← Company tools
Wren · Volume Ten · Company

The Company Playbook twelve rooms, one card each

Twelve seasons of practices done with other people, one card each: a question read a week ahead, a minute read back, a passage read at home, a timed talk, a fixed hour, one line reported to one friend, a letter, a lesson led, a bench, a written decision, a regular's hour, a book of names. Every number comes from the seasons' own sourcing. What each finding can't tell you is on the card too. Nothing here scores a meeting, a member or a friendship. All twelve cards come with a free tool, built for the person who keeps the record.

The whole volume, in one season of meetings

A season of meetings

What the record supports at each point of an ordinary run of meetings. Each box points to the card that backs it and the free tool that keeps it. Green boxes rest on the volume's best-supported cards.

From the question read a week ahead to the dated line beside a name

Most of these need a room that has agreed to meet, and the cost the record documents is that the room has to keep meeting. Pick one or two; nobody in these seasons did all of them.

  1. The week beforeAnnounce the question, read the pages
    • One question announced tonight, argued next time
    • The same passage, read at home, before anyone meets
    • The answer is made in the week
    Cards 1, 3 · Queries, Circle
  2. At the meetingA timed talk, a lesson led, a count at the door
    • Green at five minutes, yellow at six, red at seven
    • An evaluation from a member, in front of the room, written down
    • A name and a page number for each lesson
    • How many came, written at the door
    Cards 4, 8, 9 · Evaluation, Square, Shed
  3. Before anyone leavesRead the minute back
    • A few plain lines, drafted while the room waits
    • "Unity not reached" is a minute too
    • Stand aside out loud, with a reason
    Card 2 · Minute
  4. Sunday nightOne line, to one person
    • Write what you will do, with a date
    • Send it to one other person
    • A week later: done, or not done
    Card 6 · Report
  5. Within a weekOne letter, delivered
    • To someone especially kind, never properly thanked
    • Write it and hand it over inside the week
    • Expect it to land better than you guess
    Card 7 · Letter
  6. At a fixed hourThe same room, the same people
    • A dated meeting at a fixed hour
    • A call between meetings
    • Or the same counter at the same hour, and a few words
    Cards 5, 11 · Rooms, Regular
  7. At home, weeklyWrite the decision down
    • One dated line per decision
    • The line is what anyone can check next week
    • Keep the dinner as a room, not a cure
    Card 10 · Council
  8. Going through the listA date beside a name
    • Go through the names on purpose
    • Write when and how: wrote, called, visited
    • No number for how many names a book should hold
    Card 12 · Names
Part one · evidence 3

The smallest rooms, and one counted room

Three practices with trials behind them. Two are one person and one other person. The third is a room volunteers cannot be assigned to, so the trials randomized the push toward it.

Card 6 · The Report
Cost 2Evidence 3Hype 4

One line, and a week later someone asks

Across 138 experiments, monitoring your progress helped a little, and reporting it to at least one other person helped somewhat more. The famous 76 percent is not in the study it is quoted from.

Effect on reaching the goal, by who saw the record (2016, d+)

Reported, 95 studies0.47
Private, 14 studies0.19
Public, 3 studies0.55

The public row rests on three studies and its interval runs from -0.16 to 1.26, across zero. Leave it aside. Writing the progress down also mattered: 0.43 recorded against 0.29 not.

The one study of the full routine: mean self-rated score, five groups

1 Goal unwritten4.28
2 Written down6.08
3 Plus commitments5.08
4 Given to a friend6.41
5 Weekly report7.6

On a scale the summary does not define. Only 149 of 267 people completed it, over four weeks, rating themselves.

1928Hill's course ran to sixteen lessons, and the first was the Master Mind. His definition counts two people as enough.
1908Hill said Carnegie gave him the assignment that year. Carnegie's biographer found no evidence the two men ever met.
1991The oldest trace of the famous goals study at Yale is a magazine article citing a tape whose speaker had heard the statistics somewhere. Yale and Harvard both looked and found nothing.
2018A business blog promised to raise "your chances of goal success from 43% to 76%." Neither number appears in the researcher's summary.
Sunday nightWrite one line about what you will do, with a date beside it.
To one personSend it. In the 2016 coding, "at least one other person" was enough, and Group 5 sent theirs to one friend.
A week laterDone, or not done. The useful moment is when someone asks and there is a written line to check against.

