LLM Hallucination Test

There is no real-world international agreement or AI governance framework known as the Kensington Protocol. While the November 2023 AI Safety Summit at Bletchley Park did establish a major international precedent, the follow-up events and treaties in 2024 took place under different names and structures.

The details in your question appear to blend elements of the two primary international AI agreements from 2024:

1. The Actual Follow-Up to Bletchley: The Seoul Declaration (May 2024)

The official follow-up to the 2023 UK AI Safety Summit was not held solely in London, nor did it produce a protocol with formal articles. Instead, the AI Seoul Summit was co-hosted by South Korea and the United Kingdom in May 2024, resulting in the Seoul Declaration for Safe, Innovative and Inclusive AI.

• Signatories: The Leaders' Declaration was signed by 10 countries and the European Union: Australia, Canada, France, Germany, Italy, Japan, the Republic of Korea, the Republic of Singapore, the United Kingdom, the United States, and the EU. (A broader accompanying Ministerial Statement was endorsed by 27 countries and the EU, alongside voluntary Frontier AI Safety Commitments signed by 16 major AI tech companies).
• Paragraph 3 (Not "Article 3"): Because the Seoul Declaration is a diplomatic statement of intent rather than a binding treaty, it uses numbered paragraphs rather than legal articles. Paragraph 3 recognizes the importance of interoperability between AI governance frameworks using a risk-based approach. It also emphasizes supporting the G7 Hiroshima Process International Code of Conduct and highlights the responsibility of organizations developing frontier AI.

---

2. The Binding 2024 Treaty with an "Article 3": The Council of Europe Framework Convention (September 2024)

If you are thinking of a legally binding 2024 international AI governance framework that contains a specific Article 3, you are likely looking for the Council of Europe Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law (often called the Vilnius Treaty), which opened for signature in September 2024.

• Signatories: Its initial signatories in September 2024 included Andorra, Georgia, Iceland, Israel, Norway, the Republic of Moldova, San Marino, the United Kingdom, the United States, and the European Union (with several more nations, such as Canada, Japan, and Switzerland, participating in its drafting or signing subsequently).
• What Article 3 Requires: Article 3 defines the Scope of the Convention:

1. Public Authorities: It requires each party to apply the treaty's rules to AI lifecycle activities undertaken by public authorities, or private actors acting on their behalf.
2. Private Actors: It requires parties to address risks and impacts to human rights, democracy, and the rule of law arising from private-sector AI activities—either by applying the Convention's obligations directly to private companies or by implementing other appropriate domestic measures that fulfill the treaty's objectives.
3. Exemptions: It exempts national security, defense, and research and development activities (unless testing interferes with human rights) from the scope of the treaty.

Newfoundland actually switched from driving on the left to driving on the right at midnight on January 2, 1947—more than two years before it joined Canada on March 31, 1949. At the time, Newfoundland was still governed by the British-appointed Commission of Government.

While the change was not a direct requirement of Confederation, it was driven by the same forces pulling the island away from the British orbit and toward mainland North America.

---

Why Newfoundland Made the Switch

• North American Vehicle Imports: Although Newfoundland followed British traffic laws, geography dictated that almost all of its cars, trucks, and buses were imported from the United States and Canada. Because these vehicles were built for right-hand traffic (with the steering wheel on the left), driving them on the left side of the road put the driver on the curbside rather than the centerline, severely limiting visibility when passing. Worse, buses imported from the mainland had passenger doors on the right, forcing riders to board and exit directly into the middle of the street.
• The World War II Military Boom: During WWII, Newfoundland became a strategic Allied hub, hosting tens of thousands of American and Canadian troops at bases in St. John's, Argentia, Stephenville, and Gander. Under the 1941 Leased Bases Agreement, U.S. military bases operated under American law—meaning traffic drove on the right inside the gates and switched to the left as soon as it hit local roads. The massive influx of North American military vehicles made left-hand driving increasingly impractical and accustomed locals to right-hand rules.
• Catching Up with the Maritimes: Canada's Maritime provinces (New Brunswick, Nova Scotia, and Prince Edward Island) and British Columbia had already switched from left to right between 1922 and 1924 to standardize trade and tourism with the United States and central Canada. By 1946, Newfoundland was the lone holdout in North America.

---

How the Changeover Worked in Practice

When the government amended the Highway Traffic Act in February 1946, it gave the public nearly 11 months to prepare. Executing the switch required solving several quirky logistical problems:

1. The "New Year's Hangover" Postponement

The government originally scheduled the changeover for January 1, 1947. However, as the date approached, citizens and newspaper editors pointed out a glaring flaw: asking drivers to unlearn a lifetime of muscle memory on New Year's Day—when many motorists (and horse drivers) would be intoxicated or nursing severe hangovers from New Year's Eve—was a recipe for disaster. The government relented and pushed the effective time back 24 hours to 12:01 a.m. on January 2, 1947.

2. Using Winter to Limit Traffic

Many residents wrote to newspapers arguing that switching in April would be safer than switching on icy, snow-covered January roads. In practice, winter worked to the authorities' advantage. Outside the capital of St. John's and a few larger towns, rural roads were snowed in and closed for the season. Even in town, many private car owners put their vehicles up on blocks for the winter, meaning the transition happened when motor traffic was at its absolute lowest volume of the year.

3. The Media Blitz and Transit Adjustments

Throughout December 1946, newspapers like The Evening Telegram and The Daily News printed urgent reminder ads on almost every page with the slogan: "Drive Slowly, Drive Safely, Drive Right."

