How We Build Entertainment Intelligence
Entertainment information is scattered everywhere — trade press, local news, interviews, podcasts, YouTube, Reddit, social posts, permits, production documents. Digging through all of it to answer one question ("is this actor actually attached?") can eat an entire evening. We built this to do that hunting for you, connect what it finds, and explain it without pretending rumors are facts.
This page is about the second half of that: how we decide what to show you, and why you shouldn't have to just trust us on it.
We Don't Want You to Take Our Word for It
A site that just tells you "Actor X is joining the show" is asking for blind trust. We'd rather show our work: here's the claim, here's what we found, here's where it came from, and here's what it currently supports. If you disagree with our read of the evidence, you should be able to look at the same evidence we looked at and reach your own conclusion. That's the whole point of this page.
Where the Information Comes From
We track information from a range of source types — official studio or network announcements, trade and entertainment publications, interviews, podcasts, YouTube channels, local reporting, and social posts, among others. Each type carries a different kind of weight. An official studio announcement is different from a trade publication's report, which is different from a fan noticing something on social media. None of that is automatically true just because it came from somewhere — a source existing doesn't make its claim correct. What matters is what kind of source it is, and what it actually said.
A Signal Isn't a Fact
Some of what we pick up is just a signal: a raw, potentially useful observation that hasn't been investigated yet. A local social post appears to show unusual activity outside a known filming location. That's worth a second look. It is not proof of anything on its own — it's a lead, not a conclusion. We keep signals visually and explicitly separate from anything we treat as evidence, precisely so a possibility never quietly gets promoted into a fact just because it showed up.
Claims Need Evidence
A claim is an actual assertion about the entertainment world — an actor is reportedly attached, a production is filming somewhere, a show got renewed, a rumor was denied. Evidence is the material that supports or contradicts that claim: a specific article, an official statement, an interview quote. The principle we're building toward is that you should be able to inspect the evidence behind a claim yourself, instead of just receiving our conclusion about it.
Corroboration Is More Than Counting Articles
It's tempting to assume that five articles about the same rumor means five independent confirmations. It doesn't. Outlets frequently repeat each other's reporting — one trade publication breaks a story, and a dozen others repost or rephrase it within the hour. That's one source, echoed, not five sources agreeing.
When we talk about corroboration, we mean genuinely independent supporting information, not a raw mention count. To be direct about a real limitation: we don't currently have a reliable way to detect when several outlets are all repeating the same underlying report rather than reporting independently. Our corroboration counts approximate independence by counting distinct sources, which is not the same as confirming those sources did independent reporting. We're not going to pretend that's solved when it isn't.
What "Reported," "Corroborated," and "Confirmed" Mean
Where we've done the work to track a claim's history, it can move through a small set of states: Reported (at least one source has said this), Corroborated (two or more independently sourced reports support it), Confirmed (an official or studio source is behind it), Unconfirmed or Speculation (circulating, but without evidence that rises to a reported claim), and Debunked (it was reported, then contradicted).
These aren't just labels we apply by feel — each one requires a specific kind of evidence to back it up, and we don't let a status get asserted without it. A claim's state can change as new evidence comes in, and we keep the history of how it changed, not just the current answer.
What Changed?
On a show or movie page, the "What changed?" section is a timeline of the meaningful updates we've recorded. Where the data supports it, you can expand an individual entry and see why it was recorded, the primary evidence behind it (source, article, publish date, an excerpt, and whether that evidence supports or contradicts the claim), any supporting discovery signals we picked up separately, and the claim's recorded status history.
We deliberately keep "supporting signals" labeled as discovery signals, not verified evidence, and shown apart from the primary evidence list. That distinction is the point of this whole page in miniature: we don't want you to see "Reported · 2 sources" and just trust the number. We want you to be able to ask "okay, show me" — and actually see what's behind it.
Show Me the Evidence
The reason all of the above matters is provenance: information should stay attached to where it came from. Ideally, for anything meaningful we show you, you can trace what was found, where it came from, which claim it relates to, whether it supports or contradicts that claim, what state the claim is currently in, and how that state got there. "Actor X is joining the show" is much easier to trust when it comes with "here's the claim, here are the sources, here's what they actually reported, here's the current status" attached to it.
Why Reddit, X, YouTube, and Other Messy Sources Can Matter
Unconventional sources are genuinely useful for discovery. A Reddit user notices something happening near a filming location; a fan's social post catches an actor somewhere unexpected. Posts like that can become the first signal that something is worth investigating, well before any outlet reports on it.
But discovery and verification are two different steps, and we treat them that way. A signal like that can prompt us to look for an actual report, an official announcement, a permit, an interview, or other corroborating (or contradicting) information — it doesn't become a claim just by existing. Upvotes, views, reposts, and engagement are not evidence of anything being true; they're evidence that something got attention.
To be concrete about what's actually running today rather than what we'd eventually like to run: our live automated discovery currently covers a small set of curated RSS feeds and YouTube channels — not the full range of Reddit, X, and other social platforms described above. Reddit- and social-post-driven discovery is a direction we're building toward, not something already scanning those platforms at scale. Automated discovery is also deliberately conservative in what it's allowed to do: it can surface a signal or attach a "mentions" note to existing evidence, but it never changes a claim's status on its own — that stays a human decision, backed by the evidence rules above.
Where AI Fits
AI helps us investigate information. It is not the source of truth. We use it for things like extracting entities from text, drafting summaries, and — when you ask a research question — synthesizing an answer strictly from the evidence we've already collected, with every factual sentence required to cite the specific evidence it came from and to match that evidence's claim status precisely (a "reported" claim gets described as something one source reported, not as settled fact).
What AI does not do here is independently decide whether a rumor is true, whether a source is credible, whether two outlets reported something independently, or whether a claim's status should change. Those are governed by the explicit evidence rules described above, not by a model's judgment call. "Trust our AI" isn't the product. "Here's what we found, where it came from, how it relates to the claim, and what supports it" is the product.
Information Changes
Entertainment information is unusually fluid. Projects move through development, casting, production, release, renewal, or cancellation, and reports evolve right along with them. A report that was reasonable on Monday can be contradicted by an announcement on Friday — that doesn't necessarily mean the original source was wrong to report it; the available information was just incomplete, preliminary, or later overtaken by events. This is exactly why we keep a claim's status history instead of only showing its current state. We're not claiming we can always perfectly reconstruct everything that happened along the way — only that we keep what we actually recorded.
We Will Get Things Wrong
Trust goes up, not down, when a system is honest about where it can fail. Some of the ways this one can: a source can simply be wrong; a report can be incomplete; information can go stale as a situation moves on; multiple outlets can repeat one underlying report in a way that looks like independent corroboration but isn't (see above); discovery signals can be noisy; automated classification can misfire; matching a mention to the right show, movie, or person isn't always perfect; and not every meaningful development gets caught immediately.
We're not claiming we've eliminated uncertainty from entertainment reporting — nobody has. We're building a system that makes that uncertainty visible and investigable instead of hiding it behind a clean-looking headline.
The Goal
Fans shouldn't have to spend two hours cross-referencing Google, Reddit, YouTube, X, interviews, trade publications, local reporting, and obscure corners of the internet just to answer one question about a show or movie they care about. We want to do more of that hunting for them. And when we tell them why something matters, they should be able to see exactly where it came from.