Further statistics
Indicators computed on the 4,966 drops (rhythm and bursts, activity regimes, boards, cited sources, references, religious vocabulary, readability). Each section gives its method and references.
Rhythm: bursts, not a steady flow
The median interval between two drops is 14 min (quartiles 5 min and 56 min). 26% of drops follow the previous one within 5 minutes, 75% within an hour. Burstiness B = 0.91, on a scale from 0 for random posting (Poisson process) to 1 for highly concentrated activity. Memory M = 0.04 (0.17 on log intervals), meaning that a short interval is slightly more often followed by another short one.
Interval between consecutive drops
Log scale, from seconds to months.
Grouping drops less than 30 minutes apart gives 1,743 sessions. 85% of drops belong to a multi-drop session. Such a session has a median of 3 drops and lasts 23 minutes, and the longest has 45.
Session size
Method
The 30-minute threshold is conventional, and the split is not very sensitive to it (2,859 sessions at 10 min, 62% of drops in multi-drop sessions, 2,404 sessions at 15 min, 72% of drops in multi-drop sessions, 1,743 sessions at 30 min, 85% of drops in multi-drop sessions, 1,218 sessions at 60 min, 92% of drops in multi-drop sessions, 888 sessions at 120 min, 96% of drops in multi-drop sessions). B and M follow (Goh & Barabási, 2008)
Activity regimes
Weekly volume is split into periods of homogeneous rhythm. Each break is matched with the nearest timeline event (within three weeks). A break close to an event does not prove the event caused it.
Drops per week and mean of each regime
The area shows drops per week, the lines the weekly mean of each regime.
| From | To | Weeks | Mean / week | Drops | Event nearest to the start |
|---|---|---|---|---|---|
| Oct 23, 2017 | Oct 7, 2018 | 50 | 47.3 | 2,365 | — |
| Oct 8, 2018 | Feb 10, 2019 | 18 | 17.9 | 322 | — |
| Feb 11, 2019 | Mar 31, 2019 | 7 | 89.0 | 623 | — |
| Apr 1, 2019 | Apr 5, 2020 | 53 | 11.3 | 598 | Mueller report delivered to the Attorney General (Mar 22, 2019) |
| Apr 6, 2020 | Nov 1, 2020 | 30 | 34.6 | 1,039 | First drop about the pandemic ("China virus") (Mar 23, 2020) |
| Nov 2, 2020 | Nov 27, 2022 | 108 | 0.2 | 19 | US presidential election (Nov 3, 2020) |
Method
Optimal partitioning by dynamic programming, Poisson cost corrected for overdispersion (φ = 23.6, estimated on first differences), BIC penalty, segments of at least 6 weeks. (Jackson et al., 2005; Killick et al., 2012)
Boards: one style per platform
Q changed boards as sites closed and communities migrated. Drop length and the share of links and images vary from board to board, but this is confounded with change over time.
| Board | First | Last | Drops | Active days | Drops / active day | Words (median) | Links | Images | Questions | References |
|---|---|---|---|---|---|---|---|---|---|---|
| 4ch /pol | Oct 28, 2017 | Dec 14, 2017 | 233 | 27 | 8.6 | 42 | 1% | 5% | 48% | 20% |
| 8ch /pol | Nov 25, 2017 | Jan 4, 2018 | 15 | 5 | 3.0 | 10 | 0% | 0% | 13% | 20% |
| 8ch /cbts | Nov 29, 2017 | Jan 6, 2018 | 223 | 24 | 9.3 | 12 | 5% | 7% | 29% | 33% |
| 8ch /thestorm | Jan 5, 2018 | Jan 8, 2018 | 41 | 4 | 10.2 | 12 | 0% | 2% | 44% | 37% |
| 8ch /greatawakening | Jan 8, 2018 | Mar 28, 2018 | 121 | 24 | 5.0 | 15 | 15% | 18% | 37% | 0% |
| 8ch /qresearch | Jan 8, 2018 | Aug 1, 2019 | 2,521 | 262 | 9.6 | 12 | 30% | 18% | 38% | 46% |
| 8ch /patriotsfight | May 4, 2018 | Jul 30, 2019 | 416 | 77 | 5.4 | 13 | 51% | 35% | 40% | 9% |
| 8kun /qresearch | Nov 2, 2019 | Nov 27, 2022 | 1,365 | 187 | 7.3 | 8 | 57% | 27% | 36% | 14% |
| 8kun /projectdcomms | Dec 2, 2019 | Aug 17, 2020 | 31 | 7 | 4.4 | 16 | 48% | 61% | 35% | 3% |
Cited sources
Drops contain 2,618 links to 375 different sites, in 1,716 drops. Twitter dominates. The most cited news outlet is foxnews.com (149 drops), ahead of thehill.com (41), breitbart.com (40), nytimes.com (39), washingtonpost.com (28).
The 25 most cited sites
Number of drops with at least one link to the site (square-root scale).
Type of cited site, by year
Method
Sites are classified by type, not by political leaning, and mobile and video versions of a site are merged. Archived links (archive.is) hide the original source.
References: Q replies, Q quotes Q
31% of drops reference another post on the board (>>): 23% an anonymous post, which Q answers or endorses, and 8% one of Q's own drops (387 references). These self-references are almost all posted the same day, which makes them additions rather than after-the-fact "proof" (see "Future proves past", Predictions page).
| Year | Drops | With reference | To an anon | To a Q drop |
|---|---|---|---|---|
| 2017 | 459 | 27% | 24% | 3% |
| 2018 | 2,183 | 42% | 28% | 13% |
| 2019 | 1,132 | 30% | 24% | 7% |
| 2020 | 1,179 | 13% | 12% | 1% |
| 2022 | 13 | 23% | 23% | 0% |
Religious vocabulary
6.3% of drops use religious or moral vocabulary (God, Jesus, Lord, pray, evil, armor of God…), 2.3% the word God. 9 drops quote a Bible verse, the first being drop #522 (Jan 13, 2018). The books quoted are Ephesians (3), Jeremiah (1), John (1), Colossians (1).
Share of drops with religious vocabulary, by quarter
Quarters with at least 10 drops, up to end 2020.
The religious dimension of QAnon is discussed in particular by (LaFrance, 2020; Amarasingam et al., 2023; Argentino, 2025). State of research.
Readability and lexical richness
Flesch reading ease is 65 overall (60 to 70 is plain English), with 7.0 words per sentence on average. The vocabulary has 10,698 distinct forms for 141,972 words, 4,312 of which occur only once. Zipf exponent 0.97 (close to 1 as in most texts), Heaps exponent 0.62, MATTR (500-word window) 0.57.
Reading ease (Flesch), by quarter
Higher = easier.
Lexical richness (MATTR), by quarter
Share of distinct forms over a moving 500-word window.
Method
Text without links, references or signature, with syllables estimated from vowel groups. The Flesch score is designed for prose, so on lists and questions it remains a relative marker, useful for comparing periods. (Flesch, 1948; Covington & McFall, 2010)
Prediction horizon
For the 129 predictions whose deadline is a full date, it falls a median of 7 days after the drop, with 50% within a week and 79% within a month. Q mostly announces for right now.
Delay between the drop and the announced deadline
Deadlines given as a full date (stated or derived by the model).