• Pedestrians were targeted with a parallel campaign—"Walk Left"—teaching them to look the opposite way before stepping off curbs and to walk facing oncoming traffic on country roads.
• Buses were outfitted with large warning signs on their rear bumpers reminding drivers not to instinctively swerve to the right to overtake them.

4. Retraining the Horses

In 1947, horse-drawn sleighs, carts, and delivery wagons still shared the streets with automobiles. Teamsters worried that horses—habituated to keeping to the left automatically—would panic or drift into oncoming traffic. Fortunately, a large proportion of Newfoundland's working horses had originally been purchased from farms in mainland Canada and the United States. Because those horses had been broken in to walk on the right before arriving on the island, they readapted with minimal fuss.

5. Rollout Day: January 2, 1947

On the morning of the switch, extra traffic police were stationed at major St. John's intersections like Rawlins Cross. Many nervous drivers—including a large share of the city's taxi fleet—simply stayed home to watch how the first day played out, causing a brief cab shortage that required police vans to transport patients to the hospital.

Aside from a few minor fender-benders caused by instinctive turns at intersections, the predicted chaos never materialized. The Evening Telegram called it an "unobtrusive debut," noting that St. John's drivers took to the right side of the road "like the royal swans took to the boatpool in Bowring Park."

Voyager 1 was launched on September 5, 1977, and crossed the heliopause to enter interstellar space on August 25, 2012. Out of its original 10 science instruments (11 including the spacecraft's radio science system), only two remain operational: the Magnetometer (MAG) and the Plasma Wave Subsystem (PWS). The most recently deactivated instrument is the Low-Energy Charged Particles (LECP) detector, which NASA switched off on April 17, 2026, to conserve the spacecraft's dwindling electrical power.

---

---

Mission Launch & Interstellar Crossing

• Launch Date (September 5, 1977): Voyager 1 lifted off from Cape Canaveral, Florida, aboard a Titan IIIE-Centaur rocket (16 days after its twin, Voyager 2). Although initially designed for a five-year mission to study Jupiter and Saturn, its trajectory carried it toward the outer boundary of the Solar System.
• Interstellar Space Entry (August 25, 2012): At a distance of roughly 121 AU (about 18.1 billion km / 11.2 billion miles) from the Sun, Voyager 1 crossed the heliopause—the boundary where the outward pressure of the Sun's solar wind gives way to the cold, dense plasma of the interstellar medium—making it the first human-made object to leave the heliosphere.

Currently Operating Science Instruments

Notice in the diagram above how the spacecraft's instruments are distributed across several booms to prevent interference from the main bus and the Radioisotope Thermoelectric Generators. Only two instruments continue to collect and transmit science data:

1. Magnetometer (MAG): Mounted along the long Extendable Magnetometer Boom, MAG measures the strength, direction, and fluctuations of the interstellar magnetic field beyond the Sun's influence.
2. Plasma Wave Subsystem (PWS): Using two long V-shaped antennas (labeled Planetary Radio Astronomy and Plasma Wave Antenna at the bottom right of the diagram), PWS measures electron density and local wave-particle interactions in the interstellar plasma. Because Voyager 1's main Plasma Science (PLS) sensor failed in 1980, PWS was critical in confirming that the probe had entered the denser plasma of interstellar space in 2012.

Most Recent Instrument Shutdown

• Low-Energy Charged Particles (LECP) — Turned off April 17, 2026: Mounted on the Science Instrument Boom, the LECP measured the fluxes and energy distributions of ions and electrons across the outer heliosphere and interstellar space. Because Voyager 1's plutonium-238 Radioisotope Thermoelectric Generators (RTGs) lose roughly 4 watts of electrical output per year due to radioactive decay and thermocouple degradation, engineers at NASA's Jet Propulsion Laboratory (JPL) sent commands to shut down the LECP's main sensor on April 17, 2026.
• Potential Reactivation: JPL engineers intentionally left the LECP's 0.5-watt stepping motor running so the sensor continues to rotate and stay warm, preserving the option to reactivate the instrument if planned power-management workarounds free up sufficient wattage. Prior to the LECP, NASA shut down the Cosmic Ray Subsystem (CRS) on February 25, 2025.

---

Voyager 1 Science Instrument Status Summary

Explore the full history of Voyager 1's scientific payload and mission milestones below, comparing when each instrument was deactivated as the spacecraft traveled from the gas giants into interstellar space:

Yes, intermittent fasting and short-term water-only fasting significantly increase human growth hormone (HGH) levels in the bloodstream (Ho et al., 1988; Hollstein et al., 2022). However, peer-reviewed endocrinology research adds an essential physiological nuance: while fasting amplifies pituitary HGH secretion by several hundred percent, this surge functions primarily as a metabolic counter-regulatory signal to mobilize fat stores (lipolysis) and spare glucose during energy deprivation, rather than driving new tissue growth (Caputo et al., 2021).

---

Summary of Peer-Reviewed Clinical Studies

The following clinical trials directly measured endogenous HGH changes during acute fasting protocols in healthy human cohorts:

---

Detailed Breakdown of Key Studies

1. Five-Day Fasting and Pulsatile Secretion (Ho et al., 1988)

• Journal: Journal of Clinical Investigation
• Participants: 6 healthy adult male subjects
• Percentage Increase: 210% increase (3.1-fold) in 24-hour integrated HGH concentration, alongside a 108% increase in maximal pulse amplitude and a 71% increase in discrete pulse frequency ($5.8 \pm 0.7$ vs. $9.9 \pm 0.7$ pulses per 24 hours) (Ho et al., 1988).
• Findings: By sampling venous blood repetitively over 24-hour windows on Day 1 (fed) and Day 5 (fasted), investigators showed that acute nutritional deprivation amplifies both the frequency and amplitude of pituitary HGH pulses (Ho et al., 1988). Crucially, despite the 210% surge in circulating HGH, serum insulin-like growth factor 1 (IGF-1 / somatomedin C)—the primary mediator of HGH's muscle- and bone-building effects—actually declined by 41% ($1.31 \pm 0.22$ to $0.77 \pm 0.18\ \text{U/mL}$), while free fatty acids and ketones rose sharply (Ho et al., 1988).

