I'm writing a difficult paper for my PhD. My first draft is rambling and long, so I'm cutting it down for the next round of edits.

However, I find it difficult to discard old writing. Some of the paragraphs I'm discarding may still be valuable for my future PhD, so I put them in 'discard' documents which I may use for my final dissertation document, or try to fit them into drafts for future papers.

What is your strategy for discarding old writing? What's the most efficient way to utilise discards, so you don't have to rewrite the same paragraph in future?

  • 5
    Is it difficult to discard because of the effort you put in, or because it contains information/ideas that is left out of the final version that you might want in the future?
    – Kimball
    Commented Oct 31, 2022 at 0:29
  • 2
    A bit of both I think!
    – Piethon
    Commented Oct 31, 2022 at 2:31
  • 2
    We usually have a separate document where we put the discarded paragraphs - this way you still have them when writing something else.
    – aqua
    Commented Oct 31, 2022 at 6:53
  • 1
    Science: you will have to rewrite again and again and again the same thing. And everytime you will not be able to reuse your previous text because of some small changes in the context. Be prepared. Bonus point: be prepared also for the consequences ... what you write again and again as a PhD will be read as "innovative", as an assistant professor will be read as "state of the art", as full professor will be read "oh gosh, not that old pile of crap again" ;D
    – EarlGrey
    Commented Nov 1, 2022 at 8:50
  • 1
    Acknowledge that less than half what you write will get to print, either discarded by you or not accepted. Commented Nov 1, 2022 at 13:20

9 Answers 9


(This answer is based on my experience with mathematical writing. I assume that it should apply to other kinds of scientific writing too, but your mileage may vary.)

I suspect that you may be fundamentally misunderstanding the purpose of the writing that you are having a hard time "discarding".

It seems like your logic is that you spent a lot of effort writing a certain paragraph, but since the paragraph will not appear in the final version of your manuscript, this effort was wasted. So that the effort is not wasted entirely, you would like to keep your writing in the drawer so that you can pull it out later at an opportune moment. This is of course pure speculation on my part, but I cannot imagine why else you would "find it difficult to discard old writing".

The flaw in this logic is the idea that the effort it took to write the paragraph will go to waste unless you use it somewhere. This is a misunderstanding. The effort was not wasted, since the paragraph has already served its purpose, which was to serve as a stepping stone on the path towards the final manuscript. Its purpose is not to be something for you to hold on to until you find a proper place for it. The paragraph had its place in the process of exploring and stumbling towards the final manuscript, but after it has served that purpose it's time to let go of it. Imagining that it is going to be revived for some other purpose is almost always wishful thinking.

Of course, from time to time you may find yourself writing an outstandingly well-polished couple of paragraphs which for some reason or another did not quite fit into the final manuscript, or paragraphs where you have a convincing, well-articulated reason to believe that they might be useful later, rather than a generalized feeling of what if. These it makes sense to keep around. But your average run-of-the-mill paragraphs that simply didn't make it through the editing process? Kill them and don't carry their corpses around. (This does not mean that you should never back up previous versions of your work, but rather that you should not primarily do so with a view to resurrecting unused paragraphs in some future work.)

Your stated reason for keeping these paragraphs in a state of suspended animation is "so you don't have to rewrite the same paragraph in future". Again, I suspect that there is some fundamental misunderstanding here, because it's not like the act of writing a paragraph is a major undertaking. The amount of effort you are going to save by not starting from zero is going to be infinitesimal, and is more than made up for by the fact that writing the paragraph anew will almost certainly result in a better one. What is a major undertaking is polishing a paragraph over and over again (and moving it to and fro) until it fits perfectly into its surroundings. But this process is not something you can substantially speed up by cutting and pasting bits and pieces from your previous work. Whether you start from zero or not is almost irrelevant, as long as your focus is on producing good writing rather than spending a minimum amount of effort on the writing process.

