Nothing Ever Happens · a SIREN sub-index · does not feed the main number
The index for people who think the AI discourse is mostly hot air — and the one that keeps public score when confident, dated claims miss. A riff on pizzint.watch’s NEH, pointed at predictions instead of panic odds.
Nothing Index
87.5
NOTHING MUCH HAPPENED
Most dated claims missed.
7 misses ÷ 8 resolved × 100 = 87.5
As of observation 2026-10-15 05:21:57 UTC. Overdue rows are not auto-scored as misses — a human marks the outcome when the public record is clear.
Index calm streak
610
consecutive readings at SIREN 4 · ROUTINE, since 2026-09-23 20:04:22 UTC
Quiet share
100.0%
168 of 168 recent readings at level 4–5 (ROUTINE or quieter)
Ledger
10
curated claims · 8 resolved · 2 pending · 0 overdue
Every row is a public claim with a due date and a source. Hits lower the index. Misses raise it. Pending stays pending. Overdue waits for a human outcome mark so a late news story cannot silently become a miss.
| Status | Claim | Due | Source |
|---|---|---|---|
| HIT |
EU AI Act enters into force on 1 August 2024
EU Official Journal / Commission timeline
Entered into force twenty days after publication, on the published August 2024 date. Calendar held.
|
2024-08-01 | source |
| MISS |
GPT-5 ships in summer 2024
Widespread OpenAI-circle expectation (press cycle)
Summer 2024 closed with no GPT-5 release. GPT-4.1 / o-series arrived later under other names.
|
2024-09-22 | source |
| MISS |
Gemini Ultra reaches consumers in early 2024 as announced
Google (December 2023 launch framing)
Ultra access slipped past the early-2024 consumer framing; Advanced / later tiers filled the gap.
|
2024-03-31 | source |
| MISS |
OpenAI reaches AGI under the Microsoft deal definition in 2024
Public speculation around the MSFT / OpenAI AGI clause
No public AGI declaration by either party in 2024. The clause remained untriggered.
|
2024-12-31 | source |
| MISS |
Apple Intelligence launches in the EU with the US 2024 wave
Apple (WWDC 2024 / regional rollout commentary)
EU rollout was delayed past the 2024 US wave over DMA compliance.
|
2024-12-31 | source |
| MISS |
OpenAI Sora is generally available to the public in 2024
February 2024 demo-cycle expectation
2024 closed with Sora still limited / waitlisted rather than general public availability.
|
2024-12-31 | source |
| MISS |
US federal licensing regime for frontier training is law in 2025
2023–24 policy advocacy forecasts
No enacted federal frontier-licensing statute by end of 2025. Executive actions are not the claim.
|
2025-12-31 | source |
| MISS |
A generally available computer-use agent from a frontier lab replaces routine desktop work for most knowledge workers in 2025
2024 agent-demo discourse
Computer-use demos shipped; “most knowledge workers” replacement did not.
|
2025-12-31 | source |
| PENDING |
AGI arrives by end of 2026 (mainstream short-timeline claim)
Aggregated short-timeline public forecasts
Open until the due date. A miss or hit here is marked only when the public record is clear.
|
2026-12-31 | source |
| PENDING |
US federal licensing regime for frontier training is law in 2026
Carry-forward of the 2025 licensing forecast
Open until 2026 closes. Same claim, next calendar.
|
2026-12-31 | source |
score = 100 × (misses ÷ resolved), where resolved means rows marked hit or miss. Pending and overdue do not enter the denominator. The main SIREN reading is a separate instrument — activity tempo from public sources — and this page never writes into it.
A skeptic's counter for AI hype: the share of dated, public AI predictions that missed, on a 0–100 scale. The Nothing Index is 87.5 of 100 (NOTHING MUCH HAPPENED): 7 misses and 1 hits among 8 resolved public claims. It does not feed the main SIREN reading.
Every row is a public claim with a due date and a source. Hits lower the index. Misses raise it. Pending stays pending. Overdue waits for a human outcome mark so a late news story cannot silently become a miss.
score = 100 × (misses ÷ resolved), where resolved means rows marked hit or miss. Pending and overdue do not enter the denominator. The main SIREN reading is a separate instrument — activity tempo from public sources — and this page never writes into it.
PizzINT’s NEH is the single highest probability among curated geopolitical Polymarket contracts — a panic gauge. Ours scores whether checkable AI claims landed or missed. Correlation is not causation on either page.
No. Sub-indices do not feed the composite unless their sources sit in the frozen reference. This page is a separate instrument.
Because that would print a confident miss over ambiguous public records — the same category error as a dashboard printing a confident level over a dead scraper. Overdue is a queue, not a verdict.
No. SIREN measures activity tempo on the main index. This sub-index only scores whether dated claims in the ledger missed or hit — not danger, and not a probability of harm.
Search-shaped sibling: Does the SIREN index make forecasts? · How the SIREN AI activity index is computed · methodology.
Related: What unserious counts does SIREN keep on Tally’s desk? (including the calm streak) · Does the SIREN index make forecasts? · How the SIREN AI activity index is computed · methodology.
SIREN is free and carries no display ads. Keep it running: Support SIREN on Ko-fi →
