Browse the library — page 8

An alphabetical directory of published knowledge, collections and longreads. Every page works without JavaScript.

1154 entries · Page 8 of 12

• Put the buffer before the assessment and protect sleep opportunityprotocol
• Put the caveat beside the decision it can changeprinciple
• Put the explanation beside the thing it explainsprinciple
• Put the real build and validation path in repository instructionsprotocol
• Put the slow external dependency on the forecastchecklist
• Put uncertainty around consequential eval differencesprinciple
• Put variation beside the averagequestion
• Question the shortcut behind a judgmentcollection
• Rank assumptions by decision leverageprotocol
• Rank the issues before you start concedingchecklist
• Re-benchmark AI when the model or task changesprotocol
• Re-decide continuation from todaycollection
• Re-estimate when evidence changes, not when anxiety spikesprotocol
• Re-test old harness rules when the model changesprotocol
• Read a confidence interval as a procedure with assumptionsconcept
• Read a percentile as a rank, not a scoreconcept
• Read eval transcripts before trusting the scoreprotocol
• Read expected value as a weighted average of possibilitiesconcept
• Reappraise arousal as activation, not a verdictconcept
• Receive feedback in two passes: signal first, verdict laterprotocol
• Recheck the bottleneck after improvementprotocol
• Recheck the safeguard after the workflow changesprotocol
• Reconcile what changed after a bulk writeprotocol
• Reconcile what is missing and what should never have arrivedprotocol
• Reconfirm the goal when new information changes the meetingprinciple
• Record confidence before feedback, then score it against the outcomeprotocol
• Record the episode while its context is still availableprotocol
• Record the outcome and your contribution while the evidence is freshprotocol
• Record your answer before reading consequential AI adviceprotocol
• Recover from a missed session by replanning, not doublingprotocol
• Red-team the plan from the other side of the systemprotocol
• Redesign a repeat slip before scheduling more trainingprinciple
• Redesign one job resource before adding more workprotocol
• Reduce batch size when feedback delay dominatesprotocol
• Reduce judgment noise without erasing useful disagreementcollection
• Reduce simultaneous complexity when overload appearsprotocol
• Reflect on a concrete nonwork goal when work keeps reclaiming the eveningprotocol
• Refresh a dormant professional tie before making the requestprotocol
• Rehearse self-presentation against a real role requirementprotocol
• Rehearse the cue–response plan onceheuristic
• Reject the idea without closing the channelprotocol
• Reject the suggestion without closing the channelcollection
• Relearn to success on more than one dayprotocol
• Remove household hazards before they become an emergencycollection
• Remove one hindering demand before adding another learning loadheuristic
• Remove the alert, not only the urge to answer itprotocol
• Remove the interesting detail that teaches nothingquestion
• Remove worked steps graduallyprotocol
• Repair the meaning gap instead of pretending you understoodprotocol
• Repeat a high-risk concern when the first challenge is ignoredprotocol
• Repeat the behavior in a stable contexttemplate
• Repeat the message, not a memorised scriptprotocol
• Repeat the task, then change the taskprotocol
• Reperform after feedback instead of collecting correctionsprotocol
• Replace quality adjectives with observable conditionsprotocol
• Replay from the failure checkpoint, not from memoryprotocol
• Replenish from a real downstream signalprinciple
• Report how much work an abstaining system actually coverschecklist
• Require a human decision before privileged workflow executionprotocol
• Require an active choice when no safe default fits most peopleprinciple
• Rescue high-value language from input into deliberate practiceprotocol
• Reset a long thread with a current-state blockprotocol
• Resolve the contradiction instead of polishing the compromisecollection
• Respond to good news so the other person can keep the momentcollection
• Restore help when errors stop teaching you anythingprotocol
• Resume after a missed habit without inventing a debtprotocol
• Resume after an interruptionlongread
• Resume from the saved state before rereading everythingheuristic
• Resume the right task and remember the future actioncollection
• Resume work after an interruptionlongread
• Retire a goal explicitly when its case has changedprotocol
• Retrieve code before sending codeprotocol
• Retrieve context just in time instead of preloading the whole projectprinciple
• Retrieve the decision rule, not only the answerprotocol
• Retrieve tool schemas just in timeprotocol
• Retry after a state change; replan after the same failureprinciple
• Retry slower, with randomness, and stopprotocol
• Return after a delay before adding another layerprotocol
• Return credit with the ownershipprinciple
• Return medicines to locked storage after each usechecklist
• Reuse the operation identity when retrying the same intentprotocol
• Reverse the business effect without restoring stale historyquestion
• Review a risky success, not only a visible failureprinciple
• Review exception debt as a portfolioprotocol
• Review goal progress before changing the goalprotocol
• Review test changes independently from the code they excuseprotocol
• Review the dominant focus on a trigger, not on every moodprotocol
• Review the file list before reading the agent's explanationprotocol
• Review the recording with timestamps and a tiny rubricprotocol
• Review whether team learning changed the next cyclechecklist
• Reward early problem surfacing with actionprinciple
• Rewrite from the feedback, not from the corrected sentenceprotocol
• Rotate an exposed credential even after the text is removedprinciple
• Rotate ownership where complexity benefits from shared leadershipprinciple
• Round for reporting, not by accident during calculationheuristic
• Route alerts by urgency, not one undifferentiated channelprotocol
• Route cases using both human and AI confidence only after both are calibratedprinciple
• Route team knowledge with a living expertise mapcollection
• Route to the cheapest model that still passes your task evalprotocol
• Run a bounded mobile-internet block when constant phone connectivity is the suspected problemtemplate

Inspect the public connection graph. Connections explain how pages and sources relate; they are not quality scores.