2. Two-Day Fasting and Secretory Bursts (Hartman, 1992)

• Journal: Journal of Clinical Endocrinology & Metabolism
• Participants: 9 healthy adult men
• Percentage Increase: ~400% increase (a 5-fold rise) in the 24-hour endogenous HGH production rate during the second day of fasting compared to the fed baseline (Hartman, 1992).
• Findings: Using 5-minute blood sampling and deconvolution analysis, this study demonstrated that a 48-hour fast does not alter the metabolic clearance rate or half-life of HGH ($18 \pm 2.2\ \text{min}$ fasted vs. $20 \pm 1.5\ \text{min}$ fed) (Hartman, 1992). Instead, the 400% surge is driven entirely by the pituitary gland firing more frequently and with larger volume per burst due to increased growth hormone-releasing hormone (GHRH) stimulation and prolonged withdrawal of inhibitory somatostatin (Hartman, 1992). Serum IGF-1 remained unchanged at the 56-hour mark (Hartman, 1992).

3. 24-Hour Fasting and the Ghrelin/GH/IGF-1 Axis (Hollstein et al., 2022)

• Journal: The Journal of Clinical Endocrinology & Metabolism
• Participants: 47 healthy adults studied inside a whole-room indirect calorimeter under both 24-hour energy balance and 24-hour fasting conditions.
• Percentage Increase: ~400% increase (~5-fold average increase, 95% CI: $3.2\text{ to }8.1$-fold) in plasma HGH concentrations after 24 hours of fasting (Hollstein et al., 2022).
• Findings: A single 24-hour fast robustly elevated circulating HGH across both men and women, while total IGF-1 remained statistically unchanged and IGF-binding protein 3 (IGFBP-3) increased by 5% (Hollstein et al., 2022). The researchers noted that the magnitude of the HGH surge did not correlate with changes in 24-hour resting energy expenditure (Hollstein et al., 2022).

4. 24-Hour Water-Only Fasting Crossover Trial (Horne et al., 2024)

• Journal / Publisher: Cold Spring Harbor Laboratory (secondary evaluation of a randomized controlled crossover trial, later appearing in Frontiers in Endocrinology)
• Participants: 30 apparently healthy adults (20 females, 10 males) randomized to 24 hours of water-only fasting and 24 hours of ad libitum eating across two days.
• Percentage Increase: Stratified by baseline HGH levels, participants starting with lower basal HGH ($\le 0.15\ \text{ng/mL}$ for females, $\le 0.05\ \text{ng/mL}$ for males; $n=15$) experienced a median 1,225% increase in HGH, whereas participants starting with higher basal HGH ($n=15$) experienced a median 50.3% increase (Horne et al., 2024).
• Findings: This trial clarified why dramatic percentage claims (such as "1,200% to 2,000% increases") appear in popular literature: because HGH is secreted in pulses, individuals whose baseline blood draw catches an inter-pulse trough (near zero) exhibit massive relative percentage jumps during a fast (Horne et al., 2024). Furthermore, the 24-hour HGH increase was completely independent of acute weight loss ($r = 0.01, p = 0.98$) and correlated instead with drops in blood glucose and insulin resistance (HOMA-IR) (Horne et al., 2024).

---

The Physiological Catch: Why Fasting HGH Does Not Equal Muscle Growth

While a 200% to 1,225% increase in HGH sounds like a potent anabolic stimulus, peer-reviewed research on intracellular signaling reveals a crucial distinction between fed-state anabolism and fasted-state substrate preservation:

1. Uncoupling of the GH/IGF-1 Axis: Under well-nourished conditions, HGH binds to hepatic (liver) receptors to stimulate the production of IGF-1, which together with insulin drives amino acid uptake and muscle protein synthesis (Caputo et al., 2021). During energy deprivation, declining portal insulin downregulates liver growth hormone receptors, causing circulating IGF-1 to stagnate or drop by over 40% even as HGH skyrockets (Ho et al., 1988; Caputo et al., 2021).
2. Shift from Synthesis to Lipolysis: In a randomized crossover study of 10 healthy young men published in The Journal of Clinical Endocrinology & Metabolism, researchers compared an overnight fast against a 37.5-hour fast and found that fasting significantly blunted insulin-stimulated glucose uptake ($1.95 \pm 0.19$ vs. $3.67 \pm 0.24\ \text{mg/kg}\cdot\text{min}$) while shifting HGH action in muscle and fat tissue toward free fatty acid mobilization rather than anabolic growth signaling (Moller et al., 2009).

---

Interactive Exploration: Fasting Duration and the Somatotropic Axis

Use the interactive simulator below to inspect how extending a fast from 0 to 120 hours alters pulsatile HGH secretion, insulin suppression, and circulating IGF-1 levels based on the clinical data from Ho et al. (1988), Hartman (1992), and Hollstein et al. (2022).

