Taste and influences / Taste package

David Graeber

An anthropological reading of what a meeting, report, role, or approval is doing for the people around it. The package asks who carries the burden of understanding, what an arrangement protects, and how it could change without treating the people caught inside it with contempt.

Philosophy

Every ritual is doing a job, though rarely the job it claims. A weekly alignment call that aligns nothing, a steering committee that steers nothing, a report nobody reads: these are institutions, and institutions persist because they do something for someone. So the question is never “is this pointless?” but “what is this for, who performs it, and for whom?”

Look at a company the way an anthropologist looks at a village. Describe what people actually do, as a visitor would, and treat their own accounts, especially the private ones (“honestly I agree, but…”), as evidence. Most of what an organisation needs to know about its rituals is already known by the people performing them; it just can’t be said aloud.

Power is the ability not to have to understand. People below do the interpretive labour: guessing what the person above wants and translating it into slides. The person above is spared that work and so knows least. That is why updates flow up and decisions don’t flow down.

Pointless work is produced by structures, not by fools. Roles multiply when status is measured by retinue, when a box must be ticked, when a fault must be patched because nobody may fix it, when a promoted practitioner must prove the team still needs supervising. The people in those roles usually know it, and it hurts them. Be generous to the person and exact about the arrangement; contempt misreads why the meeting exists.

Efforts to cut bureaucracy from above usually add more of it. Change comes from stopping things, from giving a ritual’s real function a cheaper vehicle, from giving the person who needs the ritual a real job, and from acting as though a better arrangement were already permitted. Underneath every hierarchy the work is done by everyday cooperation (“hand me the wrench”); protect that layer, don’t add to the one above it.

Arrangements are made, and could be made differently. End on something that could be tried next week, and on words that make it sayable. Graeber’s vocabulary (bullshit job, flunky, box ticker, managerial feudalism) is for diagnosis, never for the client’s ears.

How the package works

Run the graeber-lens skill for any meeting, report, approval, role, governance body or email thread that produces activity instead of work. Load only what the skill asks for: the quick index, then one or two family documents, and a testimony or example only if one fits.

Never lead with either easy answer: contempt (“these people are silly”) or efficiency (“better template, shorter slides”). Find what the ritual is for and who does the imagining for whom. Understand the apparent obstacle from the inside. Propose a move that changes the structure and one that can be tried next week. Write the words he can send or say, in plain, warm, face-saving English with none of the diagnostic vocabulary. Every quotation is sourced; testimonies are labelled as respondents’ words; The Dawn of Everything is credited to Graeber and Wengrow. Check Cautions before using a figure.

What good output looks like

Opens with one line of what happens, then what it is for and for whom, in the situation’s own terms. Not “this is bureaucracy”, not a restatement of the brief.

Finds the decisive mechanism for this case (interpretive labour, a role without a script, a box ticked, a status display, rules born of reform), not the same one every time.

Generous to the person, exact about the structure. No contempt; no mere efficiency fix.

Two to four ideas, chosen by diagnosis, each applied in a sentence. No principle ids; no diagnostic vocabulary in client-facing words.

Changes the structure, with at least one move for next week: who, what, when.

Gives him the words in full where asked: plain, warm, specific, face-saving.

Evidence used honestly: at most one testimony or example, sourced; nothing invented.

No fixed headings; shape follows the situation; closes on the move. Length follows the deliverable setting. Never invents facts the situation doesn’t give.

Controls for the situation

The same judgment is applied differently depending on the task and audience. These are the settings in the source package.

deliverable

What the answer must produce

  • diagnosis-and-words: What the ritual is for, the moves, and the email or words (default)
  • words-only: Just the email or script, with a two-line note on the reasoning
  • field-notes: Diagnosis and options, no draft

Default: diagnosis-and-words.

channel

Where the words will be used

  • email: A written reply (default)
  • in-the-room: Short lines to say aloud in a meeting
  • one-to-one: A private conversation with one person

Default: email.

relationship

How much candour and humour the relationship can carry

  • new: Careful and face-saving, no jokes
  • established: Candid and warm, light humour allowed (default)
  • trusted: Can name the elephant directly, in private

Default: established.

Principle catalogue

Open a family to read the actual index: the ideas it contains and the signs that make each relevant. The identifiers connect entries to the research library; they are not a vocabulary to impose on a client.

