Attention Budget
Once a model can produce almost unlimited output, the binding constraint on what actually ships is the operator's finite attention — so attention is a budget to be managed, not an infinity to be spent. The source's framing: "the ability for AI tools to create productive output has gone nuclear, yet people's attention span has only gotten worse," and that divergence is the problem. This is the firehose thesis stated as a personal-workflow rule — "human attention is the bottleneck" — and it turns "how do I 10x output?" from a prompting question into an attention- allocation question.
The concept has two halves. The first is a division of labour by comparative advantage: AI is good at fast, continuous execution and pattern-matching at scale; the human is good at judgment, taste, and knowing when something is actually finished — so hand the model the execution and spend your scarce attention on the judging. (This is the same seam Cognitive Offload Cost watches from the other side: offload execution freely, but the judging is the part whose atrophy is a cost.) The second is a set of attention-conservation practices: schedule work by cognitive load (intellectually complex tasks in your freshest hours, repetitive work when depleted), consolidate every interrupt into a single hub so automations don't ping you in five directions, and cut input friction (voice-first capture, phone/remote access) so the mechanical cost of directing the model doesn't itself burn you out.
The honest boundary: the specific tactics (an AM/PM split, Slack-as-hub, a particular voice tool) are one operator's system offered from personal practice, not measured results — the durable part is the budget framing and the comparative-advantage split, not the exact schedule.
Claims
- When output is nearly free to produce, attention — not model capability — is the binding constraint on what ships; manage it as a budget. principle — durable: it is the general economics of a commoditizing input (the scarce complement becomes the constraint), applied to the operator's own hours rather than to strategy. Ties to Execution Commoditization (value moves off the cheap input) and is firehose's own operating premise.
- Divide labour by comparative advantage: AI takes fast continuous execution and pattern-matching; the human keeps judgment, taste, and knowing when something is finished. principle — durable: the split follows from what each side is actually good at, independent of tool or generation. It is the constructive half of Cognitive Offload Cost — offload the execution, keep exercising the judgment.
- Schedule intellectually complex work for your freshest hours and batch simple/repetitive work for when you're depleted. best practice — context: an individual with discretionary control over their day; the specific AM/PM split is one operator's system, not a measured optimum. The durable part is match task cognitive-load to your attention level, not the clock.
- Consolidate every automation and notification into one hub so tools don't fragment your attention. best practice — context: an operator running many concurrent automations/loops whose independent pings would each cost a context switch; the source routes everything through Slack webhooks with a channel per automation. Contingent on having a hub the whole workflow already lives in — the principle (minimize interrupt surface) survives the choice of tool.
- Cut input friction with voice-first capture and remote/phone access so directing the model doesn't itself deplete you. best practice — context: heavy daily AI use where the mechanical cost of typing/being-at-your-desk is a real drain; offered as personal practice (Whisper Flow / Hex / Claude "dispatch"), not evaluation. The durable idea is that the interface to the model is part of the attention budget, not free.
Related
- Cognitive Offload Cost — the mirror concern: this page says offload execution to reclaim attention for judging; that page warns that offloading the judging itself is where the cost lands. Same execution-vs-judgment seam, opposite side.
- Execution Commoditization — the market-level version of the same move: when execution gets cheap, value (and here, attention) relocates onto the scarce complement.
- Loop Engineering — the automation half of reclaiming attention: designing loops that remove you as the trigger is how execution gets handed off so attention can go to the boundaries.
- AI Completion Asymptote — where the reclaimed attention should be spent: on the last-20% polish AI can't reach, not on re-prompting the 80% it already delivers.
- Imagination Constraint — the adjacent binding-human-input claim: attention is the constraint on doing, task-selection is the constraint on what's worth doing.
- Agent-Shape Triage — a pre-flight estimate that protects the budget: deciding whether a task even warrants an agent (including "no AI at all") is attention-allocation before the fact.
- Distillate: You're the Problem, Not Claude — Six Fixes to 10x Output