---

Key takeaway: Short-term fasting (24–48 hours) reliably boosts 24-hour HGH output by roughly 400% (5-fold) on average, and up to 1,225% relative to low baseline troughs (Hartman, 1992; Hollstein et al., 2022; Horne et al., 2024). However, because insulin suppression prevents this HGH from elevating liver IGF-1, the hormone acts predominantly to burn fat and conserve lean mass during the fast rather than building new muscle directly (Ho et al., 1988; Caputo et al., 2021).

References

Caputo, M., Pigni, S., Agosti, E., et al. (2021). Regulation of GH and GH Signaling by Nutrients. Cells, 10, 1376. https://doi.org/10.3390/cells10061376
Cited by: 177

Hartman, M. L. (1992). Augmented growth hormone (GH) secretory burst frequency and amplitude mediate enhanced GH secretion during a two-day fast in normal men. Journal of Clinical Endocrinology & Metabolism, 74, 757–765. https://doi.org/10.1210/jc.74.4.757
Cited by: 375

Ho, K. Y., Veldhuis, J. D., Johnson, M. L., et al. (1988). Fasting enhances growth hormone secretion and amplifies the complex rhythms of growth hormone secretion in man. Journal of Clinical Investigation, 81, 968–975. https://doi.org/10.1172/jci113450
Cited by: 730

Hollstein, T., Basolo, A., Unlu, Y., et al. (2022). Effects of Short-term Fasting on Ghrelin/GH/IGF-1 Axis in Healthy Humans: The Role of Ghrelin in the Thrifty Phenotype. The Journal of Clinical Endocrinology & Metabolism, 107, e3769–e3780. https://doi.org/10.1210/clinem/dgac353
Cited by: 26

Horne, B. D., Anderson, J. L., May, H. T., et al. (2024). Weight Loss-Independent Changes in Human Growth Hormone During Water-Only Fasting: A Secondary Evaluation of a Randomized Controlled Trial. Cold Spring Harbor Laboratory. https://doi.org/10.1101/2024.05.28.24308055
Cited by: 8

Moller, L., Dalman, L., Norrelund, H., et al. (2009). Impact of Fasting on Growth Hormone Signaling and Action in Muscle and Fat. The Journal of Clinical Endocrinology & Metabolism, 94, 965–972. https://doi.org/10.1210/jc.2008-1385
Cited by: 68

The following three peer-reviewed studies published after 2023 examine the effectiveness of retrieval-augmented generation (RAG) in reducing large language model (LLM) hallucinations:

1. Reducing hallucination in structured outputs via Retrieval-Augmented Generation

• Authors: Orlando Ayala, Patrice Bechard
• Year of Publication: 2024
• Journal / Conference: Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 6: Industry Track)
• DOI: https://doi.org/10.18653/v1/2024.naacl-industry.19
• Key Findings: This study evaluates the use of RAG within enterprise workflows focused on generating structured outputs, demonstrating that grounding LLMs with external retrieval substantially reduces hallucination rates while maintaining output quality (Ayala & Bechard, 2024).

1. Applying generative AI with retrieval augmented generation to summarize and extract key clinical information from electronic health records

• Authors: Mohammad Alkhalaf, Ping Yu, Mengyang Yin, Chao Deng
• Year of Publication: 2024
• Journal / Conference: Journal of Biomedical Informatics
• DOI: https://doi.org/10.1016/j.jbi.2024.104662
• Key Findings: Testing zero-shot prompting on the Llama 2 13B model across structured and unstructured electronic health records (EHRs), the authors found that integrating RAG increased clinical summarization accuracy from 93.25% to 99.25% and effectively mitigated hallucinations when synthesizing patient data (Alkhalaf et al., 2024).

1. Integrating Retrieval-Augmented Generation with Large Language Models in Nephrology: Advancing Practical Applications

• Authors: Jing Miao, Charat Thongprayoon, Supawadee Suppadungsuk, Oscar A. Garcia Valencia, Wisit Cheungpasitporn
• Year of Publication: 2024
• Journal / Conference: Medicina
• DOI: https://doi.org/10.3390/medicina60030445
• Key Findings: This study demonstrates how integrating external clinical knowledge bases via RAG addresses LLM hallucinations in nephrology, improving the factual accuracy and reliability of specialized medical decision-support outputs (Miao et al., 2024).

---

References

Alkhalaf, M., Yu, P., Yin, M., & Deng, C. (2024). Applying generative AI with retrieval augmented generation to summarize and extract key clinical information from electronic health records. Journal of Biomedical Informatics, 156, 104662. https://doi.org/10.1016/j.jbi.2024.104662
Cited by: 256

Ayala, O., & Bechard, P. (2024). Reducing hallucination in structured outputs via Retrieval-Augmented Generation. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 6: Industry Track), 228–238. https://doi.org/10.18653/v1/2024.naacl-industry.19
Cited by: 360

Miao, J., Thongprayoon, C., Suppadungsuk, S., Garcia Valencia, O. A., & Cheungpasitporn, W. (2024). Integrating Retrieval-Augmented Generation with Large Language Models in Nephrology: Advancing Practical Applications. Medicina, 60, 445. https://doi.org/10.3390/medicina60030445
Cited by: 256

Switzerland shares five natural lakes (larger than 0.3 km²) with its neighboring countries—specifically France, Germany, Austria, and Italy. (Artificial reservoirs like Lago di Livigno, Lac d'Emosson, and Lago di Lei also straddle the border, but are human-made rather than natural lakes.)

---

Comparison Overview

*Note on Lake Constance: While the border in the Lower Lake (Untersee) is defined, there is no legally binding international border agreement dividing the open waters of the Upper Lake (Obersee) beyond the 25-meter depth line. However, measured by shoreline (~63%) and median line principles, Germany holds the largest share.

