Randall Farmer

Thirty Years of Social Computing: Are we finally ready to scale?

Biography
For more than 30 years, I’ve been connecting people with each other using computers as the mediating technology. I co-created: one of the first forums, the first Trek MUD, the first graphical MMOG with the first avatars, the first virtual MMOG currency, the first virtual information marketplace, the first fully distributed virtual world platform, the first no-plugin web session platform, Yahoo’s 360 project, and more. Now I’m working on other far-reaching social computing projects for Yahoo! as the Community Strategic Analyst.

Position Paper
Massive-scale articulated social systems create a completely different of value propositions and risks than those that are faced by smaller experimental/academic projects and start-ups. Indeed, as , there are some good reasons for many social software applications (groups, wiki, calendars) to ‘stay small’ and limit the number and nature of contributors. None the less, we need to address the social issues raised at much larger scale (thousands to millions of users.) Applications like Search and eBay require a universal aggregation points for managing user and corporate generated content.

Where reasonable, how do we make social software work at this scale? When you involve so many, how does the application change? Previously these questions were either ignored, or dismissed with statements such as “we should be lucky enough to have such problems.” That has all changed, with the large portals (Yahoo!, MSN, AOL, Google) entering the social computing space with a vengeance as they move from content aggregation alone to community editorial of aggregated professional and user published content.

Scale produces low-cost, high-value information markets
As social software gets adopted at scale, it generates information markets with very real commercial value. At their best these markets match producers and consumers efficiently. At worst, these systems bear a large burden of social abuse for commercial gain: email brought us spam, blogs bring us track-back/comment/referrer spam, and online markets battle fraud and intellectual property abuse.

MMOGs blaze one trail: Virtual Economies & Intellectual Property
Games are driving a lot of the hard virtual-world/real-would interaction questions faster than many ‘social computing’ folks may know. At a recent conference, lawyers were already talking as if the secondary virtual item market (eBay) for virtual goods implies real world value and that implies that they are real world property and, by extension, subject to real world property laws, such as estate liquidation. No kidding, a lawyer was actually thinking that an Epic Mount in Worlds of Warcraft might be an estate asset if his client dies!

Social Networking blazes another: Identity & Abuse
Early articulated social networking software has demonstrated how rapidly a million-user community can get grown via network-effects. Unfortunately, this very effect demonstrated a weakness when the community scaled: Too much public information, too easy to abuse, weakened social boundaries (friendster vs. friend), not enough services to leverage your network, and not enough control over who can see which fields.

Though these problems were seen fairly quickly, effective solutions were slow to arrive, are spotty at best, and some proposals (like FOAF) don’t seem to help much at all.

How do we socially scale micropublishing to millions?

“A good name, like good will, is got by many actions and lost by one”.

— Lord Jeffery

For social computing to scale to millions of users, we need to hand control over disclosure to the users, but also put it in a context of trust.

Consider the example of the evolution of web search. Originally we all tagged our own pages with META keywords, following the pattern of NetNews. Early search engines read these tags and used them to produce search results. This information market had very real value as dotcom start-ups competed for eyeballs. Next thing you know, every startup was putting SEX in their keywords list. The solution was PageRank: don’t trust the page owners, trust the anchor text used by the webmasters of referring pages to describe the page content.

PageRank is the first massive social reputation system to deal with the first massive scale information market. It is the first trust currency on the internet.

Clearly, we need to develop more reputation systems for applications other than search. EBay, NetFlix, and SlashDot are just scratching the surface.

Of course, reputation systems produce their own information markets, ripe for abuse. We still have a lot of work ahead of us.

Back to Social Computing Symposium 2005