I can sympathize with the fact that at the beginning of your scientific career it make take you some time and effort to write a coherent paragraph, so my comment that the amount of time saved is going to be infinitesimal might not ring true to you. If so, then that's all the more reason for you to practice the bread and butter of scientific writing – which is editing, rewriting, and experimenting with different ways of expressing the same idea, rather than clinging to existing strings of words and sentences – until it does ring true.

  • 1
    +1 A big part of writing is experimenting with different ways to think about the work, as well as different ways of organizing and presenting it. When constructing a building you often need to construct scaffolding to support parts of the building during construction. Metaphorically, you often need to build intellectual scaffolding to arrive at the profound ideas that underlie the work. Once you arrive there, the path you took to get there is often messy and obscures the final result. Your writing is only useful if people can understand it, so you discard the mess in service of clarity. Commented Oct 31, 2022 at 18:27
  • 1
    This idea is what I needed to hear: "the paragraph has already served its purpose, which was to serve as a stepping stone on the path towards the final manuscript". Thank you!
    – Piethon
    Commented Aug 13, 2023 at 6:29

I used to hate discarding paragraphs, which I'd spent lots of effort writing. And indeed, they can be useful and save time. On the other hand, hoarded paragraphs could be distracting and delay the inevitable rewrites.

My solution has been to use version control, such as GitHub. It's free and easy with LaTeX, and there must be something for other document types. There is a learning curve, but lots of advantages: easy backup and version management, clean directories, easy re-use of text (you can highlight differences between versions), collaboration with co-authors, and no regrets for discarded text.

Added Clarification (fully re-using the comment by @preferred_anon):

Just to clear up any confusion about terminology: GitHub is not a version control system, it is a third party host for files that are version-controlled by the software git. For offline work, no more is needed than just git - no server, local or otherwise. "Private" github repositories are not private from GitHub, but for truly private hosting over the internet, you are forced to do it yourself. Server software like Gitlab or Savannah provide this, but both require some know-how.

  • 8
    Similarly, Overleaf is a good service that has version control. I find while I almost never use paragraphs once they were discarded, and writing is faster without them serving as a distraction, I am much more comfortable knowing we can recollect them if we need - particular if an editor wants a section chopped, then another wants it back. Commented Oct 29, 2022 at 18:04
  • 1
    I’d add that sometimes a paragraph gets discarded to meet a page limit on a conference paper. Or an experiment is reduced to a couple of sentences and a couple of numbers. Having the original paragraphs as commented-out LaTeX text makes it easier to put them back in for a “director's cut” in your thesis.
    – Pam
    Commented Oct 30, 2022 at 19:10
  • 1
    For LibreOffice, save in the .fodt file format. It's equivalent to odt, except it's a single XML file instead of multiple zipped XML files (and can therefore be version-controlled more easily). If you want a LaTeX-based WYSIWYG editor, I'd recommend LyX; its format isn't quite TeX, but it's also version-control-compatible.
    – wizzwizz4
    Commented Oct 30, 2022 at 19:26
  • 1
    If you don't want to hand over your work to a third-party server, there are local version control solutions. I use RCS, which I think is considered a bit pass\'e these days. But if you want state-of-the-art, it is possible to run a local git server. Commented Oct 30, 2022 at 23:24
  • 4
    Just to clear up any confusion about terminology: GitHub is not a version control system, it is a third party host for files that are version-controlled by the software git. For offline work, no more is needed than just git - no server, local or otherwise. "Private" github repositories are not private from GitHub, but for truly private hosting over the internet, you are forced to do it yourself. Server software like Gitlab or Savannah provide this, but both require some know-how. Commented Oct 31, 2022 at 13:53

If I have discarded something earlier, this was for a reason.

If it was rambling then, it is still rambling now.

Because of that, I keep these drafts around to kickstart writing and to help overcome the writer's block so to not stare at the blank page, but not much else. The rest is generally easier to rewrite.

There are cases where the writing itself is reasonably good, but the decision is made to make the scope more narrow for a given paper. If that is the case, strike while the iron is hot: build up on this work lest it becomes outdated.

