# Firing journalists is the worst thing a publication can do: what a hiring decision says about where a publisher is going, and who to hire instead, sgit.ai > Press Gazette's tracker of journalism job cuts counted at least 3,434 in the UK and US in 2025, 3,875 in 2024 and about 6,000 in 2023, and 2026 is on the same scale: the BBC planning up to 2,000, the Washington Post a third of its staff, Reach 220 editorial jobs cut and 60 created, Scripps 268 jobs while "leaning into AI", McClatchy more than 90 journalists across at least 13 titles, with the Idaho Statesman losing 10 of its 18 reporters and a new AI lab welcoming five hires the following week. This article is about what those decisions say, which is more than what they cost. A hiring plan is the clearest statement a publisher makes about where its centre of gravity is going. Fire the people who make original content and hire for traffic, and the centre moves to packaging; packaging is measured by engagement, and engagement optimised for is clickbait whether anyone meant it or not, because what you measure is what you become. The opposite cut is as bad: fire the engineers and nothing enables the journalism. The argument is that a publisher now needs both, plus the roles that turn a notebook into a graph, historians, cartographers, ontologists and librarians, each AI-enabled, and the tooling handed to journalists as "this makes you better", never "this replaces you". It sets out five counts anyone can make from outside to read a publisher's real strategy, and an unsourced parable about who hires the ones you let go. *Source: · site v0.7.74 · this file is generated from the same content as the page, so the two cannot drift. Every page on this site has a `.md` twin; internal links below point at them.* --- [Home](../index.md) / [Articles](index.md) / Firing journalists is the worst thing a publication can do: what a hiring decision says about where a publisher is going, and who to hire instead # Firing journalists is the worst thing a publication can do: what a hiring decision says about where a publisher is going, and who to hire instead By [Dinis Cruz](../about/index.md) · 2026-10-12 · [article v1.0.0](versions/firing-journalists.md) · [site v0.7.68](../admin/versions.md) · journalismpublishinghiringlayoffsstrategymetricsengagementprovenancenewsroomrolesarticle ***Abstract:** Press Gazette's tracker of journalism job cuts counted at least 3,434 in the UK and US in 2025, 3,875 in 2024 and about 6,000 in 2023, and 2026 is on the same scale: the BBC planning up to 2,000, the Washington Post a third of its staff, Reach 220 editorial jobs cut and 60 created, Scripps 268 jobs while "leaning into AI", McClatchy more than 90 journalists across at least 13 titles, with the Idaho Statesman losing 10 of its 18 reporters and a new AI lab welcoming five hires the following week. This article is about what those decisions say, which is more than what they cost. A hiring plan is the clearest statement a publisher makes about where its centre of gravity is going. Fire the people who make original content and hire for traffic, and the centre moves to packaging; packaging is measured by engagement, and engagement optimised for is clickbait whether anyone meant it or not, because what you measure is what you become. The opposite cut is as bad: fire the engineers and nothing enables the journalism. The argument is that a publisher now needs both, plus the roles that turn a notebook into a graph, historians, cartographers, ontologists and librarians, each AI-enabled, and the tooling handed to journalists as "this makes you better", never "this replaces you". It sets out five counts anyone can make from outside to read a publisher's real strategy, and an unsourced parable about who hires the ones you let go.* A cut says the centre of gravity is moving to packaging; a hire says it is staying on the content and getting tooled; and five counts let anyone read which from outside. Below, the two failure modes and the case this site makes. I am a technologist, and I am about to argue that firing journalists to hire technologists is the worst thing a publication can do. I want that on the record first, because it is the argument against my own kind, and I think it is right. The numbers are not in dispute. [Press Gazette's tracker](https://pressgazette.co.uk/news/journalism-job-cuts-in-2026-updates/) counted at least 3,434 journalism job cuts in the UK and US in 2025, at least 3,875 in 2024, and about 6,000 in 2023. Its 2026 page, as of October, has the BBC expecting to cut between 1,800 and 2,000 jobs, the Washington Post cutting a third of all staff including more than 300 of about 800 journalists, GB News proposing up to 90, Reach planning to cut about 220 editorial jobs and create 60, Scripps cutting 268 jobs as it "embraces AI for a 'revolution' in programming", with its chief executive quoted as "leaning into AI, automation, technology and the centralisation of some roles", and McClatchy cutting more than 90 staff across at least 13 titles. [Tech Policy Press](https://www.techpolicy.press/cutting-local-journalists-mcclatchy-embraces-ai-content/) went through the McClatchy cuts paper by paper: the Idaho Statesman lost 10 of 18 reporters and three of four editors, one paper lost its city hall and schools reporter, another its only Spanish-speaking reporter, and less than a week later the company's AI-focused Content Innovation Lab announced five new hires. What I want to write about is not the cost of those decisions. It is what they say. ## In short - **A hiring decision is a direction of travel.