The Problems of Scaling and Scoble at Quora
January 26, 2011 § 31 Comments
By: Dan Kaplan
In the weeks since the influx began, the distribution of brilliance in the Quora feed has precipitously thinned. When I first joined the site, in February 2010, it was like gulping from a firehose of awesome. In January of 2011, it’s like sipping it through a narrow straw: the awesome still ekes through, but I have to work harder to taste it.
Under the pressure of new users, Quora’s answer voting system is also breaking down. Always vulnerable to cliquishness and groupthink but mostly held in balance by the icy standards of the early community and the Admin police, the up- and down-voting mechanism is increasingly unable to guarantee that the best answers will float to the top.
Indeed, it seems that in these transitional times, the best way to amass the most up-votes on prominent questions involves inserting photos of dubious value and composing highly digestible morsels of mass appeal. Giving yourself credit for the photo you inserted and tweeting your answer to a swarm of loyal fans who followed you to Quora from Twitter certainly doesn’t hurt.
The change in the site’s nature has aggravated some of Quora’s most dedicated early users. In backchannel discussions, members of this group have bemoaned the encroachment of Social Media Experts and the tangible decline in the quality of discourse across the site. They worry that amidst the cacophony of mediocre voices, the elites (whose answers are responsible for the majority the attention Quora gets from the press) will increasingly neglect or even abandon the community.
The danger of losing Quora’s elite core is not lost on its founders. A week after the appearance of the Mashable list, co-founder Charlie Cheever wrote the second post in as many weeks discussing the team’s efforts to maintain content quality in the face of onrushing Social Media Expertise:
The most intriguing of the three steps he outlined – which also included boosting the ranks of admins and better educating newcomers on how shit goes down in the hood – is “an algorithm to determine user quality…somewhat similar to PageRank” but not quite, “since people are different from pages on the web.” Shervin Pishevar called this concept “PeopleRank,” and why not.
A reputation system that consistently elicits brilliance from the elite core yet doesn’t alienate new users would be a marvel of engineering and interaction design. But the PeopleRank concept raises a veritable fuck-ton of issues. Some of these issues are Google-hard Computer Science problems: how do you teach a machine to recognize legitimate expertise across a theoretically infinite range of topics and questions? Others are epistemological, because what defines legitimate expertise, anway? And who gets to decide? Both the technical and philosophical questions get at the heart of what Quora is and what it could become.
Quora as it currently stands is an increasingly chaotic but utterly democratic bazaar of brain output, where every up- or down-vote counts the same, regardless of the topical sophistication or motivation of the user clicking the arrows. My vote on an answer about scaling MySQL databases carries the same weight as Adam D’Angelo’s, despite the fact that I know negligibly more than squat about SQL and D’Angelo is a technical wizard.
An accurately PeopleRanked Quora could solve this problem. It would also be inherently undemocratic and elitist, potentially hindering Quora’s ability to penetrate the mainstream. But if the goal is to populate a database with brilliant answers to an ever-expanding set of questions, perhaps elitism is a legitimate position for the short and medium term. After all, the lasting value of Quora’s database hinges on the site’s ongoing ability to attract the planet’s brightest and most articulate people on any topic, extract their best insights and make them rapidly accessible to the curious of mind.
This is a worthwhile goal, but achieving it with PeopleRank means revealing to the hoi polloi precisely where they stand.