Virtual Arena AI · Sofia Pulse

Radar

Cross-source relationships that survived evidence, counter-signal and falsification gates.

PRIMARY supports a decision; WATCH is an analytical thesis; context never moves up to fill space.

How to read Radar

Where do independent sources converge, diverge or still fail to support a decision?

PRIMARY has cleared evidence and decision gates. WATCH is analytically useful but still has an open condition. CONTEXT explains the setting without being promoted to fill space.

Decision nowUnder watchContext evidence

PRIMARY

decision

WATCH

under watch

Tests

2temporal

None this cycle: Relations · Hypotheses

PRIMARY · decision now

Decision now

Only theses that cleared the full evidence, falsification and decision contract.

WATCH

Under watch

Theses with analytical value and at least one gate condition still open. Not a decision; they show what to track and what would overturn them.

Observations

What the charts show

Facts and relationships directly visible in the data. They do not receive insight framing without analytical lift.

Pre-registered temporal tests

Pairs declared before measurement and tested on daily changes, not only on levels. Tests that do not qualify stay here as context — not a recommendation or a decision signal.

AI funding rounds (30 days) × Job postings mentioning AI (30 days)

pre-registered test · did not qualify

why it did not qualify: The source mix changed within the window; the comparison is not like-for-like.

also: the apparent association between the levels disappears once the shared trend is removed.

test figures

window: 64 continuous days (Aug 10 – Oct 12)
60 aligned daily changes
levels: correlation 0.66
daily changes: correlation 0.23
lags tested: 0–7 days · best: 3 days
corrected chance of coincidence: not computable with this history
continuous history for the coincidence test: 36 of 39 required shifts (3 missing)
earliest date this history could be complete: Oct 15, if both series keep being collected daily without gaps; this does not guarantee the test will qualify

AI funding rounds (30 days)

20 → 18

Job postings mentioning AI (30 days)

692 → 639

actual daily levels · each chart on its own scale

Vercel AI SDK npm downloads (30-day change) × Job postings mentioning AI (30-day change)

pre-registered test · did not qualify

why it did not qualify: The source mix changed within the window; the comparison is not like-for-like.

also: the apparent association between the levels disappears once the shared trend is removed.

test figures

window: 26 continuous days (Aug 20 – Sep 14)
24 aligned daily changes
levels: correlation 0.86
daily changes: correlation 0.20
lags tested: 0–7 days · best: 1 day
corrected chance of coincidence: not computable with this history
continuous history for the coincidence test: 0 of 39 required shifts (39 missing)

Vercel AI SDK npm downloads (30-day change)

25.4% → 13.4%

Job postings mentioning AI (30-day change)

27.4% → -23.7%

actual daily levels · each chart on its own scale

Guides for tracking technology, AI and security

AI Radar Cybersecurity Radar AI Job Market Technology Events Critical Vulnerabilities