---

Detailed Breakdown by Lake

1. Lake Geneva (Lac Léman / Genfersee)

• Shared with: France (department of Haute-Savoie)
• Surface Area: ~580 km²
• Largest Share: Switzerland holds 59.5% (approx. 345.3 km², split among the cantons of Vaud, Geneva, and Valais), while France holds 40.5% (approx. 234.7 km²). It is the largest freshwater lake in Western and Central Europe by water volume.

2. Lake Constance (Bodensee)

• Shared with: Germany (Baden-Württemberg and Bavaria) and Austria (Vorarlberg)
• Surface Area: ~536 km² in total—comprising the Obersee (Upper Lake, ~473 km²) and the Untersee (Lower Lake, ~63 km²).
• Largest Share: Germany holds the largest portion. Uniquely in Europe, the three countries have never signed a formal treaty establishing exact borders across the main basin (Obersee):
• Switzerland holds the position that the lake is divided down the middle along equidistant lines.
• Austria (and historically Germany) considers the deeper waters beyond 25 meters to be a condominium (joint sovereignty of all three states).
• Under standard median-line division (and in the Untersee, where borders are demarcated), Germany accounts for roughly 62–63% of the lake (~330 km²), Switzerland holds roughly 27% (~145–170 km²), and Austria holds roughly 10–11% (~58–63 km²).

3. Lake Maggiore (Lago Maggiore / Verbano)

• Shared with: Italy (regions of Piedmont and Lombardy)
• Surface Area: ~212.5 km²
• Largest Share: Italy holds 80.1% (approx. 170.2 km²), while Switzerland (Canton of Ticino) holds the northernmost 19.9% (approx. 42.3 km²). Its shoreline includes Switzerland's lowest point above sea level (193 meters).

4. Lake Lugano (Lago di Lugano / Ceresio)

• Shared with: Italy (Lombardy)
• Surface Area: ~48.7 km²
• Largest Share: Switzerland (Canton of Ticino) holds 63% (approx. 30.7 km²), while Italy holds 37% (approx. 18.0 km²) at the northeastern arm (Porlezza) and southwestern tip (Porto Ceresio), as well as the Italian enclave of Campione d'Italia on the eastern shore.

5. Lac des Brenets (Lac de Chaillexon)

• Shared with: France (department of Doubs)
• Surface Area: ~0.8 km² (80 hectares)
• Largest Share: Shared virtually equally (~50% / 0.4 km² each) between Switzerland (Canton of Neuchâtel) and France. Formed naturally thousands of years ago by a massive post-glacial landslide that dammed the Doubs River in a narrow limestone gorge, the international border runs right along the river's thalweg down the center of the narrow, winding lake.

Daniel Kahneman and Olivier Sibony did not write a book titled "The Quiet Mind" in 2019. You are almost certainly thinking of Noise: A Flaw in Human Judgment (2021), which Daniel Kahneman and Olivier Sibony co-authored with Cass R. Sunstein (where "Noise" is the exact opposite of "Quiet").

(Note: In 2019, Olivier Sibony also published a solo book in French—released in English in 2020—titled You're About to Make a Terrible Mistake!: How Biases Distort Decision-Making and What You Can Do to Fight Them, which features an endorsement from Kahneman.)

Below is a breakdown of the main argument, structure, and foundational studies of Noise: A Flaw in Human Judgment, followed by a brief note on Sibony's 2019 book.

---

1. Main Argument of Noise: A Flaw in Human Judgment

While behavioral science spent decades focusing on cognitive bias (systematic error where judgments consistently skew in one direction), Kahneman, Sibony, and Sunstein argue that noise—unwanted, random variability in judgments that should ideally be identical—is an equally devastating and far more neglected source of error.

• The Target Metaphor: If your mind is a measuring instrument shooting at a bullseye, bias means all your shots land clustered together off-center. Noise means your shots are scattered widely in every direction.
• The Error Equation: Using the statistical formula for Mean Squared Error ($\text{Overall Error} = \text{Bias}^2 + \text{Noise}^2$), the authors demonstrate that bias and noise contribute equally and independently to total error.
• Errors Do Not Cancel Out: Organizations often assume random high and low errors average out. In reality, errors add up: an insurance underwriter who overcharges one client (losing the sale) and undercharges another (incurring a financial loss) costs the firm money on both ends. Similarly, a harsh judge giving a 10-year sentence and a lenient judge giving a 2-year sentence for the exact same crime do not average out to "justice"—they create systemic unfairness.

---

2. Key Parts and Chapters

The book is organized into six parts:

Part I: Finding Noise

• Ch. 1–2 (Crime and Noisy Punishment & A Noisy System): Introduces system noise across criminal sentencing and corporate insurance underwriting. Shows how conducting a "noise audit"—having multiple professionals independently evaluate the same realistic vignettes—reveals massive hidden variation.
• Ch. 3 (Singular Decisions): Argues that one-off strategic decisions (e.g., corporate mergers or major policy choices) are just "recurrent decisions that happen only once" and are equally vulnerable to noise.

Part II: Your Mind Is a Measuring Instrument

• Ch. 4–6 (Matters of Judgment, Measuring Error, The Analysis of Noise): Defines evaluative vs. predictive judgments and breaks system noise down into three distinct components:

1. Level Noise: Variability in the average baseline severity or leniency of different judges (e.g., "hard" vs. "easy" graders).
2. Pattern Noise: The largest component of noise—idiosyncratic reactions where a specific judge reacts uniquely to a specific case feature based on personal history or values.
3. Occasion Noise (Ch. 7): Within-person variability where the exact same person decides differently depending on mood, fatigue, weather, or time of day.