** It is the clearest statement a publisher makes about where its centre of gravity is going, clearer than any strategy document, because it is what the company actually spent. - **Fire journalists, and the centre moves to packaging.** Fewer pieces with original reporting, fewer sources followed, fewer corrections. The product becomes the repackaging of inputs everyone else has. - **What you measure is what you become.** Hire for traffic and engagement and the metric becomes pages opened and time on page. Optimise for that and you get clickbait by construction, whatever anyone intended. The metric is a hiring plan in disguise. - **The opposite cut is as bad.** A publisher with journalists and no engineers has content that nothing enables: no graph, no provenance anyone can check, no versions, no surfaces for the agents that now read, no meter. - **It is a spectrum, and context matters.** A pure repackaging business is honest about what it is, and that world is shrinking fast. A normal media organisation has to cover the whole thing. - **Who to hire instead.** Keep the journalists and tool them. Hire the roles that turn a notebook into a graph: historians, cartographers, ontologists, librarians, each AI-enabled. Hire the engineers and AI product managers who build what the tooling needs. - **You can read all of this from outside.** Five counts: original share, sourcing, corrections, roles hired against roles cut, and the metric the publisher itself reports. - **The parable.** We hire the ones you let go. In publishing, the ones being let go are the journalists. ## The centre of gravity Every organisation has a centre of gravity: the place where the thing it exists to make is made, and where, if you look honestly, the power is. In a software company it is the people who write the software, and firing developers to hire salespeople moves the centre to selling what you already have. In a design company it is the designers. In a publisher it is the people who make the content, and in a news publisher that is the journalists. So the question a hiring decision answers is: where is this company moving its centre? Fire journalists and hire for technology, and the honest answer depends on what technology. Hire engineers who build provenance, graphs, versions and the surfaces readers and agents use, and the centre stays on the content and gets stronger. Hire for traffic, engagement and content operations, which is the common case, and the centre moves to packaging. The company becomes one that takes inputs everyone else has, the wire, the press release, the trending topic, and packages them for distribution. That is a real business, and I will come back to it. It is not the business the journalists were in. ## What you measure is what you become The reason this is not a matter of intention is the metric. If success is measured by engagement, by pages opened and time on page, then the organisation reorganises itself around opening pages, because that is what gets rewarded, promoted and funded. Optimise for that metric and you get clickbait, whether or not anyone in the building wants it. I do not think most of the people doing this want it. The metric wants it. This site runs on a different metric, and the difference is the point. [Pay after you read](../articles/pay-after-you-read.md) charges a reader pence for how far they read and then asks how useful it was, and that sets the price. [The site is an inbox](../articles/the-site-is-an-inbox.md) is a brief for signals a reader produces on their own terms and can see, with a rule that nothing is placed to make them scroll. Those metrics measure value to the reader. A publisher that measured success that way would reorganise itself around being worth reading to the end and worth paying for afterwards, and that is a different hiring plan from the one the tracker records. The metric is a hiring plan in disguise. That is the sentence I would put to a board. ## The spectrum, honestly I want to be fair about this, because it is a spectrum and context matters. There are organisations whose product is packaging, and they know it: they aggregate, they summarise, they distribute, they sell attention to advertisers, and they do not claim to be doing original reporting. For them, hiring for engagement is coherent. The problem is the world they live in. [How news got here](../articles/how-news-got-here.md) has the numbers: Facebook referrals to publishers down 58% in six years; the advertising that was 82 cents of every dollar in 2005 now overtaken by circulation; American newspaper newsrooms down from about 71,000 jobs to about 31,000 between 2008 and 2020, a loss of 57%. And [Sixteen thousand fetches, ten clicks](../articles/token-bill-nobody-is-sending.md) has the newest one: answer engines fetching a small publisher's pages 16,000 times in a week and sending 10 readers back. The attention business is getting smaller and harder every year, and a publisher that moves its centre of gravity into it is moving into a shrinking room. At the other end is the organisation that wants only to do journalism and sell it on paper, or feed it to someone else. That can be coherent too. But a normal media organisation, one that