Concept · Signal Over Noise
Signal Over Noise
Noise asks how you feel. Signal tells you what to do next.
Most of what reaches you today won’t change a single thing you do. Here is the physics of that, the data, and a five-question test you can run on anything.
SIGNAL: an input that changes your next action. NOISE: every other input, however true, loud or personal.
- 186phone checks a day, U.S., Q4 2025Reviews.org 2026
- 2 minone ping at work, every two minutesMicrosoft 2025
- 47 son one screen before switchingMark 2023
01Shannon · 1948
The physics of signal
In 1948 Claude Shannon worked out how much truth fits through a noisy channel. Your head is a noisy channel.
Shannon’s law: C = B · log₂(1 + S/N). How much passes through a channel without error depends on two things only: bandwidth B and the ratio of signal to noise. Not volume. Not effort.
Your bandwidth is fixed: one working memory, about 16 waking hours. You can’t buy more B. The only lever left is the ratio.
When noise dominates, the formula is generous. Capacity grows almost in proportion to S/N, so halving the noise roughly doubles what gets through. Push more than capacity and errors become unavoidable.
Noise dominates: halving the noise roughly doubles what gets through.
The three labels are illustrations, not measurements. Shannon, 1948: “semantic aspects … are irrelevant to the engineering problem.” We use the math as a model: signal is any input that changes your next action.
S/N 0.1 → 0.14 bits per hertz. S/N 10 → 3.46. Same bandwidth, 25× the throughput.
Don’t add inputs. Remove one source of noise today and watch the ratio, not the volume.
02Taleb · 2001
More data is not more truth
Look more often and you see more noise, not more truth. Taleb did the arithmetic in 2001.
Taleb’s dentist in Fooled by Randomness runs a strategy with a 15% expected return and 10% volatility. Checked yearly, it shows a gain 93% of the time. Daily: 54%. Hourly: 51.3%. Every second: 50.02%. A coin flip with a story attached.
The strategy never changed. Only the sampling rate did.
Looking more often also teaches you nothing new about the trend. Merton proved it in 1980: the precision of an estimated drift depends only on how long you watched, not how often you looked. Frequency sharpens your picture of the noise, never of the signal.
You pay for every look in emotion. A loss hurts about 2.25× more than an equal gain pleases.
- Noise per unit of signal
- 10.6
- Pleasant looks per year
- 136
- Painful looks per year
- 117
- Felt balance per year our model, λ = 2.25: gains +1, losses −2.25
- −127
| 1 year | 93% |
|---|---|
| 1 quarter | 77% |
| 1 month | 67% |
| 1 day | 54% |
| 1 hour | 51.3% |
| 1 minute | 50.17% |
| 1 second | 50.02% |
Not in the book: computed with the same formula.
What you know about the trend after one year: identical at every setting (Merton 1980). Merton 1980
Our model: with λ = 2.25 the felt balance turns red when you look more often than about every 6 weeks.
P = Φ((μ/σ)·√Δt) · μ = 15%, σ = 10% · Δt in trading years (252 days × 8 h) Taleb 2001
Investors with the most frequent feedback “took the least risk and earned the least money.”
Give every metric a check interval: Δt ≥ (σ/μ)² (our model). One sale every 10 days? Look every 10 days, not every hour.
03Kahneman · Sibony · Sunstein · 2021
The noise inside your head
Same file, same facts, different answer. The spread is you.
Kahneman, Sibony and Sunstein ran a noise audit at an insurer. Executives expected two underwriters pricing the same file to differ by about 10%. The measured median gap was 55%. Claims adjusters: 43%. 828 senior executives at other firms guessed 10% too.
208 U.S. federal judges sentenced the same 16 hypothetical cases. Mean sentence: 7 years. Two judges picked at random: 3.8 years apart. One asylum judge granted 5% of applications, another 88%.