As Hemingway famously said, "The only kind of writing is rewriting". You will be discarding a lot of old drafts, because part of what you are producing in the process of writing is knowledge and deeper understanding of the topic, not words on the paper that make it into the final version.

  • 15
    (+1) Good advice! I think Hemingway also said “write drunk, edit sober”. ;-)
    – Ed V
    Commented Oct 29, 2022 at 14:03

My take is a little different than some of the others'.

Your old rambling paragraphs will distract you and weigh you down, pulling you back into your previous thinking and approaches.

You need to improve your writing skills, so I strongly recommend you burn everything (or at least put it all away where you can't easily get at it) and just keep starting over.

You need to find your writing voice, and that takes time and a lot of attempts. It's really hard to find a better way to write if you keep focusing on your old and admittedly not good attempts.

Toss it all out.

Like the cartoon writers, each time you change your mind, pull the paper out of the typewriter, crumble it up and toss it in the garbage.

Only this way will you make a fundamental improvement in your writing skills. You will see that it gets easier and easier.


"However, I find it difficult to discard old writing." I would certainly say the same of myself. I agree with your sentiment that this stuff has to be axed from your current draft and yet, hoarder that I am, I feel your pain in not wanting to throw it away. (You might not be a hoarder like me, but that is definitely what instigates me to think in this way.)

Here is what I have found works great for me: I have a massive collection of electronic notes where I hoard all these snippets. (I use Microsoft OneNote, but you could use any note-taking system that works for you.) I save them in notes in a section that I call "Research ideas". Here is the power of this simple strategy for me:

  • Because I know that I am not throwing anything away, I can be very aggressive in cutting out these paragraphs from my current working document. There is little pain of parting because I know (or think) that it's just a goodbye; I can change my mind and bring it back whenever I want. So, I have no excuse to cut things out aggressively. That meets the very real need of aggressive rewriting if you want to move forward. For me, this is probably the single most powerful feature of this simple system.

  • Because everything is in an electronic notes system, storage space is virtually unlimited. So, I can hoard as much as I want. (That satisfies my hoarding urge with little cost.)

  • Because it is an electronic note-taking system, I can rapidly search for and recover anything that I really need. So, unlike a physical hoarding system, there is very little downside to the clutter here.

  • As I think more about these tangental ideas, the notes give me a place to develop and flesh out details. Some ideas are recorded once and then never again revisited; others I find myself coming back to again and again, sometimes over several years.

Over the ten-plus years that I've been saving snippets like this, I would estimate that probably no more than 5% of things that I put aside and save actually make it into new, active work. I evaluate this either as that the 5% justifies my hoarding reflex, or that the system frees me of 95% of deadweight that I could quickly offload from other projects. Either way, this system works for me.


So if you do find you want to use that paragraph again, it's going to be tough to find it again in a "discard" document. You will be most likely to be able to locate it again in the location you originally wrote it, because that's where you remember creating it. Much better if it can stay there.

Also, sometimes I think I can cut a paragraph, then my supervisor comes back with the comment "I really think you need more background on X", so it has to go back in.

For both these reasons, you don't really want to move the paragraph.

If you are using LaTeX, the solution is easy, just comment it out in place. If your LaTeX document is getting too large, it might be time to restructure each chapter into a separate file and pull them all together with \include.

If you are working in some other format, then the way to hide the text will have to depend on the format. Lots of formats allow some form of annotation or comment that can be hidden. But in the end my advice is still the same, keep it in it's original document, just find a way to hide it in the final output.

  • In LaTeX, I sometimes discard-but-keep text by moving it down below \end{document}. Commented Oct 29, 2022 at 21:21
  • @AndreasBlass, fine for one file documents, but once you split the document over multiple files, it would mean moving text between files.
    – Clumsy cat
    Commented Oct 29, 2022 at 21:26
  • In LaTeX you can: (a) "comment out" unwanted material using \iffalse or (b) [as Andreas mentioned] move unwanted material below the \end{document} . In both cases, that material will not appear in the final manuscript, but it remains in the source file where it can be recovered, if desired.
    – GEdgar
    Commented Oct 30, 2022 at 0:20
  • 1
    For those wanting to clean up such comments before submission, see Utility to Strip Comments from LaTeX Source
    – Anyon
    Commented Oct 30, 2022 at 17:34

As with everything you do with a computer, use version control. There's little reason to throw out anything digital in the 21st century, save security concerns.