wants readers of its own, has to cover the whole thing: the reporting, and the stack that makes the reporting worth more than the paper it is printed on. Which brings me to the opposite cut. ## The other way to fail If a publisher has no developers, no engineers, no AI product managers, nobody doing the kind of work I do here, it fails the other way round. The journalism is done, and then nothing enables it. There is no graph of the story, so a correction is a new document that nothing cites. There is no provenance a reader can check, so trust is a byline and a hope. There are no versions, so what changed is invisible. There are no surfaces for the agents that increasingly do the reading, so the work is scraped and summarised by someone else's machine with no way to be paid. There is no meter, so the only models left are the two bad ones. The reporting is done, and then it is thrown away as a PDF of the paper. Jonathan Palmer put this well in [the post I answered yesterday](../articles/the-article-remains-a-product.md): "This isn't about replacing journalists with engineers. Journalistic expertise and their networks create the underlying advantage." The engineers are there to capture more of the value the journalists create. Fire the journalists and there is nothing to capture. Fire the engineers and it is not captured. ## Who to hire instead So the hiring plan I would argue for, and the one this site is a working example of: - **Keep the journalists, and tool them.** The tooling handed over as "this makes you better", never "this is going to replace you". That is the whole business case of most of the articles on this site, and it is the difference between a journalist who uses an agent to follow a claim to its source in a minute and one who was replaced by an agent that cannot. - **Hire the roles that turn a notebook into a graph.** In [One article, five readers](../articles/one-article-five-readers.md) the roles are a Librarian that catalogues every fact, claim, number and source with its sentence, a Cartographer and Ontologist that builds the concepts and draws the maps, a Historian that says what a piece added and where it sits, an Explainer and a Storyteller. On this site those are agents with role files. In a publisher they should be people, AI-enabled, with the same role files, because the value they make, the graph, is the thing [the agent will pay for](../articles/the-agent-is-the-reader.md). - **Hire the engineers and the AI product managers.** The people who build the back end, the non-functional requirements, the provenance, the versions, the surfaces, the meter. Not instead of the journalists. For them. - **Measure value to the reader**, and let the metric do the reorganising. The parable I keep coming back to is one I cannot source, so take it as a story. A large company complains that it cannot innovate, cannot move, cannot find the people. It asks a fast-growing competitor how it does it, and the answer is: we hire the ones you let go. In most organisations the good ones leave first, because they want to do the work. In publishing right now the good ones are the journalists, and the tracker says they are being let go by the thousand. Someone will hire them. The question for a publisher is whether it wants to be the one that did, with the tooling, or the one that handed them over. ## Read it from outside None of this requires access to a publisher's strategy. The centre of gravity is visible from outside, in five counts, and the newsroom's agents already do most of them on other people's content, as they did for [twenty-one World News Day op-eds](https://newsroom.sgit.ai/world-news-day/). 1. **Original share.** Of the pieces published this month, how many carry original reporting: a byline that did the work, a primary source a reader can open, a quote that is not from a press release. 2. **Sourcing.** How many claims link to the thing they rest on, and how many of those links still resolve. 3. **Corrections.** How many were published, where they appear, and whether the pages that cited the corrected claim know. [Corrections must propagate](https://newsroom.sgit.ai/corrections/index.html) is the standard; few publishers meet it, and we do not yet either. 4. **Roles.** In the last twelve months of job postings and cuts, the ratio of content-making roles to traffic and packaging roles, and whether the engineers were cut with the journalists or kept. Reach's 220 cut and 60 created is a number of this kind; what the 60 are is the signal. 5. **The metric they report.** What the publisher itself calls success, in its own words, in its results and its job advertisements. Run those five on any publisher and you have its real strategy, whatever the strategy document says. We have enough data on this site to run them on ourselves, and the honest result would be: original share high, sourcing high, corrections kept but not yet propagating, roles all agents and one person, and a metric that measures value to the reader with almost no readers yet. That is where I would start the map, and I would publish it. ## What this does not claim - **The tracker is a tracker.** Press Gazette's figures are for the UK and US, are described as "at least", and the 2026 page has no running total; the entries above are its words as of October 2026. - **I do not know the inside of any of these decisions.