Noise lives inside one head as well. Wine judges re-tasting the same wines gave identical scores to only 18% of them.
Total error = bias² + noise². You can’t feel your own noise. You can only measure it.
Example pair from the book: $9,500 vs $10,500 expected; $9,500 vs $16,700 measured. Kahneman et al. 2021
- 43%claims adjusters
- 828executives who also guessed 10%
- ×0.5noise after averaging 4 independent estimates
Gap = |A − B| ÷ average of A and B, the book’s noise index.
Expected spread: 10%. Measured: 55%.
Write your estimate before you hear anyone else’s. Average four independent estimates and the noise halves (1/√4).
04Simon · Mark · 1971–2023
The attention economy
Information isn’t scarce. You are.
Herbert Simon, 1971: “a wealth of information creates a poverty of attention.” Gloria Mark measured the bill. Time on one screen before switching: 150 seconds in 2004, 75 around 2012, 47 by 2016–2021. You start about half of those switches yourself.
Each switch leaves attention residue, part of you still on the last task (Leroy, 2009). A notification you never open hurts focus about as much as using the phone (Stothart, 2015).
Now: 186 phone checks a day in the U.S., 84.6% within ten minutes of waking. At work, Microsoft logs a ping every two minutes.
It’s reversible. Block mobile internet for two weeks: 91% improved on at least one measure.
- 186phone checks a dayU.S., ~1,000 adults, Q4 2025 · self-report from screen-time reportsReviews.org 2026
- 84.6%check within 10 minutes of wakingU.S., ~1,000 adults, Q4 2025 · self-reportReviews.org 2026
- 42%sometimes or often avoid the news97,520 people, 48 markets, 2026 · surveyReuters Institute 2026
- 91%improved after 2 weeks without mobile internetn = 467, randomized, 2025 · objectively trackedCastelo et al. 2025
47 seconds on one screen, median 40.
Notifications off except people. Messages in three windows a day. No phone in the first hour.
05Barrett · Lieberman · Hart
Emotions are data, not orders
A feeling is a measurement of you. It is not a measurement of the world.
On a leading account, your brain builds emotions from body signals and past experience (Barrett, 2017). They carry real data about your sleep, your stakes and your history. Read them. Don’t take orders from them.
Filter one: name it. Putting a feeling into words lowers the amygdala response (Lieberman et al., 2007).
Filter two: scripts. Much of what feels like your opinion was installed by family, school, culture and the room. Under social influence, success turns unpredictable: the best songs rarely flopped, the worst rarely won, anything else was possible (Salganik et al., 2006).
Filter three: confirmation. People pick agreeing information about 1.9× more often (Hart et al., 2009). Tie the belief to a live goal and the bias flips.
Rule: if a card has a vowel on one side, it has an even number on the other. Which cards must you turn over to test it?
Tap the cards you would turn.
We test ideas by looking for agreement. Signal comes from looking for the card that could break it.
Congenial information chosen 1.92× more often across 91 studies and about 8,000 people. When the challenging fact served a current goal, people preferred it.
One line before you act: “Anxious, 7/10. The data says ___. The next action is ___.”
06Five gates
The Signal Test
Five yes/no questions, thirty seconds. Works on a headline, an email, a feeling or a billionaire.
Truth is not the bar. Relevance is. A fact can be correct, important to someone, and still noise for you this week. Run any input through five gates. The first “no” settles most cases.
People choose objectively worse tasks when those tasks merely look urgent (5 experiments).
Tomorrow, before you reply to anything, run your first five inputs through the test.
07Goldratt · 1984
The bottleneck is the signal
A system moves at the speed of its narrowest point. Information about anything else is noise this week.
Goldratt, The Goal (1984): the constraint sets the output of the whole system.
He defined throughput as the rate at which the system generates money through sales. Not activity. Not attention. Money.