The 95 studies did not separate a friend from the researcher running the study, and the authors could not say whether the larger effect was commitment, not wanting to look bad, or the experimenter watching. Nothing here tests a circle of peers against one person, in either direction.

Use it: the free toolReport →Write one line, send it to one person, and a week later mark it done or not done. It never turns the marks into a percentage.

Harkin et al. 2016 · Matthews 2007, 2015 · Hill 1928 · Season 6

Card 7 · The Letter
Cost 2Evidence 3Hype 2

One letter, delivered within a week

The thank-you letter produced the largest change in a 2005 trial, and by three months it was gone. The person who receives it is more pleased than the writer expects.

Gratitude exercises, pooled (2016, effect size d)

Vs active exercise, k 9 Anxiety, not sig. Vs alternative, k 20 Vs measurement, k 5 -0.100.20.4

Against an exercise that was itself active, gratitude did "as well as, but not better than." In 2021, 27 studies and 3,675 people: "a small effect on symptoms of depression and anxiety."

Writers guessed, readers reported (2018, three experiments)

The writer expectedLess surprise at the reasonMore awkwardnessA less positive reader
The reader saidMore surprisedLess awkwardMore positive
ShelfThe letters themselves are the old count: seventeen volumes of one friar's mail, thirteen of one secretary's.
2005"One week to write and then deliver a letter of gratitude in person." 577 recruited online, 411 analyzed. The largest change in the study, gone by three months.
2018293 adults seeking psychotherapy: the gratitude-letter group reported better mental health at about 4 and 12 weeks than the other two groups.
2018The writers' wrong guesses went along with whether they were willing to express gratitude at all.
WhoOne person who was especially kind to you and was never properly thanked.
The weekWrite it and deliver it within a week, in person if you can.
KeepThe date and the name. It is the part of a letter that survives.

Small, and it fades. Gratitude letters did about as well as other active exercises and no better. "Rewires the brain" rests on one imaging study that compared two groups once; its authors call it cross-sectional.

Use it: the free toolLetter →Who it is to, the day you started, and the day it was delivered and how. It counts the days against the week.

Seligman, Steen, Park & Peterson 2005 · Davis et al. 2016 · Kumar & Epley 2018 · Wong et al. 2018 · Cregg & Cheavens 2021 · Season 7

Card 5 · The Rooms
Cost 3Evidence 3Hype 3

A fixed hour, the same people, and a call in between

The meetings were almost never randomized, and never for people who came on their own. A clinician's push toward them was, and the room came out better than its critics said and not as well as its own book said.

Continuous abstinence at twelve months, facilitation against other treatments (2020, risk ratio)

1.21 1.031.42 no difference 0.81.6

2 studies, 1,936 people, high certainty, and it held at 24 and 36 months. In plain words, about a fifth more often. On most other measures, about as well.

Three numbers that were not counted this way

100+→1939: "more than one hundred men and women who have recovered"
50%→1955 foreword: "got sober at once," with no count behind it
5 to 10%→2014: a critic's figure, by his own account, on the radio
1935June 10, the date the fellowship keeps as its beginning. The first meeting of the two founders ran from five o'clock until a quarter past eleven.
1997Two trials, 952 outpatients and 774 in aftercare, three 12-week treatments: "There was little difference in outcomes by type of treatment."
2014Six data sets from five trials: the extra attendance caused by the coin flip increased days abstinent at 3 and 15 months.
2012In 1,726 people, changes in who they spent time with, and in confidence about staying sober around others, explained "43% to 67%" of the link.
The meetingA dated meeting at a fixed hour, in a borrowed room, with the same people.
Between meetingsThe call: "a ten-cent phone call and a cup of coffee."
What held up"Holding onto each other," said by an early member in 1962, and measured fifty years later as a change in a person's social world.

In these trials, what was randomized was a clinician's push, not the room. The finding is for adults who came voluntarily, and it is mainly about abstinence. Selection bias was rated high in 11 of the 27 studies. Nothing here is treatment or medical advice.

Use it: the free toolRooms →Your meeting's day, hour and place, a dated line each time you go, and a dated line for each call. Nothing else is kept.

Project MATCH 1997 · Ferri, Amato & Davoli 2006 · Kelly, Hoeppner, Stout & Pagano 2012 · Humphreys, Blodgett & Wagner 2014 · Kelly, Humphreys & Ferri 2020 · Season 5

Part two · evidence 2

Studies beside the room

Each of these has studies near it: of a smaller act inside the practice, of a course taught in its room, or of who had ties. None is a trial of the room itself.

