Stop Managing Time. Manage Attention.

Almost every popular productivity system is a time system. Blocks on a calendar, timers, estimates in hours. They share an assumption that an hour is an hour, and anyone who has tried to do demanding analytical work at five in the afternoon after six meetings knows that assumption is false.

Sort work by the state it requires

A more useful classification has three tiers. Deep work requires uninterrupted focus and full context: architecture, difficult debugging, writing something that has to be right. Shallow work requires attention but tolerates interruption: reviews, email, routine implementation. Mechanical work requires almost nothing: expenses, file organisation, scheduling.

Once work is sorted this way, the scheduling rule is obvious and rarely followed: match the tier to your actual state rather than to the gap in your calendar. Most people's best two hours go to email because email arrives first.

Protect the peak, spend the trough

Almost everyone has a reliable window where difficult thinking is easiest, and for most people it is earlier than their calendar suggests. Identify yours over a week of honest observation, then defend it. Everything mechanical can move to the trough, where it will take slightly longer and cost nothing important.

The most expensive scheduling mistake is spending peak attention on work that would have been fine at three in the afternoon.

Context switching is the hidden cost

The damage from an interruption is not the duration of the interruption. It is the reconstruction of mental state afterwards, which for genuinely complex work can take substantially longer than the interruption itself. This is why a day with four scattered meetings can produce almost no deep work despite having several free hours.

• Batch by type. All reviews together, all messages in two windows, all administrative work in one block.
• Leave a breadcrumb. When stopping mid-task, write the next concrete step. Restarting from a written instruction is dramatically faster than restarting from memory.
• Defend contiguity over quantity. Three uninterrupted hours beats five fragmented ones for anything hard.

The honest limit

Deep work has a ceiling, usually somewhere between three and five hours a day for sustained periods. Systems that assume eight productive hours produce guilt rather than output. Planning for four good hours and treating the rest as shallow and mechanical is not lowering the bar; it is describing what already happens and arranging it deliberately.