That makes the filter structural. A tool, tip or trend that doesn’t touch your constraint can be true and still be noise this week. 50 conversations and 3 payments? Your constraint is the offer or the close. It isn’t traffic, a new logo or a better CRM.
When the constraint moves, your signal moves with it.
“An hour lost at a bottleneck is an hour lost for the total system. An hour saved at a non-bottleneck is a mirage.”
Goldratt 1984
Illustrative numbers. Model: output = min(capacity of each stage). The five focusing steps, in the Business Academy (module 02) →
“An hour lost at a bottleneck is an hour lost for the total system. An hour saved at a non-bottleneck is a mirage.”
Write your constraint in one line. Put its number where you’ll see it before your phone.
08Parkinson · Ariely · Boyd
The Equalizer Clock
Start the clock before you act. A block with an end is a signal. A task without one is noise.
In The Equalizer (2014), Robert McCall reads the room, says “Sixteen seconds,” starts his watch and moves. Afterwards he checks it. Estimate, act, review.
Run your day the same way. Every block gets a fixed time: morning routine 45 minutes, one task 60, a deep block 120. When the timer rings, you switch. Hanging on one task grows noise and shrinks signal. Parkinson saw it in 1955: “work expands so as to fill the time available for its completion.”
Inside the block: one action, full focus. After it: three lines, then the next block.
Don’t polish forever. Version one is rough. Version twenty is excellent. Build, fix and improve faster than entropy breaks things.
The 3-line retro Under a minute. “Next” names the next block.
The 45 / 60 / 120 protocol
- 45 minMorning routine
Body, light, the 10-minute Signal OS ritual. Start at the same time every day.
- 60 minOne task
One deliverable with a visible end: an offer sent, a page shipped, a call prepared.
- 120 minDeep block
Work on the constraint. Phone in another room, notifications off.
- 25 minCold start
Francesco Cirillo’s Pomodoro® unit, for the task you’re avoiding or for small chores: 25 on, 5 off, a longer break after four.
- 5 minBetween blocks
The retro, stand up, water, then the next block. Don’t put two deep blocks back to back without a break.
The parked rule
Blocked by time, by another person or by anything outside your control? Don’t stare at it. Park it in one minute: where you stopped, what’s blocking, and the next check time (date and hour). Then switch. A one-minute “ready-to-resume” plan cut attention residue and protected performance on the next task (Leroy & Glomb, 2018).
No blocks yet. Start the clock.
Stored in this browser only, by date. Nothing is sent.
Why it works
- Parkinson, 1955
Open time fills itself. A fixed block stops the spread.
Parkinson 1955 - Ariely & Wertenbroch, 2002
Evenly spaced deadlines beat one end deadline on errors found, delays and earnings. Self-set deadlines landed in between.
Ariely & Wertenbroch 2002 - Cirillo, 1987
He started with two minutes on a tomato-shaped kitchen timer and settled on 25-minute units with short breaks.
Cirillo 1987 - Bayles & Orland, 1993
In their pottery parable, the class graded on quantity made the best pots. It’s a parable, not a study (Orland says it’s based on Jerry Uelsmann’s photography class), so use it as an illustration.
Bayles & Orland 1993 - Boyd, 1986
“…we should operate at a faster tempo or rhythm than our adversaries.” Each block is one observe–orient–decide–act loop. Boyd meant relative tempo, not frenzy.
Boyd 1986
Evenly spaced deadlines beat a single end deadline on every measure: errors found, delays, earnings.
Before you open the task, set the timer: 45, 60 or 120. When it rings, write three lines and switch.
The Equalizer (2014, dir. Antoine Fuqua) is a film reference only. Pomodoro® is a registered trademark of Francesco Cirillo; our timer is the Equalizer Clock. The Equalizer 2014 Cirillo 1987
09Routine, not an app
Signal OS
Ten minutes a day, thirty a week, seven rules. Nothing to install.
