attention is a capital allocation problem

imagine a cfo who ran your money the way you run your attention. no budget. no review. positions opened by whoever shouts loudest that morning. fifty simultaneous holdings, most of them opened by other people, none of them sized deliberately. everything on margin.

you'd fire them before lunch. yet that's the operating standard for the founder's actual scarcest resource.

money can be raised. headcount can be hired. compute can be rented. attention is the one form of capital with a hard cap that's identical for everyone: you get the same waking hours as your biggest competitor, and no term sheet can change that. which means, at the top of the game, allocation is close to the only edge left.

so run it like a portfolio. the metaphor sounds cute until you take it literally, and then it starts making decisions for you.

attention has worse properties than money

before the allocation rules, respect the asset class. attention is capital with three brutal quirks money doesn't have.

it pays a switching tax. money moves between positions for near-zero cost. attention doesn't: every context switch charges a re-immersion fee, the twenty-plus minutes it takes to get truly back inside a hard problem. a day of task-hopping can be a day of paying fees on trades that never settled.

it doesn't roll over. unallocated money waits patiently. unallocated attention expires nightly, at whatever quality your energy allowed. there is no cash pile of unused tuesdays.

and it trades against adversarial counterparties. this is the one people underweight. the feed is not a neutral venue. it's a counterparty with its own book, engineered to buy your attention below market and sell it onward to advertisers. every notification is an unsolicited trade request, and you've set your account to auto-accept. you wouldn't let a stranger execute trades from your brokerage account. your phone does it forty times a day.

the audit

every allocator starts with the books, so track a week of hours the way you'd track expenses. crude buckets are fine. most people who do this find the same thing: the majority of the portfolio is in reactive positions nobody chose. meetings that are margin calls issued by someone else's calendar. slack threads that are micro-positions opened by whoever typed your name. a "strategy" bucket that turns out to hold ninety minutes. for the whole week.

you cannot fix an allocation you've never seen. most founders are managing the fund blind and calling the resulting drift "being busy."

the allocation rules

the useful part of the metaphor is that investing already solved these problems, and the solutions translate almost one to one.

concentrate. returns follow concentration. the founders who look inexplicably productive mostly hold one oversized position: the single thing that compounds this quarter, funded like they mean it. a portfolio of fifteen equal-weight priorities is an index fund of your own todo list, and it will return the average, which in startups is zero.

size by expected value, not urgency. urgency is the market screaming. expected value is the actual math. most calendars are sized by scream.

hold illiquid blocks. deep work is locked-up capital: multi-hour blocks you cannot exit early, which is exactly why they earn the illiquidity premium. a calendar with no locked blocks is a fund holding only cash equivalents, perfectly liquid, perfectly responsive, returning nothing.

keep a cash buffer. leave genuine slack, unallocated hours, or every surprise forces a fire sale of your best position. the buffer is what makes concentration survivable.

rebalance on schedule. a weekly review, fifteen minutes: where am i overweight? what position has been open for a month with no return? drift is the default state of every portfolio, attention included. rebalancing is the countermeasure.

and learn to sell. every yes is a position, and recurring commitments are the dangerous kind: subscriptions sold against your own book, quietly compounding. the discipline that separates real allocators is exits. audit your standing meetings quarterly like holdings. most survive on inertia, not thesis.

the ai complication

agents were supposed to solve this, and they've made it stranger. delegating execution frees attention; i run my whole workflow on it. but agents also explode the number of things you can credibly attempt, and abundance of doable things makes allocation harder, not easier. i've written about wanting being the bottleneck; this is its daily, hourly form. delegation without a portfolio view doesn't concentrate your attention. it just multiplies open positions until you're a fund with a thousand tiny holdings and no thesis.

agents pay off when they're pointed at positions you chose.

the serendipity argument

"founders have to be reactive. the market doesn't schedule its surprises, and the locked-calendar guys miss the serendipity that makes startups."

conceded, and the portfolio answer is better than the hustle answer: make exploration an explicit position. size it deliberately, ten, twenty percent, whatever your stage demands, and spend it on genuine wandering: odd conversations, rabbit holes, unstructured time with users. that's the venture book inside the fund, and it's where asymmetric returns hide. the difference between exploration and distraction is a single fact: whether you chose it, in advance, at a size you meant.

reactive by choice is a strategy. reactive by default is just being someone else's portfolio.

you've been the allocator every day since you started building. the allocations are happening either way, hour by hour, to whatever shouts loudest.