Prose Is Dead. Local News Isn't.

Thesis

The only durable model for independent local journalism in the AI era is fast, systems-based, data-driven investigative reporting that holds power to account on subjects neither viral video nor sponsored influencers will pursue.

Small independent newsrooms can survive this transition, and can produce work that rivals or exceeds the national outlets on depth, but only by abandoning prose-as-product and adopting open tools they own and operate themselves. This document presents the argument and the infrastructure.

The two existential threats to local digital news

The existential crisis in the local digital news space has two parts.

2.1   The signal problem

AI slop is going to continue creating more and more noise and less signal in the digital journalism space. To survive, publications need to overcome this noise.

2.2   The ripoff problem

AI and LLMs make it easier than ever to rip off journalism, and to rip off anything published online, and to make the output technically legal or distinct enough to avoid copyright issues.

Argument

The only way I believe we as local news publishers can overcome these two things, respectively:

3.1   Response to the signal problem: depth

We need to be far louder, clearer, and deeper in our reporting to beat the signal in the space.

This means having the deepest, most well-researched, most contextual, value-add reporting. More than any AI or LLM slop factory could or would produce. Becoming not only the most accurate but also the most interesting and value-add to the reader, by incorporating historical data and rich texture into the reporting.

This system allows small newsrooms to quickly have access to historical data and contextually relevant quotes and history on a level that is deeper than NYT and anything that exists right now. This is the first-of-its-kind system with full data lineage and provenance, so facts are never questioned nor hallucinated, and any discrepancies in output can be traced to their origin and source and audited.

3.2   Response to the ripoff problem: speed and comprehensiveness

The problem here is structural. The investment in journalism has been severed from the economic benefit of that journalism. Someone has to pay reporters to dig, verify, write, and publish. The economic value of that investment used to flow back to the publication that paid for it. It increasingly does not. The value is captured downstream by aggregators, content farms, and influencers who did none of the original investment. The publisher keeps paying. The rewards go elsewhere.

This is a structural pattern, not a journalism-specific problem. The English enclosure movement spent four centuries converting common lands worked by generations of peasants into private estates. The peasants did not stop working the lands. The rules of who collected the rent changed beneath them. Journalism is being enclosed the same way. The publication keeps investing in the reporting. The economic rent is increasingly collected by parties who did none of the investment.

AI and LLM factories will now be able to rip off good reporting almost immediately.

Even though it has not happened at scale yet, it will once the economic incentives of the lower-hanging fruit currently being exploited in other sectors dry up. News is unprofitable and not sexy, so luckily we have been largely spared of total annihilation. However, in the next five years, unless we do something to put up walls or moats around local news publishing ecosystems (economic moats), all content, original journalism, etc., will be immediately ripped off and monetized, which will continue the cycle of skipping the original publishers from the economic model and dropping the rewards and money to the influencers and marketers who are experts in marketing and packaging the information for modern audiences and who are unaffiliated with the publications.

This already occurs on social media and is not inherently bad. But ultimately we want the local news publication to have the marketing and influencer depth to participate in that economy and reap rewards from their own investment in public good (reporting and original journalism). That is a future node. I will write about it separately.

The other logical way to "beat" the content farmers, and to neutralize the effect of the LLM and AI farms ripping off original journalism immediately after publishing (bearing none of the costs and reaping all of the rewards), is to publish faster, more comprehensively, and at higher quality. Local publications are often days or weeks late to publish breaking news stories due to their small teams and the general lack of tech adoption in small indy news outlets. This system will immediately change this by allowing for the publication speed of a fully funded outlet on general news and newswire stories, so journalists can have time to work on deep reporting and newsrooms gain consistency and become the paper of record for their region.

Operating premises

Two additional premises this project is built on.

4.1   Prose is no longer economically valuable to local journalism

A writer who can write beautiful descriptive prose is no longer valuable in the newsroom. I love reading and I love prose, but it has no value in the marketplace because everybody has phones and HD video in their pockets. Beautiful prose about an incredible bottom-of-the-ninth grand-slam game-winning hit in the playoffs is never going to beat out a video of the real thing.