Signal OS is a routine, not an app. Ten minutes every morning before any inbox. Thirty minutes once a week to measure your noise. Seven rules that run in the background, so you stop deciding the same things again.
It’s assembled from tested parts. If-then plans have a medium-to-large effect on reaching goals (d = 0.65 across 94 tests). Email in three batches lowers daily stress. Written decisions expose your noise, because you can compare what you expected with what happened.
- 0:00–2:00The number
Write yesterday’s one metric: money in, or the count at your constraint.
- 2:00–4:00The constraint
One line. Same as yesterday? If it moved, write where.
- 4:00–7:00The next action
One move on the constraint, written as if-then with a time: “If it’s 9:00, then I send three offers.”
- 7:00–9:00The test
Run the Signal Test on the top three inputs waiting for you.
- 9:00–10:00One cut
Mute, unsubscribe or delete one source of noise.
Email limited to three checks a day → lower daily stress (n = 124, d = 0.37).
Tomorrow, before the first message: pen, one number, one constraint, one action.
Seven hard rules
- 01News diet
Once a day, one window, primary sources first. Or once a week.
- 02One metric
The constraint’s number, where you see it first. If it becomes a target, check it isn’t being gamed (Goodhart).
- 03Decisions in writing
Before you hear the room: the decision, the expected result, your confidence %, and what would change your mind.
- 04Notifications off
Except people who pay you and people you love.
- 05Messages in three batches
Morning, midday, end of day.
- 06Opinions need skin
An opinion without skin in the game is noise until it survives a test.
- 07Check at the signal horizon
Each number at its own interval, not at the speed of your anxiety.
Plain text, ready for your notes.
The weekly noise audit 30 minutes, same day every week
- Pull data, not memories: pickups, notifications and top 3 apps from your phone’s screen-time report.
- Count the inputs that changed an action this week and divide by all the inputs you handled. That’s your weekly signal-to-noise ratio.
- Open your decision journal. For each decision, compare the expected result with the actual one. Where were you noisy?
- Self-audit: re-estimate three numbers you estimated last week, without looking. The gap is your occasion noise.
- Check the constraint. If it moved, next week’s filter moves too.
- Cut three sources. Add none.
10Kaplan & Norton · Webb & Sheeran
Signals in business
Talk is cheap data. Behaviour is expensive data. Money is the final signal.
Lagging indicators tell you what already happened: revenue, churn, profit. Leading indicators tell you what’s coming: calls booked, proposals opened, deposits paid. Watch both. Act on the leading ones (Kaplan & Norton, 1996).
Then rank evidence by what it costs the person giving it. A like costs nothing. A booked call costs an hour. A deposit costs money. Across 47 experiments, shifting intentions by d = 0.66 moved behaviour by only d = 0.36 (Webb & Sheeran, 2006). Talk moves almost twice as easily as action.
The last word belongs to money in the account. A payment is the only answer nobody gives to be polite.
- Like0
- Commentseconds
- Reply with a questionminutes
- Call bookedan hour
- Invoice requestedcommitment
- Depositmoney
- Paid in fullmoney
- Repeat or referralmoney + reputation
Weights are our own heuristic (a model), not a published scale.
Lagging
- revenue
- profit
- churn
- LTV
Leading
- conversations booked
- proposals opened
- deposits
- trials
Same experiments, two shifts Webb & Sheeran 2006
- Intention changed = 0.66
- Behaviour changed = 0.36
Intention d = 0.66 → behaviour d = 0.36, across 47 experiments.
Swap “Would you buy this?” for a payment link. Count deposits, not compliments.
11Signal Over Noise
Your next action is the signal
Feelings are data. Opinions are hypotheses. Headlines are weather. Money is the verdict. Your next action is the signal.
The world won’t send you less. You can only raise the ratio: fewer inputs, better sources, slower checks, blocks with an end, decisions in writing, one constraint, one number. Start with the test.
The same system for founders, with an exercise, a checklist and a 7-day protocol. For Academy members.