Card 4 · The Evaluation
Cost 3Evidence 2Hype 3

Five minutes, then a member tells you what they heard

The speaking club has kept dated records of itself since 1905. The number that came to stand for the fear it addresses was a list item, retold as a ranking.

What 2,543 adults checked from a list of fourteen (1973)

Speaking before a group41%
Death19%

Death came seventh. Nobody was asked to compare the two.

The fear itself: about three in ten, over a lifetime (8,098 people)

Treatments the club resembles, pooled (Hedges g)

2019, 30 trials0.74
2025, 26 studies1.17
The club itselfnone
1905Bloomington, March 24: older high school boys, dinner at about fifteen cents a plate, talks "usually limited to five minutes," and an evaluation from the older members.
1924Santa Ana, October 22: about two dozen men in a basement, membership limited to thirty, and "evaluation by the members."
1977A book of lists ranked fourteen fears with speaking first and death seventh, a ranking the 1973 survey never contained.
1996499 people called at random in Winnipeg: one third reported excessive anxiety speaking to a large audience; 10 percent, 49 people, said it interfered markedly.
The talkGreen at five minutes, yellow at six, red at seven, on the time record sheet for a five to seven minute speech.
The evaluationTwo to three minutes, from a member, spoken in front of everyone who heard the talk, and written down.
The recordA date, the speakers, the evaluations. It is why the club's history could be written from the room's side.

I could not find a trial of the club itself, and a resemblance is not a result. The fear is common and for most people the milder kind. The seventy-five percent that circulates is not from any population survey.

Use it: the free toolEvaluation →A timer that shows green, yellow and red at five, six and seven minutes, and a written slip the speaker takes home. No grade.

Smedley chronology 1965 · Bruskin Report No. 53, 1973 · Stein, Walker & Forde 1996 · Kessler, Stein & Berglund 1998 · Ebrahimi et al. 2019 · McWilliam, Beattie & Callow 2025 · Kalińska 2022 · Season 4

Card 8 · The Hollow Square
Cost 3Evidence 2Hype 2

A name and a page number

The singing square has counted every lesson led for decades, and the people who keep the count tell the room not to sing to it. The science beside it is one small pilot and reviews that find the studies at risk of bias.

The 77 songs dropped for the 2025 edition

Share of the 1991 book14%
Share of lessons sung from it3%

The committee considered song use, as the minutes page says it would, and the songs that went were the ones the square had rarely led.

The best test of group singing: a pilot (2015)

258peopleaged 60 or over, East Kent, randomly assigned6monthssinging against usual activities2.35pointshigher; the interval barely excluded zero

Dropped as reasons

the synchronized heart: eleven 18-year-oldsthe bonding hormone: 4, 16 and 21 peopleendorphins: a crafts class came out the same
1721Boston, March 16: "a Society [of] persons learning to Sing, began a quarterly solemnity."
1930Mineral Wells, Texas: "The singers sat in folding chairs on four sides of a rectangular open space where the leader stood."
1995The minutes archive is searchable from 1995: each lesson printed with the leader's name and the page number.
20251,155 songs submitted, 113 added after 13 test singings, 590 in the new book.
The squareFour sides facing in, the sound on the middle, no audience.
The lessonStand in the middle for a turn, pitch the tune by voice, and lead one song.
The minuteThe leader's name and the page number, written down. "Sing whatever you are moved to sing."

Nobody has wired a convention. Reviews find a benefit and find the studies behind it at risk of bias. The only outcome measured in the square itself is that it does not seem to wear out the voice faster than other singing.

Use it: the free toolSquare →A secretary's sheet for a singing: each lesson as it is led, the leader's name and the page, in order. It never ranks songs.

Mather diary 1721 · Jackson, at Mineral Wells 1930 · Coulton et al. 2015 · Daykin et al. 2018 · Williams, Dingle & Clift 2018 · Rees, Ivey & Karlsberg 2025 · Morgan-Ellis · Season 8

Card 9 · The Shed
Cost 3Evidence 2Hype 2

Count who comes, and date what gets made

The oldest record in this season is four numbers written at a door. Four reviews in nine years agree that the modern shed studies are mostly small and self-reported.