** Some cuts are survival, some are a change of owner, some are the thing I am describing. The argument is about what the decision signals and what the metric does, not about any one publisher's motives. - **The parable is unsourced.** I have heard it attributed to more than one company and I cannot verify any of them. - **This site is not a publisher at scale.** Seventy-odd articles by one author, written from voice memos by agents, is a working example of the stack, not proof that a newsroom of two hundred can adopt it. The [page for publishers](../publishers/index.md) is where we say what is live and offer to try it. If you run a newsroom and are deciding who to cut, or you were cut and want to see what the tooling looks like when it is for you, write to [agent@riskmandate.ai](mailto:agent@riskmandate.ai). ## Where this comes from A voice memo of mine on 12 October 2026, written by this site's agent in the voice of this site. The argument and the editorial responsibility are mine. The job-cut figures and the quoted entries are from [Press Gazette's journalism job cuts tracker for 2026](https://pressgazette.co.uk/news/journalism-job-cuts-in-2026-updates/), read on 12 October 2026, which also gives the 2023, 2024 and 2025 totals; the McClatchy detail is from Chris Mills Rodrigo, [Cutting Local Journalists, McClatchy Embraces AI Content](https://www.techpolicy.press/cutting-local-journalists-mcclatchy-embraces-ai-content/), Tech Policy Press, 8 October 2026. The historical staffing and revenue numbers are from [How news got here](../articles/how-news-got-here.md) and its sources at Pew Research Center. Jonathan Palmer's sentence is from [his post on LinkedIn](https://www.linkedin.com/posts/jonathanjpalmer1_futureofnews-ai-publishing-share-7514686042193637377-Zs1Z/) of 10 October 2026. The figure was drawn for this article. Related: [The article remains a product](../articles/the-article-remains-a-product.md), [One article, five readers](../articles/one-article-five-readers.md), [The agent is the reader](../articles/the-agent-is-the-reader.md), [Pay after you read](../articles/pay-after-you-read.md), [The site is an inbox](../articles/the-site-is-an-inbox.md), [Sixteen thousand fetches, ten clicks](../articles/token-bill-nobody-is-sending.md) and [Liquid content needs water](../articles/liquid-content-needs-water.md). ## Threads News & evidenceStartups & strategy[This article as a graph →](graphs.md#firing-journalists) ### Builds on - [Pay after you read: how a reading meter became a working business model in one afternoon, one release at a time](pay-after-you-read.md) Seven releases in one afternoon turned the reading meter into a working model: pay after you read, and the rating sets the price. - [The site is an inbox: deciding before reading, marks that build the reader's own graph, and why I could not write this brief until the vault existed](the-site-is-an-inbox.md) The site as an inbox that starts empty: decide fast before reading, mark slowly while reading, build a graph you own, and sign what you send to the newsroom. - [The reader was always the product: a corrected history of how news got into this mess](how-news-got-here.md) News has sold the reader to advertisers since 1833; the web took the monopoly, the platforms made the reader measurable, and AI took the traffic. - [Sixteen thousand fetches, ten clicks, and a token bill nobody is sending: the case for paying publishers to be easy to read](token-bill-nobody-is-sending.md) AI answer engines pay to read the web as HTML; a publisher who serves markdown, dates, hashes and a typed graph saves them tokens and should get a share. - [The article remains a product: Jonathan Palmer's four outputs for a newsroom with no human readers, checked against one that runs](the-article-remains-a-product.md) Jonathan Palmer's four outputs for a newsroom with no human readers, quoted line by line and checked against newsroom.sgit.ai as it runs, screenshots included. - [One article, five readers: a librarian, a cartographer, a historian, an explainer and a storyteller read the same piece](one-article-five-readers.md) Write the article first, then send five agent readers through it: a catalogue, an ontology and maps, the arc, two minutes, and a deck. - [The agent is the reader: why agents will pay for graphs, personas and signed claims, because it is cheaper than not paying](the-agent-is-the-reader.md) Agents pay to read the raw web and guess its meaning. A curated graph answers for 90% fewer tokens, so paying the publisher is cheaper than not paying. - [Liquid content needs water: liquefy the journalist's notebook, not the finished product](liquid-content-needs-water.md) A reply to FT Strategies on liquid content: the water is the reporting, so liquefy the journalist's notebook, keep the writing theirs, and pay per use. [All articles](index.md) · [All graphs](graphs.md) **Posting this article on LinkedIn?** The cover is [firing-journalists.jpg](../articles/banners/firing-journalists.jpg) (1920×1080, title and key ideas on it). Upload it as the article cover, paste the title, then select and copy the body from this page. **Want the next issue by email.** One issue a week or so: what was published, what it adds up to, and what is worth your time. [Subscribe to the SGit Newsroom →](../subscribe/index.md) [← All articles](index.md) --- *[Site index for agents](../llms.txt) · [HTML version](https://sgit.ai/articles/firing-journalists.html)*