I would argue that flowery, descriptive, clever writing is dead. I doubt many people think of it as technology, but it is effectively an evolution in communication technology, and it no longer contains any economic value for a local newsroom. The technology has moved through:

A short genealogy of writing as economic communication technology word of mouth → stone tablets → hand-copied scrolls (only the most important information was deemed worth writing down, and only a small group of elite intellectuals could read it) → the printing press → the Protestant Reformation → the beginnings of mass literacy → printed materials becoming economically valuable as literacy rates rise → a common consumer base for written products creates economic value in the ability to write and convey stories accurately in text → explosion of global financial markets → colonialism → bond markets, battle reporting, the Bank of England → Wall Street, sports crosstabs, Tolstoy, Dostoevsky → Joseph Goebbels, Nazi propaganda → the American Dream, Hunter S. Thompson, feminism, the SCUM Manifesto → the Vietnam War on the nightly news → Watergate in 1972, CNN and 24-hour news in 1980 → the iPhone, 4G, Adam Curtis, @crackconnoisseur (aka Jinx) → the modern influencer.

Local news publications can only survive by original reporting and by leveraging the technology and the large open datasets available. Local journalism is no longer valuable for describing massive floods in Brooklyn, or political movements, or sports. The writing that was once so important, and that is still held in high regard by many of my contemporaries, was only ever trying to recreate events and movements accurately in text. Video does that now. No amount of prose or compelling writing will ever do it better than the video. Prose writers and skilled writers are still great artists, but they are no longer critical cogs in the information economy. They are completely useless in the information economy. Local news publications must reject prose and must answer the question of why we exist.

If local newsrooms have a purpose in the twenty-first century, it is to tell the stories that are not yet being told by video, and the stories that are too complicated or too abstract to be reduced to a viral video of a rat eating a pizza slice. More sophisticated reporting requires more sophisticated infrastructure. This pipeline provides it.

4.2   The "trusted voice" argument is over. Survival is accountability journalism.

Over the past decade, local publications have argued that they are "more trusted voices in the community." However, the local influencer is now more trusted in many cases. To even begin making the "trusted voice" argument requires the frequency, depth, and consistency of publishing described above, and I do not know of any independent local publication that is currently able to publish quickly and accurately enough to claim a value like "local trust."

The modern publication can only survive by publishing original reporting that holds truth to power and power to account.

There is a massive gap in the information economy. Influencers motivated by sponsorships and money are not incentivized to go after power institutions or to uncover corruption. To survive, local news publications must leverage data reporting to uncover the corruption that runs rampant. And it must be done ethically, competently, and quickly.

Local reporting needs to become more systems-based, more data-based, less politically active, and more grounded in base facts. It must be fast and it must be accurate. With the confluence of open data and accessible LLMs and AI, local publications should and will have a renaissance of Pulitzer-worthy work.

Government agencies are overworked, a condition kicked off by COVID and worsened by the death of the local news publication. The result is unprecedented corruption in our local communities. Much of the relevant data is public. Investigative reporting can and will bring it to light.

At present, the best investigative journalism I can point to is being done on the right of the aisle. Nick Shirley and James O'Keefe are the two journalists whose recent work I can think of where the reporting itself has produced massive change and follow-on investigations. Their political affiliations should not enter the conversation about the work itself, and yet liberals have largely written it off on partisan grounds. The investigative renaissance is coming, and it is upon us. Independent local news publications have both the incentive and the economic need to adapt and to leverage the tools of the future, or they will be washed away.

Project posture and conditions of use

Local news should be free and accessible by all. This tool exists to give local newsrooms what they need to make that real, without Silicon Valley venture money, without acquisition risk muddying the waters, and without monthly fees. It is local. It is a tool. Think of it as the best calculator you have ever had.

My mission with this project is to save local journalism. Righteous local journalism.

The only condition of use I ask is that the system be credited when it is used to generate or substantially support a story, and that it be mentioned in the acknowledgments if any work produced with it wins a journalism award.

Implementation status

The codebase has been in active development for ten months. The chart below tracks cumulative lines of code by subsystem, sampled at the end of each calendar day on which a commit was pushed. Line counts are exact, computed from the git object database via git ls-tree and git cat-file.

Figure 1. Cumulative lines of code by subsystem (frontend, API, agents, scrapers, RAG core, migrations, infrastructure, docs, other). Toggle subsystems via the legend at left. Switch between overlay and stacked views with the buttons below the legend.

Significant feature commits

Click a commit to mark its position on Figure 1.