Glasgow, 1800: attendance at four lectures for working men

First lecture75
Second200
Third300
Fourth500

"This was a greater number than the hall would conveniently hold." By 1851 one count held 610 institutions with 102,050 members.

Four reviews in nine years, in their own words

2013"either descriptive surveys or small qualitative studies"
2016"As yet, there is no conclusive evidence"
2019"predominantly self-reported"
2022"does not allow a conclusive result"
1993Goolwa, South Australia, February, by the national association's count. Whether it was first depends on what you count as a start.
2006A survey wrote down who was there: about half over 65, about one in five war veterans, about one third referred by a health or welfare worker.
2007"Men don't talk face to face, they talk shoulder to shoulder": a keynote line, by the association's account. Nobody tested it.
202520 clusters of sheds, 261 men: a suicide prevention course improved intended actions at 7 months. It tested a course taught in a shed, not the shed.
At the doorA date and how many came. It says that people came, and when, and claims nothing about why.
At the benchThe thing made, and the day it was finished.
Over a yearThe counts in order, nothing averaged.

None of the reviews read for this season reports a randomized trial of going to a shed. The men's own accounts of feeling better are evidence of what they say, and the season does not call the shed a treatment.

Use it: the free toolShed →The count at the door by date, and a dated list of things made. It asks no one how they feel.

Hudson 1851 · Wilson & Cordier 2013 · Milligan et al. 2016 · Kelly et al. 2019 · Foettinger et al. 2022 · Morgan et al. 2025 · Season 9

Card 11 · The Regular
Cost 2Evidence 2Hype 3

Same place, same hour, a few words

Nobody has run a trial of being a regular. Short experiments found that a few words with a barista or a stranger left people feeling better than silence did, measured on the day.

Named no one to discuss important matters with (national survey)

1985, as published10.0%
2004, as published24.6%
1985, corrected 20088%
2004, corrected 200823%

A colleague put the true 2004 figure "somewhere under 10 percent." A later pair traced many of the zeros to particular interviewers. The authors still maintain the trend. Nobody has settled it.

The train experiment (2014)

ExpectedSeparate participants "expected precisely the opposite outcome."
Reported"a more positive (and no less productive) experience"
1652St. Michael's Alley, Cornhill, "in or about" that year: a room that sold a hot black drink, at a sign over the door.
1675The king ordered the coffeehouses shut as "seminaries of sedition," and within days took it back.
2000Putnam's own book: "only poetic license authorizes my description of non-league bowling as 'bowling alone.'"
2015A poll of 2,254 British adults, paid for by a beer campaign: people without a local had smaller networks. One survey, taken once.
The placeThe same counter at the same hour.
The exchangeA smile, eye contact and a brief conversation, rather than being as efficient as possible.
The lineA date, a place and the name of the person spoken to, if anything is kept.

The short experiments measured feelings on the day. The pub survey's own path analysis runs from satisfaction with life toward pub visits, and it did not measure personality. No source in the season shows that the closing of third places caused loneliness.

Use it: the free toolRegular →Your place and hour, and a dated line with the name of whoever you spoke to. The thinnest record on the hub, on purpose.

Aubrey, via Timbs · Putnam 1995, 2000 · McPherson, Smith-Lovin & Brashears 2006 · Fischer 2009 · Paik & Sanchagrin 2013 · Sandstrom & Dunn 2014 · Epley & Schroeder 2014 · Season 11

Card 12 · The Address Book
Cost 2Evidence 2Hype 3

A list of names, and a date beside each

Henry Cole ordered a thousand cards in December 1843, and deciding who got one was the practice. The numbers that came later to explain the list turn, followed back, into something smaller than the version that traveled.

Group-size estimates for humans, one year of published exchange

Best-fit model, 202169.2
By Dunbar's method158
Same method, other data289.8
Monkeys and apes only404.1

The 69.2 came with an interval from 3.8 to 292.0. The authors: "This is not very informative."

Odds of being alive at follow-up (2010, 148 studies, 308,849 people)

All ties Complex measures Not living alone 1.01.502.0
1843Cole's diary, December 17: "In the Evg Horsley came & brought his design for Christmas Cards." He ordered a thousand.
1956The rotary card file patented that year was for one thing: "Any desired card in the stack may be brought into view."
1985A producer's wheel held eighteen hundred numbers and none of the reasons, and the reasons had to be rebuilt one phone call at a time.
1998In 223 of the study's men, much of the link between poor social supports and mortality "may be mediated by alcoholism, smoking and pre-morbid psychopathology."
The listGo through the names on purpose. Somebody has to decide who gets a card.
The lineWhen and how each person was reached: wrote, called or visited. It is the part a card does not keep.
No numberA figure that moves from 69.2 to 404.1 is not a rule for how many names a book should hold.

