Information Theory & Network Economics
The Information
Gleick's history of information theory lands on one operational idea for investors: information is measured by surprise, not by volume — and almost everything a market feed delivers is volume.
The big picture
The book traces how information became a measurable quantity: Shannon's insight that a message carries information only to the degree it reduces uncertainty — the predictable parts are redundancy, not content. Flooded channels don't produce more knowledge; they produce more noise around the same few bits of signal. The core bet for an investor: your scarce resource is not access to information but the ability to price each item's surprise value before it consumes attention and triggers action.
Why it matters now: the AI news cycle is the highest-volume, lowest-surprise channel in markets. Model launches, partnership headlines and capex re-announcements mostly restate what prices already carry — while the few genuinely uncertainty-reducing data points arrive quietly and get equal billing in the feed.
The 3 strategic pillars
-
Information = surprise
A message's value is how much it changes what you believed — an expected announcement carries near-zero bits.
Operational test: write down your prior first. If the news doesn't move it, it wasn't information — whatever the headline size.
-
Redundancy dominates every channel
Most of any feed restates, rephrases and re-amplifies content already transmitted.
The nth story on the same event adds correlation, not knowledge — but each repetition still costs attention and still nudges System-1 confidence.
-
The flood is the modern condition
When transmission became free, the bottleneck moved from getting information to filtering it — the filter IS the edge.
A deliberate input diet with explicit scoring beats a bigger feed: fewer independent sources, each priced for surprise, actionability and half-life.
What a Closelook reader does with it
The working use is an input audit: list what you actually consume in a week, score each source for how often it changed a decision, and cut the bottom half. Per item, the pack's five-question score (surprise, specificity, actionability, half-life, independence) turns "interesting" into a number. The mistake this prevents is the quiet one — a portfolio steered by the loudest redundant channel while the two datapoints that mattered scrolled past unweighted.
The bridge to the Closelooknet approach
Filtering-for-surprise is Closelooknet's operating principle made explicit. Sentinel Tickers is a Shannon idea in equity form — a handful of names chosen because their moves carry disproportionate information about the whole system. The Wire and the Tape exist to compress the day into its few non-redundant items, and the Weekly Signal is deliberately one signal per week, not a feed. Run the pack's source audit against your own inputs and you'll see why the house publishes less, not more.
Action-Kit — from theory to practice
Tooling & data
| What you need | Where to get it | Cost |
|---|---|---|
| Your actual weekly input list The raw material of the audit — feeds, newsletters, terminals, feeds-of-feeds | Fifteen honest minutes with a text file | Free |
| A prior journal The surprise test needs a written prior — otherwise hindsight scores everything as expected | Any notes tool; the pack's scorer includes a prior field | Free |
The formulas
-
Item information score
Score = 2·Surprise + Specificity + Actionability + Half-life + Independence (each 0–2)- Surprise — did it move your written prior?
- Specificity — numbers or adjectives?
- Actionability — does any decision change?
- Half-life — does it still matter in a month?
- Independence — new source, or the nth echo?
Surprise is double-weighted on purpose — it is the Shannon term; the rest are practical multipliers.
-
Source signal rate
SNR(source) = items scoring ≥ 6 / total items consumed- A week of scored items per source
Rank sources by this and the cut list writes itself.
Applied Pack · free members
Gleick Applied Pack
The input diet as a working audit: score every item for surprise value, rank your sources by signal rate, cut the redundant half of your feed with numbers instead of guilt.
- Information_Audit.xlsx — the 5-question item scorer with the double-weighted surprise term, plus a source-ranking sheet that computes each feed's signal rate from your scored week
- input_audit.py — stdlib-only CLI: log items as you consume them, get the weekly source ranking and the cut-list suggestion from your own data
- README.txt — scoring rubric with worked examples and the educational-use disclaimer
Pack security
Macro-free Excel · plain-text Python you can read before you run it · no installers, no network access — the code works only on files you provide. Served only from closelook.net; we never distribute through download portals or email attachments. How to verify in 30 seconds →
SHA-256 22ce7f28c4c52795514646aeab89b240b7d015365702879ee382903da44ff012
The pack is a free-membership download — no card, free forever. Membership also delivers the Daily Pulse and Weekly Signal to your inbox.
Get free access → Already a member? Sign in →
Educational templates — a research diary companion, not investment advice.
Closelook publishes a market diary, not investment advice. This condensed read restates the book's ideas in our own words for education — for the author's full argument, go to the source.