Why people feel sleepy on 15–31 May 2026 — a DoWhy causal analysis: Earth's rotation is not to blame

A news headline went around the networks: "something unprecedented is happening to Earth's rotation." The popular speculation followed immediately — that is why everyone feels sleepy. We treated the question the way we treat any causal hypothesis: we built the causal graph and ran it through a formal inference model (DoWhy). The result is clear — Earth's rotation is the only candidate factor without a causal path to sleepiness. The rest of this piece shows what actually causes it over 15–31 May 2026.

1. What the rotation study actually says

The article that started the wave refers to a study by teams at the University of Vienna and ETH Zürich. Their conclusion is the opposite of what the sensationalist headline suggests: the melting of the polar ice caps — about 1,000 gigatons of ice shifted from the poles toward the oceans and the equator — redistributes the planet's mass and slows down the rotation. The effect: days are lengthening by roughly 1.33 milliseconds per century, a rate described as unprecedented over the last 3.6 million years.

Beware of a common confusion: there are two different phenomena being mixed up in the press. One is the slow, millennial slowdown caused by ice melt (the one in the study). The other is the short-lived speed-ups of a few days — 9 July 2025 was the shortest day measured with modern methods, about 1.3–1.6 milliseconds under 24 hours, followed by other short days in July and August. Both are measured in milliseconds.

2. Why rotation physically cannot make you sleepy

Regardless of which of the two phenomena we mean, the magnitude is on the order of milliseconds out of a day of 86,400,000 milliseconds — a variation below any threshold perceptible to the human body. The circadian rhythm, the one that regulates sleep, is synchronized by light (via melatonin and serotonin), not by the planet's angular velocity. There is no biological mechanism by which a millisecond more or less in the length of the day could reach the brain. The correlation in the headline is pure narrative coincidence.

3. The formal causal test (DoWhy)

Rather than rely on intuition, we formalized the question. We built a causal graph (DAG) with the plausible factors and ran the estimand identification step from the DoWhy library for each candidate cause. Identification is purely topological: it states whether there is any causal path through which a factor can influence the outcome, and what must be controlled for in order to measure it correctly. We did not estimate effects on made-up data — we tested the structure alone.

The graph starts from domain knowledge: the season (photoperiod / spring asthenia) is both a direct cause of fatigue and a common cause for the heat wave and pollen — i.e. a classic confounder. Geomagnetic activity is exogenous (it comes from the Sun). Social perception ("everyone is tired") is a direct reporting cause. And Earth's rotation was deliberately introduced as a control node, with no edge to fatigue.

Candidate factor	DoWhy verdict (estimand identification)
Earth's rotation	❌ Not identifiable — no causal path to fatigue. Nothing to estimate.
Season / spring asthenia	✅ Direct identifiable effect (exogenous root cause).
Heat wave / heat dome	✅ Identifiable via backdoor, adjusting for season (confounder).
Pollen / antihistamines	✅ Identifiable via backdoor, adjusting for season (confounder).
Geomagnetic activity	✅ Direct identifiable effect (exogenous, no confounder).
Perception / "quiet burnout"	✅ Direct identifiable effect (reporting cause).

The test's conclusion is topological and unambiguous: rotation is the only candidate for which DoWhy finds no estimand — because there is no causal arrow toward fatigue. By contrast, the heat wave and pollen require adjustment for season: this is where the illusion of correlation is born, because the season moves them all at the same time. Anyone who confuses "they appear together" with "one causes the other" falls into exactly the confounder trap.

4. The real causes, in order of evidential strength

4.1 Spring asthenia (the dominant cause)

It is not a disease but a normal reaction to the seasonal transition. Lengthening days alter melatonin and serotonin secretion, on top of the fatigue accumulated over winter. Typical symptoms: drowsiness, difficulty concentrating, irritability — exactly the picture described in this period.

4.2 The late-May heat dome

Romania's National Meteorological Administration confirmed the first heat wave of 2026: a "heat dome" formed over Europe reached Romania after 25 May, with temperatures up to 33 degrees Celsius in the shade — unusual for this time of year. Across Europe, 26 May 2026 officially became the hottest May day in France's measurement history. The fatigue mechanism is direct: heat and high atmospheric pressure ruin night-time sleep and cause daytime drowsiness. For the second half of the window (25–31 May), this is probably the strongest real physical factor.

4.3 Pollen at its peak (plus the antihistamine effect)

In May, birch pollen (declining after April) overlaps with grass pollen (rising), producing maximal symptoms in sensitized people. On top of the allergic "fog" comes the drowsiness induced by antihistamines — a factor many take daily without connecting it to fatigue.

4.4 The geomagnetic storm of 15–17 May (the only real "cosmic" cause)

The irony is that if there is an influence "from the sky," it does not come from rotation but from the Sun. NOAA's Space Weather Prediction Center confirmed a G1-level geomagnetic storm (Kp index = 5) for 15 May 2026, with a G2 watch, followed by G1 watches for 16–17 May — triggered by a coronal hole (a high-speed solar wind stream), not by a direct flare.

