Start here"Knowledge is power" gets quoted as a promise. Yet the exact Latin, Thomas Hobbes's of 1668, goes on "but a small one". So which information is worth taking in, and from whom?
“Knowledge is power” is a familiar line. Where it comes from turns out to be a good place to start, because it says something useful about how information turns into power, and how often it does not.
This page rests on one idea, and it is our own way of putting it: your mind works a little like your stomach. What you take in, again and again, shapes how you think. That is a comparison, not a scientific finding, and the page says where research bears on it and where it does not.
The verdicts
A quick filter before you believe something
Here are four questions. None of them takes long, and the last two link to pages that go deeper.
-
Who is saying it, and what do they gain? Gain is not only money: attention, followers and status count too. A motive tells you how much to lean on someone’s word. It does not settle whether their argument holds.
-
Is it a fact, an opinion or a feeling? A Pew Research Center survey of 5,035 US adults in 2018 defined a factual statement as one that objective evidence can prove or disprove. An opinion, it said, reflects the beliefs and values of whoever expressed it.
A majority correctly sorted at least three of five statements of each kind. Pew called that result only a little better than random guesses.
The third category, a feeling, is our addition. A statement can describe how someone feels about a thing and still be presented as a fact about it.
-
Where is the original source? A quote, a statistic or a study is only as good as the place it first appeared. Our guides on finding where a quote first appeared, on numbers that mislead and on reading a study cover the three cases.
-
Do other trustworthy sources agree? One quick way to find out is to leave the page and see what others say about the source and the claim. That is lateral reading, the core of the SIFT method.
The same questions apply when the answer comes from an AI chatbot. Our page on critical thinking in the age of AI looks at what studies have found.
Who to listen to
Two signals help here, in our reading: what a source loses when it is wrong, and what it gains from your attention.
People who pay a price when they are wrong
Nassim Nicholas Taleb’s book Skin in the Game, published by Random House in 2018, is built around this. The publisher’s description quotes him: “Never trust anyone who doesn’t have skin in the game”. Without it, he says, fools and crooks benefit, and their mistakes never catch up with them.
The publisher’s description puts the rule in plain words: you cannot make profits and pass the risks on to others. Our summary of Skin in the Game covers what holds up.
Our reading: someone who loses something when they are wrong has a reason to be careful. That loss might be money, reputation or their own health.
Someone with a track record you can check has given you something to check. Neither is proof that they are right, but both are better signals than confidence.
People who are paid by your attention
If a source earns from your attention, its first job is to hold that attention. Being right does not always do that.
This is not a reason to dismiss anyone. In “How to Disagree”, Paul Graham gives the example of a senator arguing that senators’ pay should rise: “Of course he would say that. He’s a senator.”
That reply, he writes, would not refute the argument, but it may at least be relevant to the case. Our logical fallacies cheat sheet draws the same line: facts about a person matter when you are relying on their word, not when you are judging their reasoning.
Watch your own bias, and find the strongest version of the other side
The filter above points outwards. The harder part points inwards, because the same shortcuts work on whoever is doing the checking. Our cognitive biases page checks fourteen of the famous findings and which held up.
Munger made a practical point about one of these shortcuts. People hold on to the first conclusion they reach. So, he noted, judges and juries must sit through long, skilful presentations of the side they would not naturally favour. He said this helps prevent a good deal of bad thinking caused by first-conclusion bias.
John Stuart Mill made the same argument in On Liberty, published in 1859: “He who knows only his own side of the case, knows little of that”. It is not enough, he wrote, to hear the other side from your own teachers. You need to meet its arguments at their most plausible and persuasive.
The levels, he notes, describe the form of a reply, not whether it is correct. A DH6 reply might still be unconvincing, but one at DH2 or lower is always unconvincing.
Graham’s essay does not use the word steelman. In our reading, the two ideas fit together: you cannot refute someone’s central point until you have stated it in a form they would accept.
Fewer, deeper sources
Seneca’s second letter is the classic case for depth. “Everywhere means nowhere,” in Gummere’s translation (2.2).
A person who spends all their time travelling, he says, ends up with many acquaintances but no friends. The same holds for readers who visit every author in a hasty, hurried way.
His advice in Latin is short: probatos itaque semper lege. In our translation: so always read the tried and tested ones. Our summary of Letters from a Stoic covers the rest of the letters.
His sorting of books into tasted, swallowed, and chewed and digested is a way to spend attention unevenly on purpose. A few books, he says, are to be read wholly, with diligence and attention.
Depth does not mean narrowness. Our page on the Renaissance man looks at what research says about range and depth together.
Where “knowledge is power” comes from
Credited to Francis Bacon
Bacon’s line was about God
The Latin words begin with Francis Bacon, in a passage about God. In February 1597 Bacon published the first edition of his Essayes. The same volume held his Religious Meditations and Places of Perswasion and Disswasion.
In brackets he adds the famous phrase: “for knowledge itself is power”. That is the English of James Spedding’s 1859 edition. The Latin of the aside reads nam et ipsa scientia potestas est.
In context, the knowledge is God’s, and Bacon’s point is that God’s knowing is part of God’s power. It is not advice about getting ahead.
Bacon’s human version came later
Bacon did write about human knowledge and power, more than twenty years later. In the Novum Organum, published in October 1620, the third aphorism reads Scientia et potentia humana in idem coincidunt. In our translation: human knowledge and human power come to the same thing.
Joseph Devey’s translation begins “Knowledge and human power are synonymous”. The reason it gives: if you do not know the cause, you cannot bring about the effect, and nature is mastered only by submitting to it.
The exact Latin is Hobbes’s, and he added a catch
In the Latin edition of Leviathan, published in 1668, Hobbes writes Scientia potentia est; sed parva. In our translation: knowledge is power, but a small one. It is small, he says, because outstanding knowledge is rare and is seen by very few (our translation of quia scientia egregia rara est).
The idea itself is older than either man. The King James Version of Proverbs 24:5 reads: “A wise man is strong; yea, a man of knowledge increaseth strength”.
Our reading: the line is less a promise than a pair of conditions. Bacon’s version needs knowledge of causes that you can act on. Hobbes’s caveat is that knowledge only counts as power when other people can see it.
↑ All verdictsYour mind is like your stomach (our framing)
This section offers a comparison, not a claim that the brain digests information the way the body digests food. What you read, watch and scroll through, again and again, becomes the material you think with.
The comparison is old. Seneca makes it in his second letter to Lucilius. Food does no good, he writes, and is not taken into the body if it leaves the stomach as soon as it is eaten. Toying with many dishes, he adds, is the sign of an over-fussy appetite (Letter 2.3–2.4, in Richard M. Gummere’s translation).
Francis Bacon used the same image in his essay “Of Studies”: “Some books are to be tasted, others to be swallowed, and some few to be chewed and digested”. The line was already in the first edition of 1597.
Modern writers have kept the comparison going. Clay Johnson’s book The Information Diet is summed up in its catalogue record the same way. Too much junk information can lead to cluelessness, just as too much junk food can lead to obesity.
Rolf Dobelli’s 2013 essay in The Guardian argued that “news is to the mind what sugar is to the body”. Both are arguments by their authors, not research findings.
Where research bears on it: repetition
One finding fits the comparison closely. The illusory truth effect is the finding that people rate statements as truer when they have seen them before.
The first paper on it was published in 1977, by Lynn Hasher, David Goldstein and Thomas Toppino. A 2010 meta-analysis pooled 51 studies of it.
Two later results matter for everyday reading.
-
In a 2018 study, people saw real false headlines as they had appeared on Facebook. Even a single exposure made a headline seem more accurate later, both in the same session and a week later. Prior exposure had no effect on entirely implausible statements, such as that the earth is a perfect square.