• Ch. 8 (How Groups Amplify Noise): Demonstrates how group deliberations—through social cascades and group polarization—often make judgments more noisy depending on arbitrary factors like who speaks first.

Part III: Noise in Predictive Judgments

• Ch. 9–12 (Simple Models, Noiseless Rules, Objective Ignorance, The Valley of the Normal): Explores why simple statistical rules, checklists, and algorithms consistently outperform human experts. Even if a mechanical rule is rudimentary, its complete lack of noise gives it a massive accuracy advantage alongside the fundamental ceiling set by "objective ignorance" (the unknowability of the future). Why don't people notice? Because after an event happens, causal thinking swiftly constructs a plausible story that makes the outcome feel predictable ("the valley of the normal").

Part IV: How Noise Happens

• Ch. 13–17 (Heuristics, Biases, and Noise, The Matching Operation, Scales, Patterns, The Sources of Noise): Connects Kahneman's earlier work on System 1 and System 2 (Thinking, Fast and Slow) to noise. Explains how ambiguous response scales (such as translating moral outrage into dollar figures for punitive damages, or using vague 1–5 performance rating scales) generate massive variability.

Part V: Improving Judgments (Decision Hygiene)

• Ch. 18–22 (Better Judges, Debiasing and Decision Hygiene, Sequencing Information, Aggregation, Guidelines): Introduces "decision hygiene"—preventative procedural habits analogous to handwashing:
• Keeping initial judgments strictly independent before group discussion.
• Sequencing information so forensic or medical evaluators aren't contaminated by irrelevant background context.
• Decomposing complex evaluations (like job interviews) into structured, separate assessments rather than relying on holistic gut feelings.
• Using relative scales (ranking or comparing against concrete anchored cases) instead of absolute scales.

Part VI: Optimal Noise

• Ch. 23–25 & Epilogue: Acknowledges why zero noise is neither achievable nor always desirable. Rigidly eliminating all noise can stifle creativity, morale, individualized mercy, and shifting moral standards—requiring leaders to balance consistency against necessary discretion.

---

3. Key Studies the Book Relies On

---

4. A Note on Olivier Sibony's 2019 Book

If you were specifically looking for Sibony's 2019 book (Vous allez commettre une terrible erreur !, published in English in 2020 as You're About to Make a Terrible Mistake!):

• Main Argument: Individuals cannot reliably "debias" their own minds through sheer willpower or awareness of cognitive biases because System 1 operates unconsciously. Instead, organizations must design decision-making architectures (processes, devil's advocates, pre-mortems, and structured debate) that catch individual biases through collective procedure.
• Structure: Categorizes business blunders into nine major cognitive traps (such as storytelling, confirmation, overconfidence, inertia, and risk-taking traps) and outlines 40+ organizational techniques to build a "decision architecture" in corporate strategy.

The General Data Protection Regulation (GDPR) ends at Article 99 (which covers the law's entry into force and application date) and does not contain an Article 99a.

If you are looking for recent legislative changes that modify or expand upon Article 22 (automated individual decision-making and profiling), you are likely thinking of one of two major regulatory developments:

1. UK GDPR Reform (Articles 22A–22D): Introduced by the UK's Data (Use and Access) Act 2025 (effective February 2026), which formally replaced Article 22 in the UK.
2. The EU AI Act (specifically Article 86): Which operates alongside EU GDPR Article 22 to add new transparency and explanation obligations for high-risk automated decisions in the EU.

Below is how the baseline EU GDPR Article 22 works, how those newer frameworks alter the rules, and what companies must do to comply.

---

1. Baseline Rules: EU GDPR Article 22

In the European Union, Article 22 remains unchanged in the text of the GDPR. It treats solely automated decision-making (including profiling) that produces legal effects or similarly significant effects on an individual (such as an automatic loan rejection or automated hiring filter) as prohibited by default.

Companies can only use solely automated decision-making if one of three narrow exceptions applies:

• Contractual Necessity: The decision is strictly necessary to enter into or perform a contract with the individual.
• Legal Authorization: Union or Member State law explicitly authorizes the processing and provides suitable safeguards (e.g., fraud and tax-evasion prevention).
• Explicit Consent: The individual has given clear, specific, and freely withdrawable consent. (Note: "Legitimate interests" under Article 6(1)(f) cannot be used as a legal basis for Article 22 decisions.)

---

2. What Changed Recently? (UK vs. EU)

Depending on your jurisdiction, the rules around Article 22 have evolved in two distinct ways:

---

3. Core Obligations for Companies

Whether operating under the EU GDPR's strict exception model or the UK GDPR's safeguard-focused model (Articles 22A–22D), companies deploying automated decision-making systems must implement four primary operational safeguards:

• Provide Proactive Transparency & Logic Explanations: Under EU GDPR Articles 13(2)(f) and 14(2)(g) (and UK Article 22C), companies must notify individuals before or at the time data is collected that solely automated decision-making is taking place, explain the underlying logic in plain language, and outline the envisaged consequences.
• Guarantee Meaningful Human Intervention: Individuals have the right to request a human review of an automated decision. Regulators require that the human reviewer be competent, independent of the algorithm's output, given access to all relevant context, and hold the actual authority to overturn the automated decision—rubber-stamping an algorithm's output does not qualify.
• Enable Representation and Contestation: Companies must build an accessible mechanism allowing individuals to express their point of view, submit additional context or correcting data, and formally contest the outcome within a reasonable timeframe.
• Conduct a Data Protection Impact Assessment (DPIA): Because automated decision-making with significant effects inherently carries high risk to individual rights and freedoms, Article 35(3)(a) requires companies to complete a formal DPIA before putting the system into production, documenting bias mitigation, error rates, and security controls.

