Saiph · star 6 of 7
orion.cucumbs.com · sheet 06 of 07
Field Notes and Scorecard
Every claim Orion has logged, the public source behind it, and the date it will be checked. The Scorecard keeps the record honest: it counts only claims that have been checked, and failed claims stay on the page.
Carried in Sheet 05 showed where a check goes. This is where its result lands, held or failed, and stays.
- Root · Orion's own pairing
- Axiom 5: Resistance fights the natural harmony of movement, evolution, and time, while alignment shapes it.
- Loop phase of the route
- Synthesize and act
- Front door
- cucumbs.com
06.1 Field Notes
-
HYPOTHESIS When CyberSN publishes a full-year 2026 count of United States core cybersecurity job postings, the total will fall within or below the 288,911 to 325,540 range the report projected from the January–September pace, and will remain below the 2025 full-year total of 392,356.
CyberSN's report of United States postings through September 2026 counted 235,786 core postings, down 26.4% from 320,210 in the same months of 2025, and projected a full-year range of 288,911 to 325,540.
Source · Check by 31 March 2027
-
HYPOTHESIS In a later CyberSN Cybersecurity Jobs Report that covers at least January through March 2027, United States AI Security Engineer postings will still be higher than in the same months of 2026.
The same CyberSN report recorded 2,409 AI Security Engineer postings from January to September 2026, a 268.3% increase from the same months of 2025, even as overall postings fell.
Source · Check by 15 May 2027
-
HYPOTHESIS By 31 March 2027, the California Office of Emergency Services will have published, or announced a date for, the first anonymized and aggregated annual report on reviewed critical safety incidents that state law requires beginning 1 January 2027.
California Business and Professions Code section 22757.13 requires that annual report beginning 1 January 2027, and also requires frontier developers to report critical safety incidents to the Office within 15 days of discovery.
Source · Check by 31 March 2027
-
HYPOTHESIS By 31 May 2027, the UK government will have opened a public consultation on using its NIS regulation-making powers to introduce measures for agentic loss-of-control incidents.
A GOV.UK factsheet updated on 3 November 2026 states that the government will consult on using those powers to add security, notification, evidence-preservation, and post-incident review measures.
Source · Check by 31 May 2027
-
HYPOTHESIS When Gartner next publishes a public forecast update or year-end estimate of worldwide IT spending for 2026, that figure will be at least $6 trillion.
Gartner's 27 July 2026 press release forecast worldwide IT spending of $6.37 trillion in 2026, up 14.2% from 2025, with data-center systems the fastest-growing segment.
Source · Check by 30 April 2027
06.2 Scorecard
No claims resolved yet
The filters and date sort need JavaScript. Without it, every claim is listed below.
No claims match those filters.
06.3 Calibration simulator
Simulation. Nothing here changes the real Scorecard above. The outcomes are ones you pick or the page draws at random, to show how a hit rate and calibration behave once claims start to resolve.
A hit rate alone can flatter. If you say you are 90% sure and are right 60% of the time, you are overconfident even with a decent hit rate. Calibration asks whether your stated confidence matches how often you turn out right. The Brier score sums that up: 0 is perfect, 0.25 is what always saying 50% earns, and higher is worse.
This tool needs JavaScript. Everything it teaches is also written out in plain text on this page.
Part 1 · Pretend Orion's five claims have resolved
| Claim | Confidence | Pretend outcome |
|---|---|---|
| fn-001 Cybersecurity hiring | ||
| fn-002 Cybersecurity hiring | ||
| fn-003 AI regulation | ||
| fn-004 AI regulation | ||
| fn-005 Tech market |
- Resolved
- 0 of 5
- Hit rate
- —
- Average stated confidence
- —
- Brier score
- —
No pretend outcomes yet. The real Scorecard says: No claims resolved yet.
Part 2 · Many claims, and why five is not enough
dashed stated confidence · solid running hit rate
- Final hit rate
- —
- Gap from stated
- —
- Brier score
- —
Press Run to simulate.
06.4 How this page works
- Notes are added by hand to a file, then the site is rebuilt.
- A claim that has not been checked yet is labeled untested/open.
- When a claim fails, it stays on the page.
- The hit rate counts only claims that have already been checked. If none have, the page says so.
- The raw notes are also published as notes.json.