Codex · 01Cheat Codes30 levers for non-linear output.
Codex · 02Red Pills17 reality checks on incentives, frames and scripts.
From 50.02% to 93%: the same strategy, checked less often.
Run the Signal Test on the next thing that pulls at you.
Questions
Signal, briefly
What does “signal over noise” mean?
Signal is any input that changes your next action. Noise is everything else, however true or loud. The terms come from engineering: Shannon showed that how much information passes through a channel depends on the ratio of signal to noise, not on volume.
How do I tell signal from noise quickly?
Run the Signal Test: does it change my next action, does it touch my constraint, is it verified at the source, will it matter in a year, can I start in ten minutes? If it doesn’t change your next action, it’s noise for now.
Is checking my numbers often a bad idea?
It’s noisier, not neutral. In Taleb’s example, a strategy that gains in 93% of years gains in only 50.02% of seconds. Merton showed that frequent checks don’t improve your estimate of the trend. Check at the signal horizon, Δt ≥ (σ/μ)² (our model).
Are emotions noise?
No. They’re data about your state (sleep, stakes, history), not orders about the world. Name the feeling, rate it from 0 to 10, then run the decision through the Signal Test.
What is a noise audit?
You give the same case to several people, or to yourself twice, and measure the spread. At one insurer, executives expected a 10% difference between underwriters; the measured median was 55% (Kahneman, Sibony & Sunstein, Noise, 2021).
How many times a day do people check their phones?
A Q4 2025 Reviews.org survey of about 1,000 U.S. adults, read from their screen-time reports, found 186 checks a day, down from 205 a year earlier. It’s self-reported, so treat it as an order of magnitude.
Is the “8-second attention span, shorter than a goldfish” real?
No. It traces to an unsourced website cited in a 2015 Microsoft Canada report, and BBC and WSJ checks found no study behind it. The measured figure is different: about 47 seconds on one screen before switching (Gloria Mark).
What is the strongest signal in business?
Behaviour that costs the customer something: a booked call, a deposit, a payment, a repeat purchase. In experiments, intentions move almost twice as easily as behaviour (Webb & Sheeran, 2006). Money in the account is the final signal.
Every number has a source
Sources
Numbers marked “model” are our arithmetic on these inputs. Shannon measured symbols, not meaning: here his math is a model.
- Reviews.org 2026Trevor Wheelwright / Reviews.org (2026). Cell Phone Usage Stats 2026. Reviews.org survey, ~1,000 U.S. adults, Pollfish, Q4 2025, weighted, ±4% open ↗
- Microsoft 2025Microsoft WorkLab (2025). Breaking down the infinite workday (Work Trend Index special report). Microsoft 365 telemetry to 15 Feb 2025 (excl. EU and Edu tenants) + survey of 31,000 knowledge workers in 31 markets open ↗
- Mark 2023Gloria Mark (2023). Attention Span: A Groundbreaking Way to Restore Balance, Happiness and Productivity. Hanover Square Press; APA Speaking of Psychology ep. 225 (Feb 2023); UC Newsroom (Jan 2024) open ↗
- Shannon 1948Claude E. Shannon (1948). A Mathematical Theory of Communication. Bell System Technical Journal 27: 379–423, 623–656 open ↗
- Shannon 1949Claude E. Shannon (1949). Communication in the Presence of Noise. Proceedings of the IRE 37(1): 10–21 open ↗
