I can’t say how useful having a hypothesis is in shaping your writing and research. It helps you to shape your research proposal, your research plan, your first draft. It may be that your hypothesis is wrong, that’s fine. It’s easier to fix a wrong hypothesis than to introduce an argument into a draft that has none.  This is because drafts that describe your research journey are exploratory writing, whereas drafts that describe your findings and analysis are explanatory writing.

Scientific writing always has a hypothesis, and this is one of the reasons science theses are, on the whole, faster to write than theses where the hypothesis can only emerge very late in the process like anthropology or projects using grounded theory. But most humanities and creative projects can be written using a hypothesis, and it’s worth doing so.

Surprisingly, it is MUCH easier to rewrite a text that had set off in a definitive direction but the direction turned out to be wrong, than to rewrite a text that had no direction at all. You may need to change words, but it’s actually quite easy, fast work. Why is it easier to rewrite an argument that was wrong rather than to construct an argument for the first time? Because arguments are linear, and descriptive research is not.

When you have piles of articles, books, field notes, discussions with peers and supervisors, undergraduate text books, potential theories… it’s all messy and networked and multivalent and full of potential. Describing this research will also be messy, distributed and full of potential lines of enquiry. However, a thesis is just that, a single hypothesis that you explore in a linear manner across 70,000+ words. (Or, for some anthology PhDs, 3-5 articles, each of which takes a single hypothesis that you explore in a linear manner across 7,000 words).

To create a linear argument out of a mass/mess of information involves an enormous amount of work of constructing knowledge. It is really hard to turn the words of a description into an argument—you are often better off starting again with a blank page (though using your existing research, obviously!) than trying to wrestle that prose into a new shape. In cases like this, typing up 20,000 new words is surprisingly easier and faster than reworking your existing 20,000 words into a structure.

On the other hand, to update a linear argument involves just a bit of shifting. And that shift can be successfully achieved with quite small changes in your language, perhaps by using modifying words like ‘partially’ or ‘in only two out of the five cases’, or even ‘not’.

For example, you might start with something like:

Scholars have agreed for about the last 60 years that Middle Earth kingship was contingent on the support of the Silvan Elves (Baumgarden 1952, Schwartz 1992, Allan 2007, Ringwald 2014).

This close analysis of 7 manuscripts from the Kloster Anduin written between 1300 and 1400 will illustrate the ways in which the royal family of Naith maintained temporal power by courting and relying upon the institutions of the Silvan Elves.

Except you got to the archive, and all the manuscripts are actually letters between the prince and the elf lord quarrelling about money and lands. You’ll need to rethink your hypothesis, but you’ll find it extraordinarily easy to rewrite any draft introductions.

While scholars have agreed for about the last 60 years that Middle Earth kingship was contingent on the support of the Silvan Elves (Baumgarden 1952, Schwartz 1992, Allan 2007, Ringwald 2014), this close analysis of 12 manuscripts from the Kloster Anduin written between 1300 and 1400 will illustrate the ways in which the princes of Naith maintained temporal power in spite of extensive opposition from the institutions of the Silvan Elves.

Two tiny tweaks and a change of punctuation, and my argument now says something completely different. And the tweaks make all my wrong ideas completely disappear.

The reader doesn’t need to know that I thought I was going to find something, but when I got to the archive I couldn’t find it, and then I spent a fortnight walking through the woods totally lost and confused and panicking, and then I talked to my supervisor, and then I talked to my other supervisor, and then I wrote three drafts of the new introduction none of which were any good and then… [You are bored by now, and I can promise you, your examiner will be too.]

The thesis is a map to the best route to the destination of your new knowledge, not a travelogue of how you got there.

PS: I made up everything about those sentences about Middle Earth research—so sorry to everyone who is a more serious Tolkienite than I am and was wincing (or wondering who Baumgarten was).