There are a few basic points I’m not seeing made quite often enough in the ongoing thumbsuck about the future of journalism. I take for granted we all understand that whether “newspapers” survive (either as ink-on-wood-pulp or as institutions) is of no real intrinsic importance; the question is whether vital forms of journalism will get done. But we can break that down into a couple of more specific questions. We can ask which aspects of journalism actually require dedicated professional practitioners and/or institutional support, and which can be supplied on a distributed basis by amateurs or volunteers. Next, we can ask which types of journalism will be adequately supplied by the market as traditional cross-subsidies (i.e. classified ads paying Woodward & Bernstein’s salaries) break down, and which will tend to be undersupplied given the significant positive externalities of an informed populace in a democratic society.

To get a handle on the first question, we need to know something about the conditions under which, in general, peer-production models (Linux) as opposed to standard market models (Windows) are viable for information goods. The best guide here is unquestionably Yochai Benkler, who lays out a general theory of peer-production in The Wealth of Networks—and who has also helpfully (albeit briefly) weighed in on the question of journalism’s future. Two central points cribbed from Benkler are germane here.

First, information production actually consists of a series of conceptually discrete functions of different types. Traditional newspapers and magazines (and online magazines, for that matter) normally bundle together not only different types of journalism, but several very different information production functions: identifying important stories, assigning them to researchers and writers, compensating those researchers and writers, providing editorial and quality control for stories submitted, compiling stories and conveying them to readers, and certifying those stories by providing the publication’s imprimatur as a reputation-backed guarantee that the information therein was (reasonably likely to be) accurate, well-written, and of interest or importance to the reader. We can probably thin-slice each of these categories into further subcomponents. For a variety of reasons unnecessary to belabor, there were good reasons to integrate these diverse functions under one roof. As technology lowers the costs of aggregating dispersed efforts and information, that will not always hold true. Moreover, there’s no reason to suppose ex ante that the same production model will be right for all stages of the process, or for the same stage in different instances.

Just to throw out some possibilities. Writers and reporters can work on salary; can be freelancers working for an agreed fee; can post articles on their own sites in hopes of garnering enough attention to make decent ad revenue; can submit their articles to a site that intermingles their own Google Ads with the writer’s (supplying many more eyeballs in exchange for what amounts to a cut of the ad revenue); can work purely for fun or fame either to produce whole articles or small pieces of writing or reporting that are later integrated into a larger work. Freelancers can self-assign to topics; editors can give staffers assignments, readers can either formally or informally vote on the topic an author should tackle next; sites could offer bounties for the best (reader or editor selected) spec article on a given topic. The New Yorker can hire fact-checkers and then place its imprimatur on a piece to certify its accuracy; dispersed swarms of bloggers can fact check individual elements of a piece and draw attention to any inaccuracies they find; individual authors can accumulate either informal credibility or some kind of formalized karma-rating over time (either via a site’s internal system or some Net-wide monitoring and rating service).

Since the whole reason we’re having this conversation is that Net-fueled fragmentation has destabilized the traditional revenue models, it seems unhelpful to then talk about the future of “journalism” tout court. Rather, we need to think about the circumstances under which each component of the information production function is viable for a given production model, either independently or in a particular combination with other components. Which are individually self-supporting? Which require cross-subsidy? Which don’t work at all in specific contexts?

The second big idea we want to draw from Benkler is that the keys to making peer-production work are granularity and modularity. A granular task can be thin-sliced into tiny pieces, each of which demands relatively little effort—little enough that you’ll find many people prepared to make a contribution without necessarily needing to be paid (much) for their trouble. (Think of folks correcting a single typo on Wikipedia just to be helpful.) A modular task can be broken into chunks that can be dispersed, completed independently (and perhaps without much active collaboration between the people working on them), then reintegrated into a finished product. An encyclopedia is relatively modular and granular, whereas a novel normally isn’t—hence, Wikipedia works, whereas online collaborative novels have mostly been horrific train-wrecks.

Different functions will have different levels of granularity and modularity, and therefore be differentially amenable to peer production as opposed to “expert” or “professional” production. Copy editing—checking for spelling and grammar errors, cleaning up clunky sentences—is usually both granular and modular. What we might call structural editing—seeing how a piece hangs together and how it flows from start to finish, whether it gives the reader all the relevant information, has a balance of differing views when necessary, and gives the proper weight and ordering to the different components of a story—is normally neither granular nor modular. Fact-checking and reporting can go either way: The task of gathering or confirming quotations from sources (whether texts or interviews) may be modular, but is not usually especially granular. For a statistics-heavy story, it may be granular (each of a dozen numbers in a paragraph can be independently verified) but not modular because consistency across them is important: Are they computed by the same method? Are dollar figures inflation-adjusted to the same year? Are they being combined in a misleading way? You might be able to split the job between several people, but they all basically need to be on the same page and paying attention to each other, or you’ll end up having to do it again from scratch at the quality-control stage.

Obviously, this is a somewhat loose way of speaking, since modularity and granularity will be defined interdependently: You might have a choice of different ways to carve up a larger task that affects how finely divisible or interdependent the different subcomponents are. Moreover, as the examples above should make clear, the modularity and granularity of some tasks will depend a great deal on the particular story. Sometimes effectively reporting or researching one aspect of a story—knowing what questions to ask, knowing where to look and what information is relevant when you find it—means being intimately familiar with all the other aspects. Sometimes not so much.