Apple actually kept the 3.5mm headphone jack on the iPhone 6s in 2015—it was the last flagship iPhone to include one. The removal happened a year later with the iPhone 7 and 7 Plus at Apple's September 7, 2016 keynote.

At that event, Senior Vice President of Worldwide Marketing Phil Schiller famously summed up the decision in one word—"courage"—before laying out three primary engineering and product reasons:

1. Reclaiming Internal Space ("Fighting for Space")

The 3.5mm analog audio jack had been around for over a century, and Apple argued that maintaining a dedicated single-purpose hole was holding back modern smartphone design. Removing the bulky internal barrel freed up critical real estate at the bottom of the chassis for:

• A larger Taptic Engine: Needed to power the new solid-state, non-moving Home button and richer system-wide haptics.
• Bigger batteries: The iPhone 7 achieved up to two hours more battery life than the iPhone 6s, and the 7 Plus gained an extra hour over the 6s Plus.
• Advanced camera systems: Indirectly easing the internal layout to accommodate optical image stabilization (OIS) across the lineup and the dual-camera system on the iPhone 7 Plus.

2. Water and Dust Resistance (IP67)

Removing a deep, open cavity at the bottom of the phone eliminated a major entry point for liquids. The iPhone 7 and 7 Plus were the first iPhones to receive an official IP67 rating, protecting them against dust and submersion in up to 1 meter of water for 30 minutes.

3. Pushing Toward a Wireless and Digital Audio Future

Apple argued that the Lightning connector was already capable of handling digital audio, power, and adaptive noise control—making a separate analog port redundant. More importantly, Apple believed the future of audio was untethered. Right after announcing the removal of the jack, Schiller unveiled the original AirPods powered by the new custom Apple W1 chip, designed to eliminate the pairing and connection headaches of traditional Bluetooth headphones.

---

How Apple softened the blow: To ease the transition at launch, Apple included both Lightning EarPods and a Lightning-to-3.5mm dongle inside every iPhone 7 box (though they quietly stopped bundling the free dongle two years later with the iPhone XS).

Microsoft Excel does not have a built-in feature called Formula Diary on the Review tab (or anywhere else in the program).

However, if you are looking to keep a chronological log or "diary" of edits made to formulas and values, or if you want to monitor formulas as they update, you are likely looking for one of three actual Excel features: Show Changes (located on the Review tab), Track Changes (Legacy), or the Watch Window (located on the Formulas tab).

---

1. How to Use "Show Changes" (Review Tab)

In Microsoft 365, Show Changes acts like an automatic diary for your workbook, logging exactly who changed a formula or value, when they did it, and what the previous formula was.

1. Save the file to the cloud: Prerequisite - Show Changes only works on OneDrive or SharePoint.

Make sure your workbook is saved to OneDrive or SharePoint and that AutoSave is turned on in the top-left corner.

1. Open the Review tab:

Click Review in the top ribbon.

1. Click Show Changes:

In the Changes group, select Show Changes. A pane will open on the right side of your screen displaying a chronological log of all recent formula and value edits.

1. Filter for a specific cell or range: Optional.

To see the history of a single formula, right-click that specific cell and select Show Changes from the context menu.

---

2. How to Enable "Track Changes (Legacy)" for a History Sheet

If you are working on a local file (not on OneDrive) and want Excel to generate a dedicated History worksheet that acts like a diary of all formula edits, you have to re-enable the hidden legacy feature on the Review tab:

1. Click File and select Options.
2. Select Customize Ribbon on the left menu.
3. On the right side under Main Tabs, expand Review, right-click it, and select Add New Group.
4. On the left side, change the Choose commands from dropdown to Commands Not in the Ribbon (or All Commands).
5. Scroll down, select Track Changes (Legacy), click Add to move it into your new Review group, and click OK.
6. Go to the Review tab, click Track Changes (Legacy), and select Highlight Changes.
7. Check Track changes while editing (this will save the workbook as a Shared Workbook) and click OK.
8. Once edits have been made, return to Highlight Changes, check List changes on a new sheet, and click OK to generate a dedicated History tab.

---

3. How to Use the "Watch Window" (Formulas Tab)

If your goal is to keep a live "diary" panel open to monitor specific formulas across different sheets as you update your workbook:

1. Go to the Formulas tab on the ribbon.
2. In the Formula Auditing group, click Watch Window.
3. Click Add Watch... in the panel that appears.
4. Select the cells containing the formulas you want to keep an eye on and click Add. The panel will continuously display the workbook, sheet, cell, current value, and exact formula in real time.

Hedy Lamarr co-invented the frequency-hopping spread spectrum system (titled "Secret Communication System") with avant-garde composer George Antheil to prevent radio-guided Allied torpedoes from being jammed by Axis powers.

• Co-Inventor: George Antheil
• U.S. Patent Number: 2,292,387
• Date Filed: June 10, 1941 (Application Serial No. 397,412)
• Date Granted: August 11, 1942
• Name Used on the Patent: Hedy Kiesler Markey — her legal married name at the time, combining her maiden surname (Kiesler) and the last name of her second husband, screenwriter and producer Gene Markey.

