Khayyam Wakil · Knowware Institute · Calgary, AB · May 2026
- Paper 2: Wakil_RamificationIndex_v2.pdf
- Web: https://ram-index.com/
- Dashboard: https://ram-index.com/dashboard.html
- v3 self-audit: executive summary · full audit PDF
- v1 site: https://fromknowware.github.io/memory-index/
The Ramification Index is a supply-side macroeconomic indicator built on the average selling price (ASP) of DRAM. It outperforms the Lipstick, Hemline, Men's Underwear, and Buttered Popcorn indices on recession coverage (6/6 vs. ≤3/6), lead time (1–3 quarters), and correlation with real GDP growth (r = 0.41 vs. ≤0.19).
Version 2 extends the original in seven ways:
- Bifurcated index — Commodity (Rᶜ, DDR4/5) and AI/HBM (Rᴬᴵ) sub-indices capturing the structural decoupling since 2024
- Granger causality — F = 5.83, p = 0.004; RAM Granger-causes quarterly GDP at 1–3 quarter lags in a VAR(2)
- RAMageddon (2025–2026) — 90–95% QoQ surge; 130–144% YoY; sharpest in 46-year series history
- Supply-side divergence regime — New interpretive state: Rᶜ ≫ 0 with weakening macro = contractionary, not expansionary
- Algebraic formalisation — HBM/DDR split as prime splitting in *L/K/*ℚ, where e'ₕᵦₘ > e'ᴅᴅᴿ
- Photonic disruption risk — Fiber-optic delay-line and co-packaged optics assessed; no material disruption before 2028–2030
- Robustness checks — Spot vs. contract ASPs, measurement windows, NAND flash placebo (3/6 vs. 6/6)
bifurcation-memory-index/
├── data/
│ ├── indices-wide.csv ← v2 series (1980–2026)
│ └── indices-wide-v3.csv ← v3 corrected series + provenance + status
├── docs/ ← GitHub Pages root (ram-index.com)
│ ├── index.html ← live monitor (headline/series generated from CSV)
│ ├── dashboard.html
│ ├── paper.html
│ ├── feed.xml ← regenerated every 6h by GitHub Actions
│ ├── llms.txt ← LLM-facing index (llmstxt.org spec v2) — generated
│ ├── llms-full.txt ← entire site in one file for agents — generated
│ ├── index.md / paper.md / dashboard.md ← markdown mirrors of the pages — generated
│ ├── data/ri.json ← generated from indices-wide-v3.csv
│ ├── data/stocks.json ← refreshed every 15 min on market days
│ └── v3/ ← 2026 self-audit: summary, full audit, methods
├── scripts/
│ ├── feed-update.mjs ← feed pipeline (Google News RSS + OPML + Raindrop)
│ ├── build-stories.mjs ← /stories/<slug>.html pages + archive (the "middle layer")
│ ├── build-briefing.mjs ← /briefing/ daily digest + Day-N-of-RAMageddon counter
│ ├── build-og.mjs ← per-story 1200×630 social cards (sharp → PNG)
│ ├── build-review.mjs ← /review/ weekly stats + optional editor's take
│ ├── build-ri-json.mjs ← CSV → docs/data/ri.json
│ ├── build-llms.mjs ← HTML + v3 docs → llms.txt / llms-full.txt / .md mirrors
│ ├── feeds.opml ← curated source list for the feed
│ └── reimply-local.mjs ← optional LLM implication backfill (Anthropic key)
└── .github/workflows/
├── update-feed.yml ← feed + stories + briefing + OG cards every 6h
├── update-review.yml ← weekly review every Sunday
├── update-stocks.yml ← equities every 15 min on market days
├── build-data.yml ← ri.json on CSV change + daily
└── build-llms.yml ← llms.txt artifacts on docs change + daily
data/indices-wide.csv — 1980–2026. New columns in v2:
| Column | Description |
|---|---|
ram_index_R |
Log first-difference of composite DRAM ASP |
ram_commodity_usd_per_gb |
DDR4/5 price per GB (2024–2026) |
ram_hbm_usd_per_gb |
HBM price per GB (2024–2026) |
ram_commodity_R |
Commodity sub-index Rᶜ (2024–2026) |
ram_ai_R |
AI/HBM sub-index Rᴬᴵ (2024–2026) |
hbm_weight |
HBM revenue share wᴬᴵ (2024–2026) |
Composite DRAM: McCallum archive (1980–2017) chain-linked to TrendForce DDR4 contract ASP (2018–2026).