  • This is reasonable for e.g. writing, but since you use the words "everything" and "anything", file size concerns may also apply. For example, if you generate terabytes of raw simulation data a month some deletion may be required.
    – Anyon
    Commented Oct 30, 2022 at 17:28
  • @Anyon: if you version control the code (incl. settings) that generate those terabytes, the terabytes are intermediate results which usually do not need to be put under version control.
    – cbeleites
    Commented Oct 31, 2022 at 13:22
  • 1
    @cbeleitesunhappywithSX -- I suppose the point is a bit more valid if it's data and not simulation data -- but a 10TB drive is under $200 today. If I were in the data business, generated TB's per week, $2.5k/year in drives (not counting things to put those drives in) or so to maintain local storage without deletion is not an unreasonable cost of doing business. Commented Oct 31, 2022 at 13:49
  • @ScottSeidman: I do measurements that generate in the order of GB/min raw data. While those so far have been for an institute with a different backup strategy (and their git server does not offer git-lfs), for another institute I've put GB of raw data under git lfs storage. Since raw data is anyways not supposed to change, and the most important metadata change (typos in file name) work well with git lfs, that is also fine. I'd recommend to consider a "release" snapshot of the important intermediate files when submitting the manuscript, similar to a copy of a binary release of solftware.
    – cbeleites
    Commented Oct 31, 2022 at 15:12
  • 1
    @ChrisH: I'm one of those some :-) ... And I'm very much aware of a) the seriously limitated IT infrastructure funding in many scientific projects and b) discussions on whether we really need to or even can store a copy of the raw data we acquire, or whether it is necessary to move on to store only the much reduced data volume after some initial aggregation. (After all, for many instruments, we never get access to the very raw signals, but only to suitably processed/calibrated output data which we define as our raw data)... but we're getting off-topic wrt the question of manucript text
    – cbeleites
    Commented Nov 1, 2022 at 14:32

Every time you remove sections from a document, simply save it under a different name.

In the morning you open up Abacus.doc. You spend the morning chopping bad sections out. In the afternoon you save Abacus(i).doc. The next morning you open Abacus(i).doc and continue writing as normal.

Take this opportunity to backup the most recent version externally.

You end up with many numbered versions of the same document. But this is no problem. The old versions are still there in case you ever need them. The only difficulty is if you remember something you removed but want to put back, you need to crawl through all the earlier versions to find it.

  • If you start working in the morning and don't save until the afternoon to a different file, won't the autosave feature (that is standard nowadays almost everywhere) overwrite the original file you opened? Commented Oct 31, 2022 at 20:04

Let them go.

Those paragraphs were not living in a vacuum, but they were heavily intertwined with the surrounding sentences and paragraphs and documents.

They costed you some effort, they may contain some valuable informations, both things are true, but:

  • it costs you less effort to rewrite them from "scratch" (see next point) rather than being burdened by carrying them with you ... don't let you drown in the burden of sunken costs;
  • the valuable informations hidden in them is there because youput that worthwhile chunk of information into them: either you can reproduce that information, or it is lost forever if you are not being able to produce it again, having that information already written will be of no help, if you cannot reproduce the "substance" underlying that information.

In short: either you can prepare a short essay, a commentary, a blog post, a tweet or some analogous microblogging post with those paragraphs, or they are more of a burden than of a potential help in the future.

Disclaimer: obviously there is a small exception: if you by any chance will become a Nobel prize in literature, those discarded paragraphs will have immense value ... afer your death. If you really worry about leaving something to your offsprings, think about how unfair is for some people to be loaded with cash and responsibility with no merit on their own ;D

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .