Most news sources tell you what happened. dissemin8ed tells you what’s happening, who’s covering it, from which direction, and whether it’s still moving — without needing to know who you are.
dissemin8ed is an intelligence layer on top of the open web of journalism. It does not produce news. It surfaces it, clusters it, contextualises it, and gets out of the way.
Every thirty minutes, the pipeline crawls hundreds of RSS feeds from outlets across the world — national papers, wire services, regional publications, specialist sites — and compresses what it finds into a four-layer editorial hierarchy:
By the time a story reaches your screen it has already been cross-referenced against dozens of other sources, labelled with bias and paywall indicators, scored for signal strength, and woven into the thread of related events that came before it.
The modern reader is not suffering from a lack of information. The problem is the structure imposed on it.
dissemin8ed addresses all three by treating news ingestion as a structured compression problem: raw articles are the input; a multi-resolution editorial hierarchy is the output.
These are structural distinctions, not feature comparisons.
Every card you see is a story cluster — the same event reported by multiple independent outlets, collapsed into one object. The question the UI answers is not “which article should I read?” but “how widely is this covered, and by whom?” Wire syndications are tracked separately from independent corroboration; a story covered by twelve outlets that all republished the same AP dispatch is scored differently from twelve independent investigations.
Story score = recency × independent source count × source quality. There is no engagement feedback loop. A story that nobody clicked yesterday but is now being covered by eight independent outlets will rise. A story that went viral but has only one original source will not. Social amplification is not a metric this platform chases.
Political and editorial slant indicators come from AllSides and MBFC, third-party ratings applied uniformly across all sources. The Cross filter — stories confirmed by both left-leaning and right-leaning outlets — is a factual observation about coverage patterns, not a claim about truth. A story covered only on one side of the spectrum is a different kind of story than one covered across the board. That difference is information.
The trending layer detects n-gram spikes across the full corpus of current stories. A phrase jumping from background frequency to 40× its baseline in four hours is qualitatively different from a phrase that’s been common for a week. Phase labels — Breaking, Developing, Fading, Dormant, Resolved — tell you where a story is in its trajectory, not just how recently it was published. Large ongoing arcs that go quiet for a day are marked Dormant rather than Resolved, so a story temporarily displaced by breaking news is not mistaken for one that has concluded.
Threads maintain the connective tissue of a developing situation across days. Arcs track how big something became over weeks or months. The Yesterday narrative stitches the prior day’s activity into a coherent editorial summary before it scrolls out of view. A reader arriving at any point in a story’s life can find the thread. Context does not evaporate after 24 hours.
Accounts exist solely for subscription management — bookmarks, source blocks, and preferences. We do not build a behavioral profile of you. There is no interest graph, no engagement model, no per-user ranking signal derived from what you read. Dwell time and reaction signals exist only to improve what the platform shows everyone; they are never sold, never used to target you, and never used to manipulate session length. If we cannot improve the product with a signal, we do not collect it.
These are operational commitments, not aspirations. Every product decision is filtered through them.
Engagement signals exist to improve the product, not to be sold or to build a behavioral profile. If we cannot improve the platform with a signal, we do not collect it.
Headlines, summaries, and narratives we generate are aids to navigation. There is always a link. There is always attribution. We are never a substitute for the original reporting.
A story reported independently by twelve outlets is more significant than one story shared twelve thousand times. sourceCount is the primary ranking signal. Social amplification is not a metric we chase.
We display editorial bias indicators not to shame sources but to inform readers. Coverage pattern across the political spectrum is itself information — we will always surface it.
If a story appears at the top of the page, a human must be able to articulate why. No black boxes. No “the model decided.” If we cannot explain a ranking, the ranking is wrong.
News is not a series of disconnected events. Stories have causes and consequences. A reader who arrives today deserves to understand how we got here. We will always provide the thread.
Choosing not to show a story is a decision. Burying an inconvenient story through scoring is editorial bias by algorithm. Scoring parameters are published. Bias and paywall filters are user-controlled, not baked into the default ranking.
A low-quality source that occasionally produces excellent journalism is preferable to a high-traffic source producing noise at scale. Quality scores attenuate — they do not exclude. Exclusion requires a deliberate human decision, recorded and explainable.
Core content — headlines, source names, links, dates — must be accessible in the HTML response without client-side rendering. Enhancements are progressive. A reader with JS disabled gets a useful page, not a blank one.
dissemin8ed is not trying to convert non-readers into readers. We are building for people who follow the news seriously and are frustrated by the current tools. Every feature is filtered through this lens.
Artificial intelligence is a tool on this platform, not a voice. We use it precisely and only where deterministic methods are insufficient.
In every case, AI output is labelled as generated, based on source material we have already independently fetched and attributed, and disposable — if the model is wrong, the underlying sources remain correct.
We are not racing to be acquired. We are building a tool that is useful for as long as journalism exists on the open web.
The principles described on this page derive from a formal governing document — the dissemin8ed Constitution — ratified February 24, 2026. It defines what the platform is, what it refuses to be, and the rules that govern every product decision.
dissemin8ed is built on the premise that an informed public requires more than access to information — it requires structure, context, and provenance. We provide those things and then get out of the way. — Article VI, dissemin8ed Constitution v1.0
Amendments to the Constitution require a written rationale, a git commit recording the change with date and author, and consistency with Articles I and II — the core identity and the ten rules may be refined but may not be reversed.
Constitution v1.0 — ratified 24 February 2026.