Here the literature exists and is not negligible, but it is correlational. Human studies have measured lower night-time melatonin excretion on days with elevated geomagnetic activity (above thresholds on the order of 30 nanotesla per 24 hours, with effects reported in the 15–80 nT range). A scoping review found significant associations between geomagnetic activity and cardiovascular events in 28 of 36 studies (the remaining 8 found no effect), and an epidemiological study across 263 U.S. cities linked geomagnetic disturbances to increased mortality. The proposed mechanism involves cryptochrome proteins — already components of the circadian clock — which could mediate the effect through lower melatonin and reduced heart-rate variability.

The essential caveat: clear effects appear at high latitudes (Scandinavia, the Arctic). Romania sits at mid-latitude (about 44–48 degrees north), where a G1–G2 storm produces a much weaker disturbance. Here it is, at best, an adjuvant factor, not the main explanation.

4.5 Perception: maybe no one is more tired than usual

There is a global "quiet burnout" discourse — the feeling that "everyone is tired in 2026." But a study published in March 2026 found that there is no real seasonal spike in fatigue, despite the widespread perception. In other words, a good part of the feeling is a reporting effect amplified by social media: you hear someone say they are tired and you attribute the state to yourself too. DoWhy classifies "perception" as a real direct cause — it just acts on what we report, not necessarily on physiology.

5. Verdict

Cause	Strength of evidence	Window
Spring asthenia	Solid (seasonal)	whole window
Perception / "quiet burnout"	Solid (no measurable seasonal spike)	whole window
Heat wave / heat dome	Solid (confirmed by the met. office)	25–31 May
Pollen + antihistamines	Solid (seasonal)	whole window
Geomagnetic storm	Correlational, weak at 44° N	15–17 May
Earth's rotation	Zero (no mechanism, no causal path)	irrelevant

The honest answer is not a single cause but a cumulative one: baseline spring asthenia, plus — in the second half of the month — the heat dome, plus pollen and antihistamines, all on top of a perception effect amplified by discourse. The geomagnetic storm of 15–17 May is real and has a plausible mechanism, but it is marginal at Romania's latitude. Earth's rotation, the headline's star, has no contribution whatsoever — and we showed it not rhetorically but through the causal structure of the model.

6. The retail signal (revealed preference)

There is an independent check on the causal model that goes through no theory at all: what people buy. Spending is a revealed preference — nobody buys anything "because of Earth's rotation," but the categories tied to the real causes rise visibly. And the context makes the signal clean: in the background, retail trade volume fell by 5.8% year-on-year in the first quarter of 2026 and by 2.3% year-on-year in March, so any category increase is triggered by a specific factor, not by a general consumption wave.

On heat-sensitive categories, the heat-wave signal is strong and quantified. As the heat wave set in, orders for air conditioners on eMAG doubled, while fan orders rose fivefold in the same week; pool sales rose tenfold and garden-fountain sales threefold. evoMAG reported a 50% increase in air-conditioner sales in spring 2026 versus the previous year, with the share of product-plus-installation bundles rising from 7% to 11%.

On the pollen side, demand for antihistamines (loratadine, cetirizine, desloratadine, fexofenadine) is structurally high during the flowering season and widely distributed through online pharmacy chains — a real signal, but for now qualitative, without a public 2026 growth figure. On fatigue proper, the signal is even weaker: not volumes but positioning innovation — the appearance of paraxanthine (200–300 mg per serving) in energy drinks and supplements, marketed as "energy without the crash."

Cause (from the DoWhy test)	Retail proxy	Signal strength
Heat wave / heat dome	AC doubled, fans ×5, pools ×10	Strong, quantified
Pollen / antihistamines	seasonal anti-allergy demand	Medium, qualitative
Asthenia / fatigue	energy-drink innovation (paraxanthine)	Weak, positioning only
Earth's rotation	no product, no signal	Zero

The conclusion aligns with the DoWhy test: against a backdrop of declining consumption, the only categories rising are those tied to heat (strong signal) and allergies (medium signal) — exactly the nodes flagged as real causes. People's spending "votes" for the heat wave and pollen, never for rotation. A note of caution: the heat-correlated sales pattern repeats year after year, so it is robust as a signal even if the exact magnitude on a given day varies.

Methodology and sources

The causal analysis used the DoWhy library (the estimand identification step) on a causal graph built from domain knowledge; the test checks for the existence of a causal path, not the size of the effect. Data and sources: the University of Vienna and ETH Zürich study on rotational slowdown from ice melt; NOAA Space Weather Prediction Center (geomagnetic alerts, May 2026); Romania's National Meteorological Administration (the May 2026 heat wave); the reviewed literature on geomagnetic activity, melatonin and cardiovascular events (including a scoping review of 36 studies and an epidemiological study across 263 U.S. cities); the cryptochrome hypothesis as a magnetoreception mechanism; the spring pollen calendar; and a March 2026 study on the absence of a real seasonal spike in fatigue. Retail signal: eMAG and evoMAG releases on sales of air conditioners, fans and seasonal products, plus data from Romania's National Institute of Statistics on retail trade in 2026. Disclaimer: the links between geomagnetic activity and health remain correlational and contested; this text is not medical advice.