Every published text (academic paper, story, novel, poem) is built on another text that is buried underneath, an abandoned text. And this is useless in a way, but it’s a sacrifice we must make; papers cannot be created without destroying in the process ideas, abandoning possible paths. The paper that never sees the light of publication such that the actual paper emerges is the equivalent of “kill your darlings”. and it has to be like this, because writing as thinking, writing for creativity is incredibly inefficient and roundabout.
For example, this is how I work:
a) optimism phase: I have an idea for a paper. I write it down, then sketch a possible outline. the paper seems doable, reachable, not a lot of work
b) i start reading around this idea. i immerse myself in the literature and realize that – my idea is not novel – it’s too coarse, needs refinement – and many others wrote about it in more interesting ways.
c) i think while re-outlining the initial idea into a more realistic outline, that pays tribute to what is out there, published. this is outline 2. Same idea, but contextualized. good enough, still looks doable. Still in the optimism phase.
d) i start writing following the outline. as I put words into phrases, lots of cracks and gaps appear that need to be filled. and not with bullshit. the passage from one idea to another requires serious work. so i have to read more, look even further in what is out there. this is the despair stage. At this stage, the author has two choices: either they continue with the initial outline, and mask the gaps (logical, conceptual gaps) with bullshit words (LLMs are happy to help here, they offer ammunition for every possible idea and angle, blind words to fill pages leading to any desired conclusion. LLMs are guns for hire: you provide the idea, they execute.) or ignore them, or they abandon the initial idea as uneasible and look for inspiration in the friction of writing. As we write, adjacent ideas emerge, and they are cool, we cannot chase all of them, but they appear. We just need to pick one, a branch that springs from the initial idea. This is the secondary idea, emerging only while writing and struggling to cover the conceptual gaps with words.
e) I write the other paper, the secondary one, because this one looks more feasible now. I send it to the journal.
f) The peer reviewers come back with a lot of concerns and new gaps. at that point, i could buckle down and defend my second idea, and cover all objections with some bullshit, answers to reviewers: I am right, they are wrong and here is why. Stomp my feet in the ground, sort of. But I don’t. Most of my life I was lucky to have good peer reviewers who genuinely helped me think further. I take their insights and run further with them. This changes the paper substantially. I now want to argue for an idea adjactent to the one I had submitted. But this one is more organic and obvious now, yet it was hidden under layers of text. Once I need to respond to reviewers, this idea jumps up. This is what actually I wanted to argue for, this! This is the one.
g) this is the revelation moment when I actually find the paper I wanted to write all along, buried under outlines and changes of mind. This one is steady and i know it. It’s a feeling of having come home. I rewrite the draft completely and send it out to the journal. From my perspective, the paper is done, because the idea has been fixed by a revelation.
Looking at all this inefficient process, I could have written 3 different papers along the way – at least – and working on any of these seems a waste of time. But I needed to do all this digging in the trenches of text to find what I really wanted to say. So in this roundabout way, I do find my voice and my standpoint. All this process would have been killed if I had co-opted LLMs to help ghostwrite this and give me ammunition for carrying out the initial project. LLMs help us only get stuck in the first insight, and not even the most interesting one, and hence we write predictable papers.
I need this inefficient and wasteful process to kill papers along the way so i can find that one worth writing. And the less drafts I produce, the worse the final paper becomes. Can this process generalise to other writers? I suspect it does. I don’t think there is anyone out there who just produces text, straight out of their thoughts, and likes it as it lands on the page. Academia has trained us to be critical and self-critical, we are bound to not like what we write. it is in the urge to write better that ideas get produced. Also known as friction.
There is no easy way of writing a text for others’ eyes to see. And if we agree on this point, LLMs as writing aids will kill our ideas.
Leave a Reply