Browse the library — page 5

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

1154 entries · Page 5 of 12

• Join historical work to the version that was valid thenprotocol
• Judge a productivity system by protected outcomes, not calendar densitychecklist
• Judge AI confidence by whether it separates right from wrongprinciple
• Judge readiness after the answer has fadedheuristic
• Judge the expensive error separately from the common onetemplate
• Judge the next tranche, not the money already goneprotocol
• Keep a bounded payment fallback for short outagesheuristic
• Keep a clear expertise spine before widening your public topicsheuristic
• Keep a fresh dataset for the discovery's next testprotocol
• Keep a human baseline for important AI tasksprotocol
• Keep a hypothesis–test–result ledgertemplate
• Keep a list of high-confidence AI failuresprotocol
• Keep a map of who knows whatprotocol
• Keep a portable vaccine record and review gaps locallychecklist
• Keep a proportion's value separate from its displaychecklist
• Keep a return list instead of switching nowprotocol
• Keep a secure copy outside the place the original can failprotocol
• Keep acceptance examples out of the prompt workshopprotocol
• Keep agent code changes behind a branch and pull requestprotocol
• Keep an eval set that can embarrass the agentprotocol
• Keep an explicit bypass for truly urgent signalschecklist
• Keep assessment preparation executable under pressurecollection
• Keep CI agents away from production secrets they do not needprinciple
• Keep creation, transformation and claim layers separateconcept
• Keep critical service contact details outside the service they are meant to recoverprotocol
• Keep debriefs close enough to preserve evidenceheuristic
• Keep difficulty in the diagnosable zoneheuristic
• Keep every rerun result when the failure is intermittentprotocol
• Keep evidence certainty attached to its outcomeconcept
• Keep extensive and intensive reading as different modesheuristic
• Keep fluoride in the brushing routine instead of rinsing it straight awaychecklist
• Keep headroom for variability and recoveryprinciple
• Keep knowledge so you can find, trust and reuse it latercollection
• Keep mitigation work out of the hidden bufferprinciple
• Keep one cold benchmarkprotocol
• Keep one inventory of the critical documents your household depends onchecklist
• Keep one option-opening move in the career planheuristic
• Keep one portfolio artifact that does not belong to your employerprotocol
• Keep one review point that does not become micromanagementprotocol
• Keep one spontaneous speaking sample every cycleprotocol
• Keep partial terms provisional until the package worksprinciple
• Keep priority pressure separate from feasibility evidenceprinciple
• Keep raw evidence behind AI memory summariesprotocol
• Keep reversible experiments behind a boundaryprotocol
• Keep secrets out of the system promptchecklist
• Keep several issues alive long enough to tradeprinciple
• Keep source, locator and access date beside a material claimprotocol
• Keep task disagreement from becoming identity conflictheuristic
• Keep task stewardship when AI does the middlechecklist
• Keep the break free of the same cognitive demandheuristic
• Keep the condition attached to a p-valueconcept
• Keep the cue steady while the action becomes familiarprotocol
• Keep the durable artifact separate from the distribution postprotocol
• Keep the exact query that produced the evidenceprotocol
• Keep the field’s identity after the user starts typingprotocol
• Keep the headers attached when the table leaves your screenprotocol
• Keep the ingredients of a ratiochecklist
• Keep the meaning intact when data movescollection
• Keep the old decision when you replace ittemplate
• Keep the unit attached when moving the numberprotocol
• Keep useful work flowing when demand exceeds capacitycollection
• Label the message mode: decision, action, blocker or FYIheuristic
• Label the proposal's status before someone relies on itchecklist
• Language that survives a new situationcollection
• Layer safeguards so one miss is not the last chanceprinciple
• Lead with the decision-relevant factheuristic
• Lead without formal authority: stakeholder decisions, delegation, voice, and team learningcollection
• Learn a word with the company it keepsheuristic
• Learn and explain without hiding the difficult partcollection
• Learn by preparing to teach, then face real questionscollection
• Learn new skills, turn them into career evidence, and protect the main development edgecollection
• Learn the usable package, not only the translationprotocol
• Leave a quiet interval after learningheuristic
• Leave room for the learner to repair the sentenceprotocol
• Let a colleague rehearse the runbook without your missing contextprotocol
• Let AI do useful work without giving it accidental authoritycollection
• Let coding agents move fast inside a reviewable security boundarycollection
• Let ranges narrow only when uncertainty is actually removedprinciple
• Let reliability data slow feature velocity when the budget is spentprinciple
• Let retrieved text supply evidence, not new authoritychecklist
• Let subagents write durable artifacts instead of relaying everything through the coordinatorprotocol
• Let the change bake before stacking the next oneheuristic
• Let the model propose; let deterministic code commitprinciple
• Let the other account exist before you rebut itprotocol
• Let the units check the formulaprotocol
• Let users group the content before naming the categoriesprotocol
• Let voice commands use the label people can seechecklist
• Limit parallel starts when completion is the goalprotocol
• Limit weekly goals enough that progress can actually be monitoredprinciple
• List the load-bearing assumptions before adding confidencechecklist
• Listen for the point, the feeling and what mattersquestion
• Listen once before opening the transcriptprotocol
• Log what the agent did, not only what it saidchecklist
• Look at the problem through nine windowsprotocol
• Look for incentives that reward the failure modeprotocol
• Look for policy resistance before scalingchecklist
• Maintain a capability profile outside the résumé formatprotocol
• Maintain a high-use tool before replacing itprotocol
• Maintain a local forecast-error baselineprotocol
• Make a commitment device earn its frictiontemplate

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