Design, a new discipline is needed

iSchool Research Conversation - October 24th 2008 10

As HCI researchers, we have been neglecting the development of tools to help designers solve

sociability problems

iSchool Research Conversation - October 24th 2008 11

Small, personal blogs read largely by friends and family

Social networking appsHighly targeted blogs aimed at a

professional audienceSites cultivating return audiences

rather than drive by, search engine traffic

12iSchool Research Conversation - October 24th 2008

Large sitesSites with high traffic per pageSites with low traffic per pageEntertainment sitesSites with an influx of new usersSites with controversial topicsSites with low community

involvement

iSchool Research Conversation - October 24th 2008 13

Poor explanation:

There’s a lot of stupid people out there

14iSchool Research Conversation - October 24th 2008

Better explanation:

15iSchool Research Conversation - October 24th 2008

Human behavior is governed by sets of constraints

iSchool Research Conversation - October 24th 2008 16

Laws of PhysicsLaws of Physics

Evolution/GeneticsEvolution/Genetics

Cultural Norms/Legal SystemsCultural Norms/Legal Systems

Contextual NormsContextual Norms

Other PeopleOther People

Free WillFree Will

= Hard constraint

= Soft constraint

iSchool Research Conversation - October 24th 2008 17

With virtual systems, there are an additional set of constraints:

iSchool Research Conversation - October 24th 2008 18

HardwareHardware

InterfaceInterface

SoftwareSoftware

Social SystemsSocial Systems

= Hard constraint

= Soft constraint

Individual constraints can be accomplished at different layers

iSchool Research Conversation - October 24th 2008 19

20iSchool Research Conversation - October 24th 2008

Works Doesn’t Work

Personal Blogs Large Sites

Social Networks Mostly new users

Professional Audiences Entertainment Sites

Returning Users Low community Involvement

Comments work when there are strong social constraints and weak interface constraints

iSchool Research Conversation - October 24th 2008 21

HardwareHardware

Interface

SoftwareSoftware

Social Systems

= Weak layers

= Strong layers

Comments fail when there are weak social constraints and weak interface constraints

iSchool Research Conversation - October 24th 2008 22

HardwareHardware

Interface

SoftwareSoftware

Social Systems

= Weak layers

= Strong layers

Could Comments work when there are weak social constraints and strong interface constraints?

iSchool Research Conversation - October 24th 2008 23

HardwareHardware

Interface

SoftwareSoftware

Social Systems

= Weak layers

= Strong layers

Increasing barriers to participation (registration, captchas etc.)

ModerationDisemvowellingROBOT 9000RatingSorting/Filtering

24iSchool Research Conversation - October 24th 2008

For conversation to happen, there must be a narrative thread.

If I refer to Comment X, you must have read Comment X to understand my reference

Rating systems must keep the chronology intact while making highly rated comments visible

iSchool Research Conversation - October 24th 2008 25

This is most often done via threading

Heavyweight approach, lots of intricacies

iSchool Research Conversation - October 24th 2008 26

Initial site developed over 12 hoursVery simple concept

27iSchool Research Conversation - October 24th 2008

28iSchool Research Conversation - October 24th 2008

iSchool Research Conversation - October 24th 2008 29

How do we prevent this site from being just a flash in the pan?

iSchool Research Conversation - October 24th 2008 30

Build remarkable content

Remarkableadjective Pronunciation: \ri-ˈmär-kə-bəl\

Something which people will remark about.

31iSchool Research Conversation - October 24th 2008

Initial user impressions are generally very positive but the site can quickly lose it’s appeal

In order to support long term growth, we needed to build conversation around each question

Design constraints: Anonymous everything Heavy reliance on external/search engine traffic Many questions, long tail of quality Primary navigation mode is random

iSchool Research Conversation - October 24th 2008 32

Large sitesSites with high traffic per pageSites with low traffic per pageEntertainment sitesSites with an influx of new usersSites with controversial topicsSites with low community

involvement

iSchool Research Conversation - October 24th 2008 33

Commenting system designed from the ground up.

Goals Drive out poor comments Support remarkable content Support real conversation Give people a sense of ownership

34iSchool Research Conversation - October 24th 2008

Every comment has vote up and vote down buttons

35iSchool Research Conversation - October 24th 2008

New comment Registered users: 2.0 points Unregistered users: 1.0 points

Vote button Registered users: 1.0 points Unregistered users: 0.1 points

36iSchool Research Conversation - October 24th 2008

Progressively degrade comment readability at lower scores

Show undegraded comment to original poster

37iSchool Research Conversation - October 24th 2008

Voting is unbounded both up and down but…

Makes it easy to find and promote remarkable comments

We are using the lack of information as a design feature

Actual Score Displayed Score

-2.3 -2.3

1.7 1.7

3.5 3.0+

98.6 3.0+

100.1 Awesome

38iSchool Research Conversation - October 24th 2008

Non threadedComments are sorted by displayed

score, then by date postedReply button on all comments

(coming!)All replies to your comment are

highlighted

39iSchool Research Conversation - October 24th 2008

Original posters are highlighted

Identity can be established while maintaining anonymity

iSchool Research Conversation - October 24th 2008 40

Visiting a question/comment you made automatically upvotes all it’s children

Returning to a question/comment rewards your repliers

You are personally in charge of voting your repliers correctly

41iSchool Research Conversation - October 24th 2008

All comments at 3.0+ are displayed in chronological order

Conversation can happen at 3.0+This is the default if you post as a

registered user (2.0) who replies to a poster who regularly returns (1.0)

42iSchool Research Conversation - October 24th 2008

Commenting systems are generally horribly designed from a sociability perspective

Better conversation is a force multiplier for the social web

Awesome comments is a ground up rethinking of how commenting should be implemented

Still very much an experiment, details will change as the system matures

43iSchool Research Conversation - October 24th 2008

Bumblebee Labs: http://blog.bumblebeelabs.com

Am I Normal or Not?: http://www.aminormalornot.com

Email me: hang@bumblebeelabs.com

Need a designer? michael@bumblebeelabs.com

iSchool Research Conversation - October 24th 2008 44

top related