The effect holds up, with limits.
-
A 2022 systematic map of 181 studies found that all were run at Western universities, and most used trivia statements in a single session. It also noted a lack of replications by outside teams.
The 2021 study’s authors point out a conflict. Their finding that the effect fades does not fit an earlier meta-analysis, which found no link across studies between the size of the effect and the delay between sessions.
In short: repetition makes claims feel truer, reliably in the lab. How long the effect lasts is not settled.
Where research bears on it: attention
The economist and computer scientist Herbert Simon put the other half of the problem in a 1971 essay, “Designing Organizations for an Information-Rich World”. In an information-rich world, he wrote, the wealth of information means a shortage of something else: the attention of those who receive it. “Hence a wealth of information creates a poverty of attention”.
If attention is the scarce thing, then choosing what to feed your mind is also choosing where your attention goes.
Popular claims, checked
Our attention span is now shorter than a goldfish’s
Doesn't hold up The comparison is vivid and the numbers sound precise, but no research has been found behind them. The claim says the average attention span fell from 12 seconds in 2000 to 8, less than a goldfish’s 9.
The BBC statistics programme More or Less traced it in 2017. The figure led back to a 2015 industry report, which credited it to another website. The sources that website listed could find no research behind it.
One attention researcher told the BBC that attention depends very much on the task. The programme concluded there was no evidence that human attention spans are shrinking.
That is time on one screen at work, not a general attention span, and it is not the 8-second figure. Goldfish do not have short memories either, as our real or myth page explains.
False news spreads faster than true news
Partly A widely cited study is behind this one, though a later reanalysis explains its result differently. Published in Science in 2018, it looked at about 126,000 stories on Twitter from 2006 to 2017. Six fact-checking organisations had judged each one true or false. Falsehood, it found, spread “farther, faster, deeper, and more broadly than the truth”.
The study also found that bots spread true and false stories at the same rate. The authors took this to mean that people, not bots, did the extra spreading.
A 2021 reanalysis in PNAS compared stories that spread to the same size. It found that the reported differences in how they spread can almost entirely be explained by the false-news cascades being larger.
So false stories reached more people in that data. Whether they travel by a different mechanism is less clear. Both studies are about Twitter, and about stories fact-checkers had checked.
Echo chambers and filter bubbles trap most people
Still argued Researchers disagree here, and the studies point in different directions.
A 2022 literature review by the Reuters Institute for the Study of Journalism found echo chambers much less widespread than is commonly assumed. It found no support for the filter bubble hypothesis. Studies in the UK put the share of people in politically partisan online news echo chambers at between six and eight percent.
On the other side, a 2023 study in Nature used data from 2020 for all active adult Facebook users in the US. Content from like-minded sources made up the majority of what people saw, although political information and news were only a small fraction of it.
In the same study, cutting that content by about one-third for 23,377 users had no measurable effect on eight measures of attitudes.
A companion 2023 study in Science used aggregated data for 208 million US Facebook users. It found that ideological segregation in political news is high. It also grows as you move from what people could see, to what they did see, to what they engaged with.
Our reading: one-sided feeds are common on some platforms, but being fully sealed off is rare. The effect of one-sided exposure on beliefs is small or unproven in the studies so far. Both Facebook studies covered US users during the 2020 election.
Following the news makes people anxious
Partly The short-term part has support, which makes the larger claim easy to believe. Small experiments show that watching negative news can raise anxiety in the short term.
Longer-term effects rest on surveys. After the Boston Marathon bombings, a 2014 study in PNAS looked at people’s media exposure in the week after. Six or more hours a day of bombing-related media exposure was associated with more acute stress than direct exposure to the bombings.
A 2023 study of US adults described what it called “problematic news consumption”. It found greater mental and physical ill-being among the 16.5% with the most severe pattern. A 2025 replication over two waves found 7.5% in that group, and the link with physical ill-being did not hold over time.
