From a practical point of view, if a book has a linear narrative, is written in a single voice, these things improve the odds of completion: You are unlikely to run into structural problems, so if you return to the book after a long gap the only challenge is resuming the well-established voice. If it involves no research the odds are even better: It is not burdened with a mass of notes, once fresh to the mind, which must be gone over again before work can be resumed. The practical is not, of course, the only point of view.
This kind of book can offer a kind of formal satisfaction: The reader learns the rules of the game. Constraints can give it intensity, momentum, energy. A book with many voices offers more variety; it does offer a view into many possible worlds. It seemed to me, after the success of The Last Samurai, that I would be able to look for a publisher who could handle technically challenging books. I was interested in Edward Tufte's work on information design; I thought this could be used to present mathematical ways of understanding chance in fiction. There was a book about poker, Stolen Luck, that required collaboration with a designer; my publisher agreed to give me this, in a deal for SL and Lightning Rods in 2003, but then changed his mind after the contract was signed. So practical considerations do end up counting for more than they should.
— It's Good To Be Pragmatic: An Interview with Helen DeWitt