[This post is a response to a thread on the ONA discussion list. It got too long to email.]

When talking about the best use of technology for uses like computer-mediated communications, a skeptical philopsophy is invariably voiced along the lines of "technology can't solve all problems; humans can." This is sensible, but the statement is problematic due to a different understandings of what exactly "technology" means in this sense.

To wit, it can mean:

i) A fancy, shiny marketable thing which claims to solve all problems.
ii) A set of standards, protocols, conventions, agreed upon by a large number of people and which has a written governing process for evolution.

The skeptics often think of it as (i), as that describes a lot of software technology, but the case for (ii) has been made convincing by Larry Lessig in Code and Other Laws of Cyberspace— by far the most important book about the Internet.

Understanding Code

Lessig breaks down Code as four forces which regulate behavior: Legal, Economic, Moral, Architectural.

What is different with architectural code (things like the environment, software) is that they cannot be broken, whereas the other three all assume punishment for enforcement.

His secondary thesis is that because of the robustness of software code, there should be a more understanding by civic leaders of what the code does, and a demand to be involved in the code design process. This is why code should remain open– open source, open documented, open designed– so that it does not something forced onto its users.

So if we substitute code for technology, we can now avoid making judgments about technology as an actor. Instead, code is the object which we ourselves control:"We use code to solve our problems." We create financial incentives to reward behavior; we develop laws to criminalize it; we develop a sense of ethics to foster the type of behavior we want. And most important, we build our software systems to be able to facilitate the process. All of these are fundamentally human behaviors.

So think about effective governance meetings. You might not call Robert's Rules of Order a "technology" but you would certainly agree that it is a code. Ditto for court proceedings, legislative hearings, the editorial process, etc. If these can work in the offline world, it's possible to code discussion in the online world to work this way.

Goals for Online Conversations

The question is not whether to have online conversations on a news site (as the panel of purported "experts" batted around); it's what type of conversations to have. Here are two such types:

• There are social conversations which exercise the emotions.
• There are constructive conversations which lead to judgments or decisions.*

The social model is tempting, and it appears to "work" in places like the Daily Kos– if we measure success by controversy, numbers of posts, size of audience attracted– until, of course, emotions turn against you. It happens in more places than the Post. In October I wrote The Scales of Discourse about a case in Delaware where anonymous posters had smeared a local city councilman and his wife; the state Supreme Court ultimately ruled that the speech was not defamatory since no one would take Internet speech seriously. ("Well, it got the community talking," a local columnist explained to me).

Nonetheless, If the Great River of Id model works for the Daily Kos, or for AM talk radio, so be it. Render unto Kos what is Kos's.

Judging by the comments on the ONA list, it appears that the constructive approach is preferred.  (Tom M.: "More is not always better, and quantity does not equal quality."; Steve Y.: "We have an obligation to help people understand one another." Vin: "What we need are policies and practices to govern how our readers utilize these online technologies.") And the corresponding aquatic metaphor may be a swimming pool. There are dedicated swimming lanes, lifeguards on duty, boundaries clearly marked, rules posted, no diving in shallow end.

Returning to the path of rationality

The Blogging Priests assert "the age of controlled conversation is over. The age of open conversation is here. But that is damned hard for the controllers to get used to." No, that's just fear mongering. It's a statement from a mindset of unworkable "technology" instead of controllable code. It's not that the publisher wants to control the message; far from it. It's that they can control the flow of the discussion. And there's many different mechanisms that could be done. I'd like to bring up one such example– developed at Harvard Law's Berkman Center before they got into blogging. It's called the Rotisserie. This excerpt captures its essence; emphasis is mine.

The Rotisserie implements an innovative approach to online discussion that encourages measured, thoughtful discourse in a way that traditional threaded messaging systems cannot. In contrast to the completely asynchronous, broadcast-to-broadcast mode of existing threaded messaging systems, the Rotisserie adds structure to both the timing and the flow of the discussion. The timing of the discussion is broken into semi-synchronous rounds. Users are allowed to post responses at any time, but their responses are not published to other users until the deadline for the current round passes. This structure allows users to put significant thought into their responses rather than competing with other participants to post first. More important, this structure allows the system to control the flow of the discussion by distributing responses to specific users for further discussion at the end of each round, ensuring that every post is distributed to at least one other user for comment and that each user has exactly one post to which to respond.

I'm not saying that this is best example for a large newspaper. But it is one way of how to handle the scaling issues that arise with hundreds or thousands of readers wanting to reply on a single thread.

What the Post really needed in this case was a rational conversation model. A reader complains of an error. Make a note of it. Allow other commenters to mark agreement (see ViewPoints). Should this get to a certain number of plaintiffs, by all means, trigger an internal alarm and get an editor to comment back. It's the difference between a petition and a rally– the latter should only be necessary when the former doesn't work.

I myself have submitted a couple corrections to the Times back in November. One took ten minutes: the columnist was able to correct it on the web right away in the wee hours of the night. On a separate occasion, it took the paper eleven days to admit that it failed to disclose a conflict of interest of a contributor. I didn't summon a lynch mob, though in the second case I wished the website had marked the column with some sort of flag. (and to his credit, the contributor acknowledged the disclosure after I posted to his blog).

Yes, this is all very systematic. I work in IT after all. But hey, a year ago, Scott Rosenberg of Salon passed along an column from the Sacramento Bee alerting readers that several newspapers had begun using bug-tracking systems. I saw little follow-up around this on the web from the Blogging Revolutionaries. The social conversation is much more important to them.

And that's how it's been; for the last few years, there's been a push towards the social aspect of conversations. I've documented the shift in The Yin to Social Software's Yang. It's not the result of inevitability, it is instead, as Lessig informs, the result of specific decisions made by software designers, and those tend to be reinforced by the "inevitable" crowd. But if that's not how we want it, we can pick a path of our choosing– and write our own code.

Updated, April 10, 2007— The second point I had originally written as "rational conversations which exercise political actions." Those words can be interpreted too widely, so I updated that to reflect what I certainly wanted to say. By "rational," I mentt constructive; by "political action" I meant judgment or decision-making.