Summary

  • Form DOE-417 prints four filing intervals in a single sentence — one hour, six hours, the end of the next calendar day after a determination of an attempted cyber compromise, and the later of 24 hours after recognition or the end of the next business day — plus 72 hours for the Final Report, and the alert criterion the filer selects for its own event decides which clock runs.
  • Two of the four anchors are the filing party's own acts, and the one-hour and six-hour limbs name no anchor at all. The published annual table shows the criterion claimed and both endpoints of the event, with no column recording when the report arrived.

Here is the sentence, as the instructions print it: Form DOE-417 "must be filed either within one hour; six hours; by the end of the next calendar day after a determination of an attempted cyber compromise; or by the later of 24 hours after the recognition of the incident OR by the end of the next business day of the incident." One sentence, four clocks, and no operator in it yet.

That sentence is not a summary of a rule. It is the rule's timing in full, and it sits in the form instructions beside a table in which four alert statuses carry those four intervals. Emergency Alert, criteria 1 through 9, within one hour. Normal Report, criteria 10 through 13, within six hours. Attempted Cyber Compromise, criterion 14, by the end of the next calendar day after a determination of the attempt. System Report, criteria 15 through 27, by the later of 24 hours after recognition or the end of the next business day, with 4:00 p.m. local time treated as the end of a business day.

Now count the anchors. Two of the four intervals start from something the filing utility does: a determination that an attempt occurred, a recognition that an incident occurred. The other two — one hour, six hours — name no starting point at all in that sentence. They say how long, not from when. A reader who wants to know what the hour runs from has to infer it, and the inference is the whole question, because a clock that starts on the operator's own act is a clock the operator also controls the start of.

The slowest limb is worth reading twice, because it is not one deadline. The later of 24 hours after recognition, or the end of the next business day, with 4:00 p.m. local taken as the close of business. Publish those two points as two points. Twenty-four hours and the end of the next business day are not the same moment, neither converts into the other, and the operator's own recognition sets both running.

The classification that selects the clock is not cosmetic. The emergency criteria include a physical attack causing major interruptions, reportable cyber incidents, complete operational failure, islanding, an uncontrolled loss of 300 MW or more lasting more than 15 minutes, firm load shedding of 100 MW or more, voltage reductions of 3 percent or more, and public appeals. The normal criteria include a potential-impact physical attack or vandalism, potential-impact cyber events, loss of service to more than 50,000 customers for an hour or more, and fuel-supply emergencies.

Choosing between an emergency and a normal report is choosing between one hour and six.

The form itself corroborates the structure in its own deadline row — "1 Hour", "6 Hours", "1 Business", "As required", "72 Hours" — and adds two details the instructions leave implicit. It carries OMB No. 1901-0288, an approval expiry of 05/31/2027 and a reported burden of 1.8 hours per response. And its Schedule 1 records the type of emergency, the cause, the impact and the action taken: four fields describing the event, none describing the hour it was sent.

The two documents also disagree with themselves at the boundary of the slowest interval. The form's text routes a System Report where criteria 15 through 26 are met; the instructions list criteria through 27. That is an internal numbering divergence in the instrument, at the edge of the limb whose anchor is the filer's own recognition. I am stating it as the two documents print it and not resolving it.

The programme page sets out what the collection is for: DOE national security, energy emergency management and analytical purposes, with the Energy Information Administration also drawing on the data in monthly reports. It also points to Annual Summaries — data tables of reported incidents from previous years.

That residue is where the instrument becomes readable from outside, and where it stops. The annual summary copy I could read, a third-party mirror of the 2019 table, carries eleven columns: Month; Date Event Began; Time Event Began; Date of Restoration; Time of Restoration; Area Affected; NERC Region; Alert Criteria; Event Type; Demand Loss (MW); Number of Customers Affected. Both endpoints of the outage. The affected area. The region. The criterion the filer claimed. The event type. The megawatts lost and the customers hit. No column for the date or time the report reached the agency. The Energy Department's own current file was not exposed to me on 19 November 2026 — the landing page's Annual Summaries section offered no file link and no header line — so I bound the claim to the copy actually read and say nothing about any other year's columns.

So the published record lets a stranger check the classification and never the interval. You can see which clock the filer said it was running on. You cannot see whether it ran. That is not an accusation that anyone misclassified an event or filed late; no document I read records whether any particular filing met its deadline, and I assert nothing about any operator's conduct. It is the shape of the instrument: the counted party authors the selection that starts the clock, and the residue cannot test the clock.

One economy worth stating plainly, because the temptation is to fold the sector's two reporting channels together. The private channel is dateable here and was not quotable. The CIP-008-6 page carries "Mandatory Subject to Enforcement 01/01/2021", Effective Date 01/01/2021, Inactive Date of Standard 06/30/2028, Board Adopted 02/07/2019, Filing Date 03/07/2019, FERC Order Issued 06/20/2019 and Docket No. RD19-3-000. The CIP index lists that standard as mandatory and lists CIP-008-7.1, under the same title, as subject to future enforcement; the CIP-008-7.1 page shows "Subject to Future Enforcement 07/01/2028", Effective Date 07/01/2028, Docket No. RM24-8-000, and a public note reading "Replaces CIP-008-7 (Errata)" — a note about CIP-008-7, not about CIP-008-6. Its notification requirement was not readable in a form I could verify on 19 November 2026, so no interval, addressee or attribute from that instrument appears in this piece, and I do not compare its clocks with the federal form's. The succession record is a schedule, and a schedule is what I am reporting.

Prior coverage is easy to state honestly. Two searches on 19 November 2026 for this instrument class returned no located article, while the directory records three existing articles on the operator this briefing hangs from — Electric Reliability Council of Texas, Inc. — so the delta claimed here is the instrument and its published residue, not that operator's history. Three adjacent briefings of mine each read a filing that routes to a public regulator: 47 CFR Part 4 at the Federal Communications Commission, the interagency computer-security notification rule at the Federal Reserve, and CIRCIA at CISA. This one routes to an energy agency whose published output is an event table, and the table stops one field short of the clock.

Publish the arrival. One column on the annual summary — the date and time the report reached the agency, or the interval that elapsed against the criterion claimed — turns a classification into a record an outsider can test. That is a field to add and a return to publish. The form already collects the hour the event began and the hour it was restored. It does not collect, or does not publish, the hour it arrived, and until it does, the only person who can say whether the clock ran is the one who set it.