So keeping that schema in the back of our heads, we can start in on the second question: What types of reporting do we have reason to worry will be underproduced without the old-fashioned bundling and cross-subsidy? People are, naturally enough, focusing their concerns on types of journalism that are seen as generating significant positive externalities, especially relative to their intrinsic interest. So, for instance, most of us may not especially want to read (or, a fortiori, pay for) an article exposing waste and corruption in a government program, unless it’s of a particularly sexy or scandalous sort. Even when a story is salable or sexy, most of us probably don’t care a great deal whether we read it from the original source or not, making it harder for publications that invest in the intensive up-front investigation to recoup that investment when competitors without that overhead can easily and quickly free ride on those efforts. Still, most of us probably do want to live in a society where those articles get reported and written, as a means of holding public officials accountable.

None of this is a problem for domains and production functions where the amateur or peer model is able to pick up enough of the slack that we get an adequate supply of these public goods. So, for instance, Jesse Walker suggests that “gadflies”—the folks who make it a personal mission to show up at board of ed hearings and town council meetings—will fill the lacuna in local coverage left by dwindling reportorial staffs. When we want in-depth analysis of a public policy issue of importance, it may fall to academics to put their thoughts out in a form suitable for public consumption. Much research and fact-checking can be done by the cloud, while various types of social networking software help to certify stories as relevant or credible to particular readers. If you pay any attention to new media enthusiasts, most of this rosy vision should sound pretty familiar.

But with this groundwork laid down, it should also be easier to identify the spaces where there is some reason for concern with a bit more specificity. Instead of talking about “journalism” or even “investigative journalism,” we should be talking about discrete stages of the production process for specific types of stories. What we need to worry about, I want to suggest, are particular production functions that are :

• insufficiently modular/granular, or not linked to a sufficiently large user base, to be amenable to distributed peer production
• produce positive externalities that are hard to monetize (because the benefit doesn’t accrue directly to the consumer and because facts aren’t copyrightable)
• Aren’t likely to be supplied adequately on a voluntary or subsidized basis

That last one’s worth bearing in mind, though I haven’t said much about it. It’s always been my understanding that most political magazines don’t actually make money, but keep putting out issues because there are people who care enough about having this or that perspective represented in public discourse to underwrite it. There are all sorts of interest and advocacy groups working to draw attention to legislative or policy developments they support or oppose—Benkler mentions some of these in his New Republic correspondence. That said, I’m perhaps naive enough to think it’s still orth having cadres of people, at the local as well as the national level, who are at least aspiring to do something that looks like objective reporting.

With these features in mind, we can zero in on the specific functions and spheres that are going to have trouble self-sustaining and worry about how those might be enabled. In each such case, we can ask: Is there some mechanism available by which to internalize more of the public benefit? Is there some way to re-divide the functions so a problematic one becomes more modular and granular? If there are only a few key stages in the large process that get undersupplied by either peer or market mechanisms, might it turn out that charitable or nonprofit resources insufficient to maintain a whole journalistic enterprise are nevertheless sufficient to make the problematic function viable?

So just very briefly, since this is already too long, let’s think about a couple of the harder-to-replace functions that traditional journalistic institutions and full-time professional reporters serve. One that certainly shouldn’t be undervalued is trust and reputation, both individual and institutional. Busy sources need to know that it’s in their interest to spend a half hour talking to a reporter, that they can feel comfortable speaking candidly on background, and so on. Readers need to know that information that’s impossible to cloud-verify (information from anonymous sources, scenes the reporter witnessed firsthand) can nevertheless be relied upon.

A few possible solutions. First, while traditional newspapers combine the production and credentialing functions (take his call, he’s from The New York Times) they can be disaggregated in principle. You could have a low-overhead institution with no function but credentialing, where the writers voted most reliable by readers get a sort of press pass from the Internet. Readers could signal their support for an author with a click on an embedded button in an article, much as they do with Digg. There’s an obvious chicken-egg problem here, which is one reason credentialing and production traditionally have been bundled, but if it were spun off by a consortium of credible institutions (papers, eminent J-schools) it could probably get off the ground pretty quickly. On the other side, you could have long-term players in a particular field (perhaps at one of those nonprofits or advocacy groups) acting as trusted brokers between sources and writers. That is, you could have an individual or institution primarily specializing in some non-journalistic function, but also acting as an intermediary who can certify that some anonymous bit of information does indeed come from a reliable source with direct knowledge of the situation. (You’d also need a mechanism to confirm that “independent” confirmations of a claim were genuinely independent without disclosing the source identity to any more people than necessary.)

That’s off the top of my head; I don’t mean to suggest that these particular mechanisms are necessarily feasible, let alone optimal. But I’m fairly confident that something like this is the right framework for thinking about how journalism adapts to a new media environment—which is to say, not thinking about “journalism” at all, but particular production functions in particular coverage areas. One seemingly overwhelming problem may turn out to be a lot of relatively manageable ones.