- Hartley 1928R. V. L. Hartley (1928). Transmission of Information. Bell System Technical Journal 7(3): 535–563 open ↗
- Taleb 2001Nassim Nicholas Taleb (2001). Fooled by Randomness, ch. 3 'A Mathematical Meditation on History', Table 3.1. Texere 2001; 2nd ed. Random House 2004 open ↗
- Merton 1980Robert C. Merton (1980). On Estimating the Expected Return on the Market: An Exploratory Investigation. Journal of Financial Economics 8(4): 323–361 (Appendix) open ↗
- Tversky & Kahneman 1992Amos Tversky, Daniel Kahneman (1992). Advances in Prospect Theory: Cumulative Representation of Uncertainty. Journal of Risk and Uncertainty 5: 297–323 open ↗
- Thaler et al. 1997Richard H. Thaler, Amos Tversky, Daniel Kahneman, Alan Schwartz (1997). The Effect of Myopia and Loss Aversion on Risk Taking: An Experimental Test. Quarterly Journal of Economics 112(2): 647–661 open ↗
- Kahneman et al. 2021Daniel Kahneman, Olivier Sibony, Cass R. Sunstein (2021). Noise: A Flaw in Human Judgment. Little, Brown Spark open ↗
- Clancy et al. 1981Kevin Clancy, John Bartolomeo, David Richardson, Charles Wellford (1981). Sentence Decisionmaking: The Logic of Sentence Decisions and the Extent and Sources of Sentence Disparity. Journal of Criminal Law and Criminology 72(2): 524–554
- Ramji-Nogales et al. 2007Jaya Ramji-Nogales, Andrew I. Schoenholtz, Philip G. Schrag (2007). Refugee Roulette: Disparities in Asylum Adjudication. Stanford Law Review 60(2): 295–411 open ↗
- Reuters Institute 2026Reuters Institute for the Study of Journalism (Egan, Robertson, Ross Arguedas, Newman, Nielsen, Mukherjee, Fletcher) (2026). Digital News Report 2026. YouGov online survey, mid-Jan to end-Feb 2026, 97,520 respondents, 48 markets open ↗
- Castelo et al. 2025Noah Castelo, Kostadin Kushlev, Adrian F. Ward, Michael Esterman, Peter B. Reiner (2025). Blocking mobile internet on smartphones improves sustained attention, mental health, and subjective well-being. PNAS Nexus 4(2): pgaf017 open ↗
- Simon 1971Herbert A. Simon (1971). Designing Organizations for an Information-Rich World. in M. Greenberger (ed.), Computers, Communications, and the Public Interest, Johns Hopkins Press, pp. 37–72 (quote pp. 40–41) open ↗
- Leroy 2009Sophie Leroy (2009). Why is it so hard to do my work? The challenge of attention residue when switching between work tasks. Organizational Behavior and Human Decision Processes 109(2): 168–181 open ↗
- Stothart et al. 2015Cary Stothart, Ainsley Mitchum, Courtney Yehnert (2015). The attentional cost of receiving a cell phone notification. Journal of Experimental Psychology: Human Perception and Performance 41(4): 893–897 open ↗
- Barrett 2017Lisa Feldman Barrett (2017). How Emotions Are Made: The Secret Life of the Brain. Houghton Mifflin Harcourt open ↗
- Lieberman et al. 2007Matthew D. Lieberman et al. (2007). Putting Feelings Into Words: Affect Labeling Disrupts Amygdala Activity in Response to Affective Stimuli. Psychological Science 18(5): 421–428 open ↗
- Salganik et al. 2006Matthew J. Salganik, Peter Sheridan Dodds, Duncan J. Watts (2006). Experimental Study of Inequality and Unpredictability in an Artificial Cultural Market. Science 311(5762): 854–856 open ↗
- Hart et al. 2009William Hart, Dolores Albarracín, Alice H. Eagly, Inge Brechan, Matthew J. Lindberg, Lisa Merrill (2009). Feeling Validated Versus Being Correct: A Meta-Analysis of Selective Exposure to Information. Psychological Bulletin 135(4): 555–588 open ↗
