I Joined Startup School to Build the Trusted Map of Events

I joined Startup School this week.

The idea I am working on: a trusted map of all events, past, current, and future. Not another aggregator. Not a scraper wrapped in a calendar UI. Something structurally different.

The problem with every events product I have seen is that it treats "is this event real?" as a content moderation problem. Someone submits an event. A human reviews it or a filter catches obvious spam. The event appears or does not.

That model does not scale. It also hides the real question: who thinks someone is an authority to say "this event is real, this one is not," and why?

That is the product. Not the events. The graph of trust around events.

Every event claim comes from a source. That source has a history: how often it has been right, how often wrong, what types of events it covers, what its incentive structure is. A venue announcing its own show has a different reliability profile than a local press outlet covering regional culture, which has a different profile than a government tourism board publishing its annual calendar.

None of those sources are binary. They are not trustworthy or untrustworthy. They are trustworthy to a calculable degree, in a specific domain, over a specific time horizon.

The vision:

• Ingest event claims from every source we can reach.
• Model each source as a node with a trust score, domain coverage, and track record.
• Propagate uncertainty through the graph: corroboration raises confidence, silence or contradiction from high-trust sources keeps it flagged.
• Surface not just the event but the confidence level and the reason for it.

At scale, this becomes something no events product has been: an honest index. Not "here are the events" but "here are the events we are confident about, here is why, and here is what we are uncertain about."

The hard problem is bootstrapping the trust graph. You need historical data to score sources, and you need source scores to validate historical data. The exit from that loop is starting with sources that are already publicly auditable: official government calendars, verified venue feeds, press archives with timestamps. You score those first, then use them as anchors to evaluate everything else.

I do not know yet if this is a business. I know it is the right structure for the problem.