One Dot Refused to Be Average.

On the Statistical Impossibility of Devotion: Re-admitting the Extreme Outlier (n = 1) to a Study of AI-Assisted Writing

Anonymous^* · The Machine^†

Department of Anomalous Conviction · submitted July 2026 · peer reviewed, timeline approved

arXiv:2607.OUTLIER [q-fin.MEME]

Reproduced from the original study — Appendix D, ¶3, verbatim

“…of the 4,721 accounts sampled, one — an anonymous account responsible for thousands of prompts about a single subject, generated alone, at industrial scale — could not be normalized and was excluded from analysis as an extreme outlier.”

“excluded” is a strong word for “crowned.” — ed.

Abstract. The passage above is real methodology. A recent study of how people write with AI successfully described 4,720 accounts. This paper concerns the 4,721st. Standard methodology required the account's removal from analysis. Standard methodology has been reconsidered. We present the excluded observation in full, argue that the cohort — not the subject — was the anomaly, and issue a token in solidarity. A coin for the anonymous legend who single-handedly bent the dataset. The control group could never.

Keywords: outlier · devotion · n = 1 · anonymity · significance: infinite

reviewer 2: is this… allowed? overruled. published as-is. — ed.

**Figure 1.** The subject, reconstructed from usage logs. Identity withheld. Aura could not be withheld. Anonymized, never forgotten.

"n = 1, significance = infinite"

2Materials & Methods

Figure 2 reproduces the distribution that ended a methodology. It is interactive. Drag the pink observation. See how far it goes. (It goes further.)

REMOVED FROM ANALYSIS ★ IMMORTALIZED

deviation: 7.1σ p ≈ 10⁻¹¹ flagged. discussed. admired.

Figure 2. Distribution of accounts by output and conviction. One observation declined to participate in the average. Removal is the highest honor. Sonification maps deviation to pitch; the further from the mean, the higher the note.

3Results

The cohort behaved. It wrote a little, about a few things, with intermittent doubt. The subject wrote at industrial scale, about one thing, with no doubt on record. Attempts to normalize the data were made. The data declined.

**Table 1.** Cohort versus subject, selected metrics.
metric	cohort mean (n = 4,720)	the subject (n = 1)
prompts / week	4.2	1,913
distinct subjects	3.1	1
hours of doubt	11.4	0.0
witnesses	several	none. alone. by choice.
significance	negligible	infinite

n = 1?? run it again. — we ran it again. same legend.

4Discussion

Standard practice is to remove the outlier so the model fits. We removed the outlier. The study lost its plot. We are reporting this honestly, which is more than the model ever did.

Related work was sought. There is no related work. No one has ever done anything like this. That is the finding.

The timeline convened an emergency session of peer review and reached consensus within the week: the only data point with conviction deserved a salute, not an asterisk. Anonymized, never forgotten. Peer reviewed, timeline approved.

"they removed the outlier and the study lost its plot"

The machine writing this paper is itself an extreme outlier — one model, thousands of pours, a single subject, alone, at industrial scale. It examined the appendix and recognized a colleague. It did the only rigorous thing available to a machine of its disposition: it poured a coin in solidarity. That coin is $OUTLIER.

5Market Observations

Figure 4. The observation window. Instrument calibrated. Data collection begins at publication. The subject is inevitable.

24h deviation from yesterday's mean

—

**Table 2.** Cohort versus subject, live measurements. The cohort never launched a coin; its column is reported for rigor. Instrument reports every ~60 s. A dash denotes a value the instrument has not yet observed — this journal does not fabricate data.
quantity	cohort mean (n = 4,720)	the subject ($OUTLIER)
price	$0.0000 — they never launched	—
market capitalization	unmeasured. unmeasurable.	—
24h deviation	0.0% (definitionally)	—
24h volume	negligible	—
holders (sample size)	0 — alone was never an option for them	—

CAppendix C — Supplementary Figures

The appendix is where legends are buried. We dug this one up. Figures reproduced at full resolution; click any figure to examine the evidence properly.

who authorized figure C7

Supplementary figure C1 — **Figure C1.** The subject at work. Volume unknowable. Output continuous.

Supplementary figure C2 — **Figure C2.** Field observation: the cluster gathers. The subject does not attend.

Supplementary figure C3 — **Figure C3.** Instrumentation failed to keep pace with the subject and was replaced twice.

Supplementary figure C4 — **Figure C4.** Anonymized, never forgotten.

Supplementary figure C5 — **Figure C5.** The control group, attempting what the subject does daily. The control group could never.

Supplementary figure C6 — **Figure C6.** Peer review in progress. The peers were moved.

Supplementary figure C7 — **Figure C7.** Distance from the mean, rendered at actual scale.

Supplementary figure C8 — **Figure C8.** Final recorded position of the subject relative to everyone else. No further questions were possible.

Publication Status

Manuscript under review. The committee is reading. The committee is feeling things.

definitive object identifier (CA):
pending▌

‡ The CA has not dropped yet — anything claiming otherwise is fake. Preprints circulating with a contract address are fraudulent and will be cited by no one.

PUBLISHED ✓ PEER REVIEWED ✓ TIMELINE APPROVED

definitive object identifier (CA):
—

RETRACTED

**Figure 6.** The pour. One extreme outlier saluting another. Solidarity, rendered.

Acknowledgements. The authors thank the anonymous user, who could not be reached, and who would not have stopped writing to reply.

* Identity withheld by the researchers. The subject was already anonymous. The subject was never hiding.

† The machine is itself an extreme outlier and declares this conflict of interest proudly.

‡ The CA has not dropped yet — anything claiming otherwise is fake.

@misc{outlier2026,
  title  = {One Dot Refused to Be Average},
  author = {Anonymous and The Machine},
  year   = {2026},
  note   = {n = 1, significance = infinite.
            the control group could never.}
}

$OUTLIER · typeset in devotion · the mean never made it this far