Benchmarking
How Roger Wilco Imagery's output measures up — against Prime Rogue's published analysis, against other coverage, and against itself. Each entry grades the work, names what it got wrong, and records what changed.
Data-source benchmark · subject: NASA FIRMS area API on the new map key, agent Gazda (docs/AGENTS.md) · benchmark: RWI FIRMS watch (global 24 h files, every ~3 h, data/firms-watch/hits.csv) · grades: RWI A / benchmark B
First run on the FIRMS map key. Over the same four boxes and seven days, the key returned every detection the watch had (48 of 48) plus 25 more. 24 of the 25 predate the western Libya box joining the watch; the other is one southern Lebanon detection the watch missed. In the period the watch was running it caught 48 of 49 (98%). The key's gain is reach back in time and any box on demand, not a better daily feed.
Coverage benchmark · subject: Postmedia's coverage of the CFB Wainwright fire (Edmonton Journal, Calgary Herald, National Post, Sun titles), 17–26 Oct 2026 · benchmark: Global News (Ashleigh Stewart and desk), CBC, CTV, The Globe and Mail, Canadian Press, Prime Rogue Inc. · grades: benchmark C+
In every Postmedia headline and print teaser we could find, there is no wrong death toll, no explosion and no gas line. Its own Edmonton reporter covered the 22 Oct briefing for the Journal and both Sun papers, and the Journal and the Calgary Herald found injured soldiers' families. Grade C+, provisional and low-confidence: we could read no Postmedia article in full, and we found no documents, data, timeline work or accountability reporting, and no column on the fire.
Coverage benchmark · subject: The Globe and Mail's coverage of the CFB Wainwright fire, 17–26 Oct 2026 · benchmark: Global News (Ashleigh Stewart and desk), CBC, CTV, Canadian Press, Prime Rogue Inc. · grades: benchmark B
The Globe's Wainwright coverage was careful and well sourced: no wrong death toll, no gas or explosion claim, an internal "fire index low" memo on 18 Oct, and the first attributed report we found that DND put ignition at about noon. Grade B, provisional: it showed no data or imagery, relayed "around 4 p.m." and "30 to 60 seconds" untested, and most of its text could not be read.
Coverage benchmark · subject: Global News coverage of the CFB Wainwright fire beyond Ashleigh Stewart, 17–26 Oct 2026 · benchmark: CBC News, CTV, Canadian Press, Prime Rogue Inc., Ashleigh Stewart (Global News) · grades: benchmark D+
Outside Ashleigh Stewart's work, Global News was fast and reached into four provinces, but its first-night story put three wrong claims into the world's coverage: seven dead, an explosion, and a gas line or well. Grade D+: none was corrected with a notice, the seven-death headline is still live nine days later, and the desk did not use Stewart's noon finding.
Coverage benchmark · subject: CTV News coverage of the CFB Wainwright fire, 17–26 Oct 2026 · benchmark: Global News (Ashleigh Stewart and desk), CBC, The Globe and Mail, Canadian Press, Prime Rogue Inc. · grades: benchmark C+
CTV was fast, never reported seven dead or a gas line, and published the only visible correction notice we found at a major outlet. Grade C+, unchanged from the preliminary grade: it put "explosion" in its own headlines on 17–18 Oct, wrongly attributed an explosion to the Prime Minister, never corrected either, kept "4 p.m." as the start, and produced no documents and no imagery analysis.
Coverage benchmark · subject: The Canadian Press's coverage of the CFB Wainwright fire, 17–26 Oct 2026 · benchmark: Global News (Ashleigh Stewart and desk), CBC, CTV, The Globe and Mail, Prime Rogue Inc. · grades: benchmark B
The Canadian Press never reported seven dead, an explosion or a gas line in its own voice, attributed every early claim, and wrote the clearest "what we don't know" of any outlet. Grade B, unchanged from the preliminary grade: it found no documents, used no data, never reported the noon fire, narrowed the spread claim to "about 30 seconds" once, and left a geography error in its own fact check uncorrected. Where member outlets ran CP copy straight, they avoided the errors; where they blended it with their own sources, the errors came back.
Coverage benchmark · subject: CBC News coverage of the CFB Wainwright fire, 17–26 Oct 2026 · benchmark: Global News (Ashleigh Stewart and desk), CTV, Canadian Press, Prime Rogue Inc. · grades: benchmark C+
CBC had the widest and most humane coverage of the fire: fast live blogs, the families, the burn units, and the earliest burned-area estimate from imagery. Grade C+: it put a wrong death toll of seven into the world's wires, and on air the next morning it presented raw satellite fire detections as live and floated a gas well as a possible cause. It found no documents and never tested the official "around 4 p.m." timeline.
Self-evaluation · subject: RWI SITREP 005 (CFB Wainwright special incident report) · benchmark: Prime Rogue Inc., CFB Wainwright series (21–24 Oct 2026) · grades: RWI C / benchmark A−
SITREP 005's satellite work reproduced Prime Rogue's numbers independently, but its analysis trailed the published modelling, put troop contact 30 minutes late, paired the origin with the wrong range and skipped the gas-line question. Corrected in SITREP 005 revision 1.
Coverage benchmark · subject: Global News reporting by Ashleigh Stewart, 18–22 Oct 2026 · benchmark: Canadian Press, CBC, CTV, Al Jazeera, Canadaland, Prime Rogue Inc. · grades: benchmark A−
Stewart's two bylined analyses were the most original mainstream reporting on the fire: first to show it had burned since about noon, and first to obtain the Range Standing Orders. Grade A−; gaps are an untimed gust and an unreconciled "4 p.m." line.