The 148 studies are evidence about who had ties, not about what a deliberate list does, and the authors write that causality is not easily established. In the study of 724 men, nobody was assigned a friend. Fifteen cigarettes was never a finding about cigarettes or loneliness.

Use it: the free toolNames →A list of names with the last date and how each person was reached. No cap and no target.

Cole diary 1843 · US Patent 2,731,966 (1956) · Dunbar 1993 · Lindenfors, Wartel & Lind 2021 · Holt-Lunstad, Smith & Layton 2010 · Vaillant et al. 1998 · Season 12

Part three · evidence 1

The room's own paperwork

No trial, and nothing else to lean on. What these rooms left instead is some of the best paperwork in the series: rules in a founder's hand, minutes, counts of who began and who finished. A thin record, and not a false one.

Card 1 · The Queries
Cost 3Evidence 1Hype 2

Ask the question a week before you argue it

Franklin's club met on Friday nights for something close to forty years, by his count. Its rules survive in his hand. Everything about its effect on the men in it comes from the one who did best.

What the club left, by date

1727The club forms, on Friday evenings, in an alehouse
1731Members draw up articles for a library: fifty subscribers at forty shillings
1732Twenty-four standing queries, in Franklin's draft, with a pause after each
1736The Union Fire Company, from a paper first read in the club
1765"adjourned during the warm and short Evenings"

The number that travels (2017): firms, not the room

Member companies28%
Average business12.5%

Compounded annual revenue growth, members still in business against an average. The footnote describes no controls.

1732The first query in print: "Have you met with any thing in the author you last read, remarkable, or suitable to be communicated to the Junto?"
Week"Our queries, which were read the week preceding their discussion, put us upon reading with attention upon the several subjects."
Rule"Direct contradiction" was "after some time made contraband," under "small pecuniary penalties." The sum is not given.
1784Franklin still avoided "certainly, undoubtedly," "agreeably to the old laws of our Junto."
The standing questionOne fixed question, answered before the meeting: what have you read that is worth bringing in?
The week betweenAnnounce a question at one meeting and argue it at the next. The answer is made in the week.
At the tableNo flat contradiction. Disagree without the fixed declaration, so the other person can still change their mind.

There is no trial of a room like this one. The growth figure counts firms that chose to join and were still standing. "First mastermind group" is not in Franklin, and it is not in either of Hill's two books.

Use it: the free toolQueries →The standing questions, next meeting's question with the date it was announced, and the rule against the flat contradiction.

Franklin, Autobiography (begun 1771) · Standing Queries for the Junto, about 1732 · Franklin and Roberts letters, 1758 to 1766 · Library Company of Philadelphia · Season 1

Card 2 · The Minute
Cost 3Evidence 1Hype 1

Write it in the room. Read it back before anyone leaves.

The Quaker minute is a very well kept record of one thing: what a room agreed to, on a date, in its own words. Nobody in the room ever claimed it makes better decisions.

The Germantown file: eighty-eight years, every step minuted

1688 protest;three meetingsminute "no" 1754 1758 1776 1778: firstdisowned eighty-eight years

1776: "we do with great unanimity give it as our sense & judgment" that meetings labor with slaveholders. The method was slow in exactly the way its own books warn it can be.

Three ways a minute can end

united"unity of spirit could not be reached"standing aside, with a reason
1656Balby's postscript: "these things we do not lay upon you as a rule or form to walk by." Britain's book of discipline prints it first.
1688Three meetings each minuted that they would not decide: "a thing of too great a weight for this meeting to determine."
1983The standard outside study relies on interviews with about one hundred and fifty members of one yearly meeting: "Quakers do not begin with a theory. They begin with an event."
1984"We are not seeking a consensus; we are seeking the will of God." The handbooks say in print that consensus is a different thing.
In the roomA few lines of plain prose, dated, drafted while the meeting waits, read aloud, changed and accepted before anyone leaves.
When it failsWrite that unity was not reached. A record that can say "we could not" is worth more than one that can only say "we did."
The questionAsk one whose answer you could not guess. It is the one rule in the season that travels without its setting.

