Scientific publishing

[hans]

So, by having an LLM "assist"in writing paper they can now ever more closely resemble existing papers, which are already mostly so badly written as to be unintelligible

 

I did two Master's degrees in very different fields (Germanic Languages and Electrical Engineering) and in both fields a certain "academic language style" is expected in order for a paper (or presentation / lecture) to be taken seriously. 

Ah yes. In Dutch we say "wollig schrijven" meaning, in a literal sense, "wooly writing". Many words are added to sound more formal.  E.g. something like:
"I've built an IoT mote with energy harvesting for perpetual wireless sensor networks"

would be translated to something as:
"In order to investigate the feasibility of perpetual wireless sensor networks, in this paper we propose a new design of an IoT mote which utilizes energy harvesting such that it can be powered indefinitely."

Its actually amazingly easy to fill half a dozen pages of 2-column paper without any punchlines to continue new paragraphs For me, I tend to limit this writing style to introductions only, as those are an overview/collection of related literature which I tend to avoid criticizing too harshly in the fear that I may hurt some reviewers feelings (maybe its their work).

[pdenisowski]

Part of my University course I had to write a dissertation and it was something I really struggled with. If there was a LLM that I could have used I think I might have been tempted to help me write coherent sections as I have a habbit of rambling and repeating myself, ideas come in faster than I can write them and by the time I am half way through the paragraph I have lost what I was thinking of writing.

A lot of my current role at R&S involves writing / presenting on technical topics:  I've produced over 200 pieces of content (videos, webinars, whitepapers, etc.) in the last 5 years alone.

My motto (attributed to Cato the Elder):  rem tene, verba sequentur.  Loosely translated it means "grasp the thing / topic / subject ("re")" and the words will follow"  In my experience, the problem most people have with writing is that they start typing before they have a clear picture in their mind of what they want to say, how the want to say it, and the order in which they want to say it.  I don't even begin typing until I can outline or even recite the entire piece verbally, from memory. 

I blame a lot of this on modern word processing:  it's far too easy to simply sit down and type, backspace, copy/paste, etc. without first spending sufficient time "grasping" or organizing the material.  I did my first Master's thesis on a typewriter (you know, the kind with ink ribbons and paper) and when correcting even a single mis-typed letter is a significant hassle, one tends to spend most of one's time planning, drafting, and organizing:  the typing itself is essentially a mechanical task.

And my second favorite motto: c'est en forgeant qu'on devient forgeron - it's in blacksmithing that one becomes a blacksmith.  Everyone starts their life as a bad writer.  Although there are some (notable) exceptions, most peoples' writing improves the more they write.  Using an LLC as a substitute for developing one's own writing skill may be expedient, but -- in my opinion -- it can be counter-productive in the long-term.

[Nominal Animal]

In my experience, the problem most people have with writing is that they start typing before they have a clear picture in their mind of what they want to say, how the want to say it, and the order in which they want to say it.

Agreed!

I approach technical writing as a construction task.  I collect the points and details and images that are important (usually using a separate scratchpad text file with a sentence or paragraph each, with any related citation references if academic).  Then, I consider the structure, and what kind of sequence and approach I believe will work for the readers/listeners (based on what I personally prefer, really).  Only then do I start really writing.

I blame a lot of this on modern word processing:  it's far too easy to simply sit down and type, backspace, copy/paste, etc. without first spending sufficient time "grasping" or organizing the material.

For over a quarter century, I've fought against teaching people how to use a specific word processor like MS Word –– exactly because it leads to them doing that, taking the quickest options and shortcuts without learning how to use the tool in an organized fashion making them more efficient ––, and to switch to teaching them how to use our word processing technologies to assist them in technical writing.

Initially, I started by explaining that the way the text is laid out is best managed using the conceptual styles –– heading, body text, listing, etc. ––, and not by picking a font you like from the convenient drop-down menu shown, so that their output would stop changing the font mid-way.  Then, I showed how the other related features like page header and footer can be used to add page numbers, and how using the proper style for headings and figures, you can make the word processor generate the indexes for you automatically.  However, all this is for the writing part, not the construction part.

Most word processors like MS Word and OpenOffice/LibreOffice Writer do have an Outline view or Navigator, so that if one uses descriptive headings, one can move them around very quickly in the Outline/Navigator view without having to cut-paste the actual text.  However, I think using the same tool for the construction tends to 'push' the user too early and too easily into 'writing mode', so I do prefer using simpler, cruder text editors and interfaces for the construction part.  Sometimes even Post-It notes on a window/door/wall just work better.

All that said, I'm not an author/speaker/presenter.  I've only made a couple of dozen presentations and talks along the years, and while they've gone well and sparked interest and interesting follow-up questions on the topic, I put that mostly on the topics themselves, and not my presentation/formatting.  (And no, I don't tend to overshoot my allotted time, unlike in writing...)

[TimFox]

After becoming familiar with WordPerfect and then Word software, I approached my technical memo writing without using all the features included in those programs (since I never had a manual for any of them).
I would just start by generating an outline for the material to be reported, subdividing the "1." heads into "1.1", etc. between the heads, and then translate my notes into the prose between them.
Cut-and-paste was useful for pulling material out of other memos or sources, and for re-arranging the prose into a more readable style, with useful punctuation.