Surveys like these show links. They cannot settle whether news causes lasting distress, or distressed people follow more news.
Taleb coined “skin in the game”
Doesn't hold up The phrase is the title of his book, which explains the link, but it is older than the book. The Oxford English Dictionary dates the phrase from 1976, decades before Taleb’s 2018 book. It labels the phrase colloquial, first used in North American business.
What Taleb did was turn a business phrase into a broad rule about risk and fairness.
Munger’s misjudgment talk was a single 1995 talk
Munger’s book Poor Charlie’s Almanack describes it as selections from three of his talks, combined into one talk never given. The book’s note on the talk calls it a checklist of 25 standard causes of human misjudgment.
Further reading
These books take up the same theme, and each has a summary on this site:
- Amusing Ourselves to Death (Neil Postman, originally published in 1985).
- The Demon-Haunted World (Carl Sagan, 1995).
- Thinking, Fast and Slow (Daniel Kahneman, 2011).
- Digital Minimalism (Cal Newport, 2019).
- Factfulness (Hans Rosling with Ola Rosling and Anna Rosling Rönnlund, 2018).
- The Shallows (Nicholas Carr, 2010).
Three more on the same ground:
- Verified by Mike Caulfield and Sam Wineburg (2023), a guide that builds on techniques such as SIFT and lateral reading.
- Stop Reading the News by Rolf Dobelli (2020).
- The Filter Bubble by Eli Pariser (2011).
Quick answers
Who said "knowledge is power"?
Francis Bacon wrote the Latin in 1597, as a bracketed aside in a passage about God's knowledge and power; James Spedding's 1859 edition gives it as "for knowledge itself is power". The exact form scientia potentia est is Thomas Hobbes's, from 1668, and it goes on sed parva: in our translation, but a small one.
Is our attention span shorter than a goldfish's?
No research has been found behind that claim. When the BBC's More or Less traced the 8-second figure in 2017, its sources could find no research behind it, and the programme concluded there was no evidence that attention spans are shrinking.
What is steelmanning?
To steelman an argument is to make the strongest possible case in favour of it, the opposite of a straw man. John Stuart Mill made the same argument in 1859: "He who knows only his own side of the case, knows little of that".
Does false news spread faster than true news?
A 2018 study of about 126,000 Twitter stories found falsehood spread farther and faster. A 2021 reanalysis found the differences can almost entirely be explained by the false-news cascades being larger.
Do echo chambers trap most people?
Researchers disagree. A 2022 Reuters Institute review found echo chambers much less widespread than commonly assumed. A 2023 study of US Facebook users found most of what people saw came from like-minded sources, but cutting that content by about a third had no measurable effect on attitudes.
Use this
Take in fewer sources, and check a claim before you believe or share it.
Who gains if I believe this, and where did it first appear?
Quick quiz
Nothing is scored or saved. Recalling it once now is what makes it last.
1. In Bacon's 1597 text, what was "knowledge itself is power" about?
Answer: God's knowledge being part of God's power.Read it again ↑
2. What does the illusory truth effect describe?
Answer: People rate statements as truer when they have seen them before.Read it again ↑
3. In Pew's 2018 survey, how well did US adults sort facts from opinions?
Answer: Only a little better than random guesses.Read it again ↑
How we checked this · Last checked October 2026
- We read Bacon, Hobbes, Seneca and Mill in the editions and translations in the Sources; where the English is ours, the page says "our translation".
- For the studies we read each abstract and, where it was open, the full text.
- The books under Further reading were not read for this page; their dates come from their publishers' pages and Open Library records. Simon's 1971 essay is quoted through Oxford Essential Quotations.
- Bacon's "Of Studies" is quoted from a modern-spelling version of the final edition of 1625, The Essays or Counsels, Civil and Moral.
- Munger's talk was read on the Stripe Press page for Poor Charlie's Almanack.