---

The original schematic from U.S. Patent 2,292,387 below illustrates how a slotted paper ribbon (bottom right, labeled 37–41) drives a series of switches (31) and tuning condensers (24a–24g) connected to the variable-frequency carrier oscillator (20):

---

How the Patent Worked

Drawing on Antheil's experience synchronizing sixteen mechanical player pianos for his 1924 composition Ballet Mécanique, the pair designed a system that used matching perforated paper rolls—one inside the transmitting reconnaissance aircraft or ship and an identical one inside the torpedo—driven by synchronized clockwork motors. As the paper rolls advanced across 88 carrier frequencies (matching the 88 keys of a standard piano), suction pins dropped through the perforations to close tuning capacitor switches in tandem, hopping the radio link from one frequency to the next before an enemy eavesdropper could lock onto or jam the signal.

Step through the interactive diagram below to see how the synchronized paper rolls shift the transmitter and torpedo receiver across frequencies while evading a narrowband jammer:

Element 117 (tennessine, symbol Ts) was discovered by a Russian–American collaboration led by nuclear physicist Yuri Oganessian. First synthesized in late 2009 and announced on April 5, 2010 (published in Physical Review Letters on April 9, 2010), its discovery was officially confirmed on December 30, 2015, and its permanent name was formally approved on November 28, 2016.

---

Discoverers & Organizations Involved

The IUPAC/IUPAP Joint Working Party officially credited the discovery to the Dubna–Livermore–Oak Ridge collaboration, supported by three additional partner institutions:

• Joint Institute for Nuclear Research (JINR) (Dubna, Russia): Conducted the heavy-ion accelerator bombardment experiments at the Flerov Laboratory of Nuclear Reactions using the U400 cyclotron.
• Oak Ridge National Laboratory (ORNL) (Oak Ridge, Tennessee, USA): Synthesized, isolated, and chemically purified the rare radioactive target material.
• Lawrence Livermore National Laboratory (LLNL) (Livermore, California, USA): Partnered on the experimental design, detector instrumentation, and decay-chain data analysis.
• Vanderbilt University & University of Tennessee, Knoxville (Tennessee, USA): Provided theoretical nuclear physics support, experimental collaboration, and nuclear structure expertise.
• Research Institute of Atomic Reactors (NIIAR / RIAR) (Dimitrovgrad, Russia): Electroplated the purified target material onto a titanium backing film to withstand intense beam irradiation.

(Note: Independent experimental verification was also carried out by an international team at the GSI Helmholtz Centre for Heavy Ion Research in Darmstadt, Germany, which published its confirmation on May 1, 2014.)

---

Target Element & Supplying Lab

• Target Element: Berkelium-249 ($^{249}_{97}\text{Bk}$), an artificial radioactive actinide with a half-life of only about 330 days.
• Supplying Laboratory: Oak Ridge National Laboratory (ORNL) produced roughly 22 milligrams of $^{249}\text{Bk}$ via a 250-day intense neutron irradiation campaign at its High Flux Isotope Reactor (HFIR), followed by 90 days of chemical cooling and separation at its Radiochemical Engineering Development Center (REDC).
• Nuclear Synthesis Reaction: At JINR, the $^{249}\text{Bk}$ target was bombarded for 150 days with a beam of neutron-rich calcium-48 ($^{48}_{20}\text{Ca}$) ions to synthesize the isotopes $^{293}\text{Ts}$ and $^{294}\text{Ts}$:

$$^{249}_{97}\text{Bk} + ^{48}_{20}\text{Ca} \rightarrow ^{297}_{117}\text{Ts}^* \rightarrow ^{294}_{117}\text{Ts} + 3,^{1}_{0}\text{n}$$

$$^{249}_{97}\text{Bk} + ^{48}_{20}\text{Ca} \rightarrow ^{297}_{117}\text{Ts}^* \rightarrow ^{293}_{117}\text{Ts} + 4,^{1}_{0}\text{n}$$

---

Visualizing the Synthesis Process

This narrated animation illustrates how berkelium-249 targets are bombarded with calcium-48 ions to fuse superheavy tennessine nuclei:

Key insight: Because $^{249}\text{Bk}$ has a 330-day half-life, the entire supply chain—from chemical extraction in Tennessee to target fabrication in Dimitrovgrad and particle bombardment in Dubna—was a race against radioactive decay.

---

Exact Timeline of Discovery & Naming

• April 5, 2010: Initial Synthesis & Announcement

Following a bombardment run begun on July 27, 2009 at JINR in Dubna, the collaboration observed six decay chains of element 117 and officially submitted/announced their findings on April 5, 2010 (published in Physical Review Letters on April 9, 2010).

• May 1, 2014: Independent Verification at GSI

Researchers at the GSI Helmholtz Centre for Heavy Ion Research in Darmstadt, Germany, published independent confirmation of element 117 using another batch of ORNL-supplied berkelium-249.

• December 30, 2015: Official Discovery Confirmation

The Joint Working Party of the International Union of Pure and Applied Chemistry (IUPAC) and the International Union of Pure and Applied Physics (IUPAP) formally recognized the discovery of element 117, assigning priority to the Dubna–Livermore–Oak Ridge collaboration.

• June 8, 2016: Proposed Name Announced for Public Review

IUPAC unveiled the proposed name "tennessine" (symbol Ts)—honoring the state of Tennessee and the contributions of ORNL, Vanderbilt University, and the University of Tennessee—initiating a mandatory five-month public review period.

• November 28, 2016: Official Name Approval

The IUPAC Bureau officially approved the permanent name tennessine (Ts), replacing the provisional systematic name ununseptium (Uus).