Every read now ends with specific events that would break it, each tied to a real level from the evidence: a close below the 50-day average, the next filing restating a figure, a drawdown past a stated point. The desk can only cite levels it actually read, so it cannot invent one. Those events are then watched for you. Each weekday morning we check the ones that can be checked, and if a risk the desk named actually happens, you hear about it. A stated risk materialising is a reason to read again, never a signal to act. The risk review is computed rather than written. You now see how much of a read rests on a single source, how it ranks against its industry peers, and how common each of its signals is across the whole universe, all dated and all labelled "computed, not argued". A report now opens with what changed since your last read: the new filings, the added risk factors, the signals that rolled, and where the score movement came from. Click any date on the timeline to open that earlier read as it stood. Every finished report carries a timestamped record of the run that produced it. Which analyst reported when, what the chair ruled, how long it took. It is the actual record, not an animation. When an analyst is quiet, the report now says why, and it tells apart two things that used to look identical: there was nothing to find, versus we could not check. A data source being down is our gap, and it is no longer written as a fact about the company. The desk now reads every ki…