- Wason 1968P. C. Wason (1968). Reasoning about a rule. Quarterly Journal of Experimental Psychology 20(3): 273–281 open ↗
- Ragni et al. 2018Marco Ragni, Ilir Kola, Philip N. Johnson-Laird (2018). On Selecting Evidence to Test Hypotheses: A Theory of Selection Tasks. Psychological Bulletin 144(8): 779–796 open ↗
- Nickerson 1998Raymond S. Nickerson (1998). Confirmation Bias: A Ubiquitous Phenomenon in Many Guises. Review of General Psychology 2(2): 175–220 open ↗
- Zhu et al. 2018Meng Zhu, Yang Yang, Christopher K. Hsee (2018). The Mere Urgency Effect. Journal of Consumer Research 45(3): 673–690 open ↗
- Tetlock 2005Philip E. Tetlock (2005). Expert Political Judgment: How Good Is It? How Can We Know?. Princeton University Press open ↗
- Goldratt 1984Eliyahu M. Goldratt, Jeff Cox (1984). The Goal: A Process of Ongoing Improvement. North River Press open ↗
- Parkinson 1955C. Northcote Parkinson (1955). Parkinson's Law. The Economist, 19 November 1955 open ↗
- Ariely & Wertenbroch 2002Dan Ariely, Klaus Wertenbroch (2002). Procrastination, Deadlines, and Performance: Self-Control by Precommitment. Psychological Science 13(3): 219–224 open ↗
- Cirillo 1987Francesco Cirillo (1987). The Pomodoro® Technique (method developed 1987; book editions 2006, 2018). Cirillo Consulting GmbH / pomodorotechnique.com open ↗
- Bayles & Orland 1993David Bayles, Ted Orland (1993). Art & Fear: Observations on the Perils (and Rewards) of Artmaking. Capra Press / Image Continuum Press
- Boyd 1986John R. Boyd (1986). Patterns of Conflict (briefing, Dec 1986; ed. C. Richards & C. Spinney, 2007). also in A Discourse on Winning and Losing, Air University Press, 2018 open ↗
- The Equalizer 2014Antoine Fuqua (dir.), Columbia Pictures (2014). The Equalizer (RU release: «Великий уравнитель»). feature film; scene in Slavi's office open ↗
- Leroy & Glomb 2018Sophie Leroy, Theresa M. Glomb (2018). Tasks Interrupted: How Anticipating Time Pressure on Resumption of an Interrupted Task Causes Attention Residue and Low Performance on Interrupting Tasks and How a 'Ready-to-Resume' Plan Mitigates the Effects. Organization Science 29(3) open ↗
- Gollwitzer & Sheeran 2006Peter M. Gollwitzer, Paschal Sheeran (2006). Implementation Intentions and Goal Achievement: A Meta-Analysis of Effects and Processes. Advances in Experimental Social Psychology 38: 69–119 open ↗
- Kushlev & Dunn 2015Kostadin Kushlev, Elizabeth W. Dunn (2015). Checking email less frequently reduces stress. Computers in Human Behavior 43: 220–228 open ↗
- Goodhart 1975Charles A. E. Goodhart (1975). Problems of Monetary Management: The U.K. Experience. Papers in Monetary Economics, Reserve Bank of Australia
- Strathern 1997Marilyn Strathern (1997). 'Improving ratings': audit in the British University system. European Review 5(3): 305–321
- Kaplan & Norton 1996Robert S. Kaplan, David P. Norton (1996). The Balanced Scorecard: Translating Strategy into Action. Harvard Business School Press
- Webb & Sheeran 2006Thomas L. Webb, Paschal Sheeran (2006). Does changing behavioral intentions engender behavior change? A meta-analysis of the experimental evidence. Psychological Bulletin 132(2): 249–268 open ↗
- BBC 2017Simon Maybin (BBC More or Less) (2017). Busting the attention span myth. BBC News, 10 Mar 2017 open ↗
- WSJ 2017Jo Craven McGinty (2017). Is Your Attention Span Shorter Than a Goldfish's?. Wall Street Journal, 17 Feb 2017 open ↗