Sources
Knowledge is power
- Francis Bacon, Essayes. Religious Meditations. Places of Perswasion and Disswasion (London, 1597), Meditationes Sacrae, “De Haeresibus” and “Of Studies” (Internet Archive scan); Latin text and English translation in James Spedding (ed.), The Works of Francis Bacon, vol. VII (London, 1859), pp. 241 and 253 (archive.org scan).Primary
- Francis Bacon, Novum Organum (1620), Book I, aphorism 3: Latin text, The Latin Library; translated by Joseph Devey, Project Gutenberg.Primary
- Francis Bacon, The Essays or Counsels, Civil and Moral, “Of Studies” (Project Gutenberg).Primary
- Stanford Encyclopedia of Philosophy, “Francis Bacon” and “Thomas Hobbes”.Scholarly
- Thomas Hobbes, Leviathan (1651), ch. X (Project Gutenberg); Latin Leviathan, Part I, cap. X, in Opera Philosophica, ed. W. Molesworth, vol. 3 (London: Bohn; dated 1839 in the Internet Archive record) (Internet Archive).Primary
- Proverbs 24:5, King James Version (Wikisource).Primary
Food for the mind
- Seneca, Moral Letters to Lucilius, Letter 2, translated by Richard M. Gummere (Wikisource); Latin text (The Latin Library).Primary
- Clay Johnson, The Information Diet, Open Library record.Popular
- Rolf Dobelli, “News is bad for you – and giving up reading it will make you happier”, The Guardian, 12 April 2013.Popular
- Hasher, Goldstein & Toppino, “Frequency and the conference of referential validity”, Journal of Verbal Learning and Verbal Behavior 16 (1977), doi:10.1016/S0022-5371(77)80012-1.Scholarly
- Dechêne, Stahl, Hansen & Wänke, “The truth about the truth: a meta-analytic review of the truth effect”, Personality and Social Psychology Review 14 (2010), doi:10.1177/1088868309352251.Scholarly
- Fazio, Brashier, Payne & Marsh, “Knowledge does not protect against illusory truth”, Journal of Experimental Psychology: General 144 (2015), doi:10.1037/xge0000098.Scholarly
- Pennycook, Cannon & Rand, “Prior exposure increases perceived accuracy of fake news”, Journal of Experimental Psychology: General 147 (2018), doi:10.1037/xge0000465.Scholarly
- Henderson, Simons & Barr, “The Trajectory of Truth”, Journal of Cognition 4 (2021), doi:10.5334/joc.161.Scholarly
- Henderson, Westwood & Simons, “A reproducible systematic map of research on the illusory truth effect”, Psychonomic Bulletin & Review 29 (2022), doi:10.3758/s13423-021-01995-w.Scholarly
- Herbert A. Simon, “Designing Organizations for an Information-Rich World”, in Martin Greenberger (ed.), Computers, Communications, and the Public Interest (1971), pp. 40–41, as quoted in Oxford Essential Quotations.Primary
Who to listen to, and how to disagree
- Penguin Random House, Skin in the Game by Nassim Nicholas Taleb.Popular
- Oxford English Dictionary, “skin in the game”, in skin, n.Scholarly
- Charles T. Munger, “The Psychology of Human Misjudgment”, Poor Charlie’s Almanack, Talk Eleven (Stripe Press).Primary
- Paul Graham, “How to Disagree”, March 2008.Primary
- John Stuart Mill, On Liberty (1859), chapter II (Project Gutenberg).Primary
- OpenStax, Introduction to Philosophy, “2.3 Developing Good Habits of Mind”.Scholarly
- Pew Research Center, “Distinguishing Between Factual and Opinion Statements in the News”, 18 June 2018.Primary
Popular claims
- American Psychological Association, Speaking of Psychology, “Why our attention spans are shrinking, with Gloria Mark, PhD”, February 2023.Popular
- Simon Maybin, “Busting the attention span myth”, BBC News, 10 March 2017.Popular
- Vosoughi, Roy & Aral, “The spread of true and false news online”, Science 359 (2018), doi:10.1126/science.aap9559.Scholarly
- Juul & Ugander, “Comparing information diffusion mechanisms by matching on cascade size”, PNAS 118 (2021), doi:10.1073/pnas.2100786118.Scholarly
