Summary of "Web 2.0 Expo NY: Clay Shirky (shirky.com) It's Not Information Overload. It's Filter Failure."

Clay Shirky: From “Information Overload” to “Filter Failure”

Clay Shirky argues that the popular idea of “information overload” misses the real problem. Using the long-running “up and to the right” information-growth charts as a starting point, he says information has always been increasing—yet the lasting, practical issue isn’t sheer volume.

Instead, it’s that the filtering systems society relies on to manage information flows are failing.

From “Overload” to “Filter Failure”

Shirky frames “information overload” as a recurring story—dating back centuries and appearing in journalism for decades—that persists even though it is predictable.

He traces today’s environment back to the economics of earlier media:

In the printing era (Gutenberg):
- Publishers filtered content because publishing involved upfront risk and costly production.
In the internet era:
- The cost to produce and publish collapsed, so filtering can no longer be safely assumed to happen “upstream.”

Therefore, the real shift isn’t a flood of information, but a collapse or break in downstream filters—including both automated and social filtering.

Evidence via Spam: Not More Data, Broken Systems

Shirky uses spam as a concrete example:

His perception that spam doubled didn’t match measurement—spam he had to delete increased by only about 25%.
The major change was that his existing filtering approach failed when spam volume rose enough to “break” the system.

He generalizes the lesson:

Effective filtering requires multiple layers (automation + user intervention).
Filters need continuous retuning—there’s no “set it and forget it.”
The problem is mainly about incentives targeting people cheaply (spam), not about unmanageable information volume.

Privacy as a Filter-Design Problem (Not Just a User Mistake)

Shirky recounts a story about a failure in an outbound information flow:

A friend changed her Facebook relationship status (engaged → single) after ending an engagement.
Even though she used Facebook’s privacy settings tools and understood them, the status change broadcast immediately to her entire network—including her ex’s friends—creating a “privacy meltdown.”

The key interpretation is that the failure isn’t only “bad settings.” Instead:

Privacy preferences are unnatural for users because they don’t match how privacy historically worked.
Historically, privacy relied on inconvenience and shaped social behavior—communication was hard to do publicly at scale.
Modern systems make privacy explicit and engineered, requiring careful management that people aren’t trained for.

Conclusion: The privacy challenge is about designing filters that match real social needs—not about perfecting information management at the source.

Institutional/Organizational “Filter Clashes”: Facebook and Academia

Shirky then describes another case showing confusion about the direction of information flow:

An 18-year-old created a large Facebook study group for a chemistry class (“The Dungeon”) and allegedly triggered campus discipline because Ryerson treated it as cheating.

His analysis:

Colleges operate in two different modes:
- internal scholarly conversation
- external quality-control credentialing
The practical “hassle” of organizing meetings keeps these modes from colliding.
Facebook merges them:
- the institution treats the activity like media-like publishing/credential control
- the student experiences it as an extension of group conversation

He argues that the conflict is less about access to information and more about flows, incentives, and “free-rider tolerance” in large groups.

Finally, he suggests that neither side can be fully “right” under the old metaphor of media vs. real life:

Facebook requires rethinking institutional design to accommodate both conversation and individual effort.

Practical Takeaway

Shirky warns that focusing on “overload” distracts from the real design question. The useful question is:

“What filter just broke?” rather than “What changed in the information volume?”

Some fixes are technical (e.g., search/ranking, tagging, moderation), but deeper fixes require adjusting social norms and institutional structures—creating new “filters” rather than merely updating old ones.

Presenters / Contributors

Clay Shirky (speaker)
Tim (mentioned as “Tim backstage,” likely a contributor not otherwise identified)
James Grimmelmann (privacy analyst mentioned)
Chris Avanier (student case subject)
Zak Rabine (quoted)
Gutenberg (historical figure mentioned)
Varian (mentioned in relation to an information-growth chart)
Peter Lyman (mentioned in relation to an information-growth chart)