The absence of structure is the danger
Everything else you've done has had weekly pressure. A dissertation has none — so it competes, every single week, against things that are due on Friday, and it loses every single week, for entirely rational reasons. Then it is March.
Manufacture the deadlines nobody gives you
- A meeting with your supervisor every two weeks, booked in advance. This is the single highest-leverage thing on the page. An impending meeting where you'll have to show something is the only reliable weekly pressure that exists.
- Something to show at every meeting. Even something bad. Especially something bad — a bad draft can be corrected and an absent one cannot.
- Interim deadlines with real consequences. Tell someone. Write them down. A deadline you can silently move is not a deadline.
Write from day one, not after the reading
The classic and near-universal failure is: read for four months, then write for one. It doesn't work, for two reasons. First, reading has no stopping condition — there is always one more paper, and 'not enough reading yet' is the most respectable form of procrastination available.
Second, writing *is* thinking. You do not know what you think until you try to write it down, and if you defer that discovery to month five, you will discover in month five that your argument doesn't work. Write badly, early, and often — the bad draft is not a failure, it's the mechanism.
And on the reading
Use the three-pass method — most papers deserve five minutes. Keep the sources somewhere you can interrogate directly, because in month four you will remember that *someone* addressed your confounder and have no idea who, and that hunt can eat a day.
This is also the one context where an elaborate note-linking system genuinely earns its keep: connections between sources are the actual output here, which is not true of a taught module.