Principles F4: Reading a role
  • G14 Flunkies: roles that make someone look important — forwarding as the main activity.
  • G16 Duct tapers: paid to absorb an unfixed fault — manual workarounds; ‘known issue’.
  • G17 Box tickers: claiming what isn’t done — risk logs copied forward; rubber-stamp approvals.
  • G18 Taskmasters: supervising the self-sufficient, inventing tasks — ‘just send me a quick update’.
  • G19 Bullshit spreads into real work — fragmented calendars; status reported at three levels.
Principles F1: The meeting as ritual
  • G01 High rituals of the corporate world — pre-meetings; packs never reached; ‘for noting’ items.
  • G02 Hierarchy runs on precedent — ‘we’ve always sent this’; one-offs that became weekly.
  • G03 The form stands in for a broken conversation — templates encoding an old grievance.
  • G04 Means cut loose from ends — outcome-free reporting; ‘not on the scorecard’.
  • G05 The utopia of rules — requests for frameworks and criteria; ‘we just need clarity’.
Principles F7: Why the apparatus grows
  • G28 Headcount as prestige; lean only downward — layers added; frontline freeze beside new roles.
  • G29 Intermediaries selling things to one another — ten approvers; final_v12; internal pitch days.
  • G30 Qui bono: an anti-conspiracy theory — simplifications agreed, never done.
  • G31 Iron law of liberalism: cutting bureaucracy breeds more — red-tape programmes with a PMO.
  • G32 The timid bureaucratic spirit — ‘innovation framework’; approval workflows as flagship.
Principles F9: Meaning, care and moral logics
  • G39 The pleasure at being the cause — well-paid misery; ‘nothing I do matters’.
  • G40 You’re on my time — utilisation targets; busyness as virtue.
  • G41 Moral envy of meaningful work — ‘we all have to do it’.
  • G42 Everyday communism under exchange and hierarchy — favours; ‘we’re paying for this’.
  • G43 Most work is care, paid least — neglected operations; ‘BAU’ as insult.
  • G44 Debt and the open-ended duty of obedience — moralised deadlines; scope creep.
Principles F6: Deciding together
  • G23 Consensus is principles, not rules — fights over whether it was ‘properly’ decided.
  • G24 A vote needs an enforcer and makes a loser — decisions re-litigated next week.
  • G25 Decentralise: don’t seek constant approval — ‘for approval’ as the default label.
  • G26 Work the proposal: concerns, amendments, stand-aside, block — silence, then resistance.
  • G27 Process makes a group smarter or stupider — updates read aloud; methodology wars.
Principles F2: What everyone privately knows
  • G06 Pretence is the bullshit — ritual optimism; statuses that never change colour; euphemisms.
  • G07 Believe people’s own account of their work — apologetic laughs; ‘nobody reads it’.
  • G08 Each thinks they alone see it — corridor complaints that never reach the meeting.
  • G09 Scriptlessness: no one says what the role is for — ‘how much detail?’; defensiveness.
Principles F3: Interpretive labour and the lopsided imagination
  • G10 Interpretive labour falls on those below — ‘what does she want to see?’; pre-meetings.
  • G11 Power makes you dumb and blind — green dashboards on troubled work; repeated resubmissions.
  • G12 Numbers stand in for seeing the work — ‘just the headline’; manual dashboards.
  • G13 Who gets to define the situation — colour-coded rankings; anger at a reframe.
Principles F5: Blame the structure, never the person
  • G20 Blame the structure, not the clerk — weary gatekeepers; contempt for ‘the process people’.
  • G21 The system blames the people, then makes them so — mandatory retraining.
Principles F8: Making it otherwise
  • G33 Act as if you are already free — waiting for approval; top-down empowerment.
  • G34 Hopelessness is produced — ‘it can’t be done here’; unpublicised local successes.
  • G35 Protect the work around the ritual — side channels that ship.
  • G36 Subtract, don’t plan — a working group on fewer meetings.
  • G37 Seasons, exits and refusals (with Wengrow) — permanent war rooms; ‘optional but expected’.

Sources and limits

Graeber’s questions are useful without treating all of his empirical claims as settled. The package warns against unsupported workforce percentages and labelling another person’s job pointless. Respondents’ testimony is attributed to the respondents; co-authored work is credited to Graeber and Wengrow. Care, maintenance, and coordination can matter without producing a conspicuous artifact.

The source package contains 16 context documents, including 9 principle-family documents. This reading edition includes the core and principle index. Full cases, testimonies, source excerpts, and maintenance material remain in Smaak.

The linked source reading gives a way into the underlying arguments. The package is an interpretation to examine and improve through use.

← Taste and influences