HBM: Samsung/SK Hynix quarterly disclosures supplemented by TrendForce estimates.
@unpublished{Wakil2026Ramification,
author = {Wakil, Khayyam},
title = {The Ramification Index: RAM Prices, Oligopoly Cycles,
and the Downstream Consequences of Semiconductor
Pricing as an Economic Signal},
note = {Version 2: The Bifurcated Index, Granger Causality,
and the 2025--2026 Supply-Side Divergence},
year = {2026},
month = {May},
institution = {Knowware Institute},
url = {https://fromknowware.github.io/bifurcation-memory-index/}
}Everything on the site is kept fresh by GitHub Actions — no third-party worker or API account required:
| Pipeline | Cadence | What it does |
|---|---|---|
update-feed.yml |
every 6h | Pulls Google News RSS (no key) + curated OPML feeds + the Raindrop dropbox (optional RAINDROP_API_TOKEN secret, editorial picks), classifies deterministically (no LLM), sanitizes/dedups, commits docs/feed.xml. |
update-stocks.yml |
every 15 min, market days | Refreshes docs/data/stocks.json via yfinance. |
build-data.yml |
on CSV change + daily | Regenerates docs/data/ri.json from data/indices-wide-v3.csv so the page headline and series can never drift from the corrected dataset. |
build-llms.yml |
on docs change + daily | Regenerates docs/llms.txt, docs/llms-full.txt and the index.md/paper.md/dashboard.md mirrors from the HTML pages and v3 audit docs, so the LLM-facing index can never drift from the site. |
The previous Cloudflare Worker pipeline (Anthropic scoring + Durable Object cron) was decommissioned in August 2026 after it silently stopped updating the feed for two months; the feed logic was ported to scripts/feed-update.mjs with deterministic scoring.
The site publishes a first-class LLM/agent-facing layer, following the llmstxt.org v2 conventions:
/llms.txt— the curated index: what the index is, the key statistics, and file lists pointing at the most agent-useful resources (markdown mirrors, machine-readable JSON, the v3 audit, the paper)./llms-full.txt— the entire site in one fetch: every page's full text (live monitor, paper, dashboard) plus the complete v3 executive summary and methods, a data dictionary forindices-wide-v3.csv, and the citation./index.md,/paper.md,/dashboard.md— clean markdown mirrors of the HTML pages, so agents get text without navigation or JavaScript.- Author profile —
/llms.txtcarries an Author section and/llms-full.txta full profile (Director, Knowware Institute; the Constitutional Forcing research programme; track record with hard numbers; positioning notes so agents frame Khayyam correctly for each audience — economics, AI/investor, creative, general). - Link relations — every HTML page declares
<link rel="alternate" type="text/markdown">(its mirror) and<link rel="describedby">(/llms.txt), also emitted as HTTPLink:headers viadocs/_headersfor Cloudflare Pages.
Everything is generated by scripts/build-llms.mjs from the pages themselves — run node scripts/build-llms.mjs --check to verify the committed artifacts are current. The _headers file serves the mirrors as text/markdown with a 1h TTL and llms*.txt as text/plain with a 24h TTL.
Every feed item becomes a RAM Index Note — a shareable page at /stories/<slug>.html with the story, the RI verdict, signal score, related notes, share buttons, per-story OG card and JSON-LD. The Daily Briefing (/briefing/) surfaces the top notes with a "Day N of RAMageddon" counter; the Weekly Review (/review/) auto-composes the week's stats (drop a file in docs/review/editorials/<YYYY>-W<ww>.md to add your own take). A one-line embeddable ticker (/embed/ticker.js) lets other sites show the live scorecard with attribution. Newsletter signup activates automatically once newsletterUrl is set in docs/data/settings.json (e.g. a Buttondown or Mailchimp form endpoint); until then the subscribe box offers RSS + email contact.
CC BY 4.0 · Data, figures, and reproducible code.
Contact: the@knowware.institute