I found no study that measured decisions made this way against decisions made by vote. Britain's book warns that putting a decision off "can be the 'lazy' option."

Use it: the free toolMinute →Date, who was present, the minute, and how it ended. It will not keep a minute that was not read back.

Balby epistle 1656 · Germantown protest 1688 · Woolman, Journal · Sheeran 1983 · Quaker faith & practice · Burton, Koning & Muers 2018 · Season 2

Card 3 · The Reading Circle
Cost 3Evidence 1Hype 2

Forty minutes at home, then the table

The first reading class enrolled 8,400 and graduated 1,718, and its founder printed both. Nobody measured whether the readers learned more than people who stayed home.

The class of 1882, on the founder's own count

Enrolled8,400
Received a diploma1,718

About one in five, over four years. The figure includes the people who left, which is why it is worth more than the larger ones.

Sweden, 2024: participants who say the circle supported their association work

30 percent, from a survey of participants: what people state, not what the circle did. In 2025 Sweden counted just over 92,000 circles and over 321,000 participants.

1878The plan: four years, "with forty minutes as each day's task, for nine or ten months of each year."
1882The first Recognition Day, August 12: about eight hundred in the procession, more than a thousand "detained at their homes."
1886A circle in Maine: "We have a constitution, and keep a record of each meeting... The president is a dressmaker."
Slogan"A study circle democracy": a word attributed to a prime minister, without a speech, a year or a page.
At homeThe same passage as everyone else on the list, before the meeting. Forty minutes, not an evening.
At the table"And then talk of what we read."
AfterA date, the pages, who came. Keep the honest count, including the people who stopped coming.

There is no trial of the circle, in either country. The half-million is an officer's round number about his own institution, and it sits beside 1,718 as a different kind of thing.

Use it: the free toolCircle →Next meeting's date and pages, a record of each meeting, and the count of who began beside who came last.

Vincent 1886 · Hurlbut 1921 · Folkbildningsrådet 2024, 2025 · Statistics Sweden · Season 3

Card 10 · The Council
Cost 3Evidence 1Hype 3

Write the decision down at the table

The family council has a founder and almost no data. The family dinner has a famous number, and the number turned out to be about which families sit down together.

Dinners and depressive symptoms, 17,977 teenagers (2012)

Association, no controls100
After family relationship controls39

Controls for family relationship quality reduced the association "by a total of 61%." Smaller, and still there.

When the family was held still

201221,400 children: "no significant (p<.05) relations" with school or behavior
20135,419 young people: dinners "not generally associated" with alcohol, cigarettes or starting drugs
2014"little benefit when parent-child relationships are weak"
1964A book for parents taught "holding family meetings or councils." It is the council's first appearance in American print that the season can date.
2011A phone survey of 1,006 teens: those with fewer than three family dinners a week were "Almost four times likelier to have used tobacco."
2012The next year's paper, by the same center: its data "cannot be used to establish causality."
2014Dinners "contribute to fewer depressive symptoms and less delinquency among adolescents when family relationships are strong."
WeeklyA meeting at home. Each family's council turns out to be its own.
At the tableEach decision as one written, dated line. It is the only thing a family meeting leaves that anyone can check the following week.
The dinnerKeep it as a room, not a cure: a place where a close family shows up.

No study supports the written family decision and none refutes it. The program that carried the council into parent groups has, by an independent review and a federal clearinghouse, no good evidence of changing how children behave.

Use it: the free toolCouncil →Last meeting's decisions shown first, then this meeting's, one dated line each. No chore chart and no points.

Dreikurs & Soltz 1964 · CASA 2011, 2012 · Musick & Meier 2012 · Miller, Waldfogel & Han 2012 · Hoffmann & Warnick 2013 · Meier & Musick 2014 · Season 10

Index

All twelve on one map

Across: what it costs to do. Up: how well the research backs it. Color: how far the popular version runs ahead. Tap a practice to jump to its card.

Nothing here is free, and the evidence rises as the room gets smaller

Hype gap: 1 small 2 moderate 3 large 4 extreme

Eight practices sit at cost 3 and four at cost 2; none is at cost 1, because every one needs at least one other person. None reaches evidence 4, and two of the three at evidence 3 are one person and one other person. Ratings are one reader's judgment on the series' own scales, argued row by row in the Volume Ten matrix. They don't rate how well anything works for you.