Method, written 18 Nov 2026
How mood is measured
CISO Mood reads public posts about security, gives each a mood from −100 (alarm) to +100 (upbeat) and prints the average.
The figure is a reading of what people wrote in public, not a survey.
The scale
Five steps, printed in three inks. Pink is worry, green is relief, petrol is the record.
- Alarm −100 to −60 People are frightened or angry.
- Tense −60 to −20 Worry outweighs everything else.
- Steady −20 to +20 Matter-of-fact talk, or worry and relief in balance.
- Curious +20 to +60 More interest and relief than worry.
- Upbeat +60 to +100 People are pleased, proud or relieved.
Two kinds of voices
Security leaders are people who hold a security seat, such as a chief information security officer or a head of security. They count in this layer only when they post under their professional identity on an open network. This layer is small. On the sheets it is drawn as a ring.
The security community is everyone else who works in security and talks about it in the open: analysts, engineers, researchers, consultants. This layer carries most of the volume. On the sheets it is drawn as a dot.
Every figure says which layer it comes from. A sheet says "security leaders feel" something only when the figure comes from the first layer.
Which posts count
Only posts about security count. A classifier reads each post and marks it on-topic when the post itself is security talk: an attack, a breach, an outage, a flaw, how it was handled, defences, identity, rules, or the security job. A reply about politics, prices or anything else in the same thread is left out. It does not count toward any mood and it never appears on a sheet.
The posts on the site today come from Bluesky, Hacker News, Mastodon (hachyderm.io), Mastodon (infosec.exchange) and Mastodon (mastodon.social). The site reads only networks that publish posts openly and allow them to be read. Its reader identifies itself as CISOMoodBot and never signs in, never goes behind a paywall and never pretends to be a browser.
How a post is scored
Each on-topic post gets three labels: a topic, a mood on the five-step scale, and the organisation, product, rule or flaw it is about. A fast language model gives the labels. A stronger one reads again the posts the first is unsure about.
Jokes and sarcasm get their own label. They are counted as posts but left out of the average, because they cannot be read as plainly good or bad.
The floors under every figure
A figure from a handful of posts is not a reading, so the sheets hold to these floors.
- The world figure needs 100 posts. When the last hour has fewer, it is read over the last 6 hours, or the last 24, and the label says which.
- A topic, an incident, an hour or a day shows a mood only at 10 posts or more. Below that the sheet prints the count and says there are too few posts to read.
- An incident gets a sheet of its own once 15 different voices have posted about it, and a topic or a day once 15 have.
Where a post sits on the globe
At the city the voice gives in its own public profile, and never anything finer than a city. Or at the head office of the organisation the post is about, when the post is part of an incident or a rule story. A post with neither still counts in the mood; it is just not pinned to a place.
People
- Nobody is named on a sheet. A voice is shown as its layer, and its role and sector when it states them.
- The sheets print short excerpts with a link to the post itself, never the whole post.
- A post deleted at its source leaves the site when the reader next checks it. It checks up to 200 posts every five minutes, the ones printed on a sheet first; a deletion on Bluesky usually arrives within minutes.
- No sheet follows one person over time. People appear in the stream; history is kept as totals.
- Anyone can ask to be left out. It applies at once, except when more than 30 requests arrive in an hour: then later ones wait until the earlier ones are applied.
Visitors
The site sets no cookies and runs no tracking script. Every file a page loads comes from the site itself.
Its servers keep a log of requests: the address a request came from and any address a proxy passed along, the time, the page asked for with anything after a question mark, the page that linked to it, what the browser says about itself, and technical details of the answer such as its size. A log is deleted after 90 days.
Reading the figures elsewhere
The same totals are published as plain data. The current reading, as of 17:10 UTC, 21 Nov 2026, is on the front sheet.