- Arguedas, Robertson, Fletcher & Nielsen, “Echo chambers, filter bubbles, and polarisation: a literature review”, Reuters Institute, 19 January 2022.Scholarly
- González-Bailón et al., “Asymmetric ideological segregation in exposure to political news on Facebook”, Science 381 (2023), doi:10.1126/science.ade7138.Scholarly
- Nyhan et al., “Like-minded sources on Facebook are prevalent but not polarizing”, Nature 620 (2023), doi:10.1038/s41586-023-06297-w.Scholarly
- Johnston & Davey, “The psychological impact of negative TV news bulletins”, British Journal of Psychology 88 (1997), doi:10.1111/j.2044-8295.1997.tb02622.x; Szabo & Hopkinson, “Negative psychological effects of watching the news in the television”, International Journal of Behavioral Medicine 14 (2007), doi:10.1007/bf03004169.Scholarly
- Holman, Garfin & Silver, “Media’s role in broadcasting acute stress following the Boston Marathon bombings”, PNAS 111 (2014), doi:10.1073/pnas.1316265110.Scholarly
- McLaughlin, Gotlieb & Mills, “Caught in a Dangerous World”, Health Communication (2023), doi:10.1080/10410236.2022.2106086; “Problematic News Consumption and Its Relationship to Mental and Physical Health: A Replication Study”, Health Communication (2025), doi:10.1080/10410236.2024.2434955.Scholarly
Reading
- Penguin Random House pages for Amusing Ourselves to Death, Digital Minimalism and The Filter Bubble; Open Library records for first publication years.Popular
- Hodder & Stoughton, Stop Reading the News; University of Chicago Press, Verified; Macmillan, Thinking, Fast and Slow and Factfulness.Popular
Collect what you can act on
In Bacon's human version, knowledge is power when it shows you causes you can act on. Hobbes called it a small power, because outstanding knowledge is rare and few people see it. Our view: choose information you can use, and let your work show what you know.
Plan a month of better inputs
- List what you take in. Write down every feed, channel and newsletter you check on a normal day.
- Drop one repeater. Cut one source that mostly repeats what you've already heard. Repetition can make a claim feel truer without making it true.
- Pick one deep source. Choose one book or long read for the month, and plan to finish it.
- Add one source you can check. Look for someone with a track record you can check, or who loses something when wrong. Neither proves them right, but both beat confidence.
- Review once a week. Ask which sources changed something you did or decided. Keep those, and trim the rest. This is our rule of thumb.
This week: run the filter on three claims
Pick three claims you saw shared this week. Put each through the four questions, and write one line on what you found. Then note which ones you'd still share.
Bring the other side into your decisions
- Hear it before you decide. Munger noted that people hold on to the first conclusion they reach. Hear the opposing case before yours sets.
- Ask someone to argue against you. Before a big choice at work or home, ask one person to make the strongest case for the other option.
- Write their case so they'd accept it. Show your version to someone who holds the view, and ask if they'd accept it.
- Answer the central point. Graham's ladder rates the form of a reply. Aim for the top rung, not a reply to tone or to the person.
Goes with: Watch your own bias, and find the strongest version of the other side ↑
Go further on this site
- Numbers that misleadTen statistical traps, for the claims that come with a figure attached.
- How to read a studyNine questions to ask when a claim says 'a study found'.
- Use emotion honestlyHow emotional triggers work, with a short quiz to practise spotting them.
- Digital Minimalism, summarisedCal Newport's idea of knowing how much technology is just enough, and what holds up.