Right now I’m in the middle of writing a long paper. This is a rather intense and time consuming effort that I don’t find particularly gratifying. On a good day, I can write a lot. But when I get annoyed at how something is organized I usually copy/paste and end up reorganizing things and that is usually not bad (it’s easier than retyping). However…
when I do that, some snippets here and there assume another part of the text that comes after the new pasted location. So what am I to do? I have to spend (one could even say waste) time getting the concordances right and each copy/paste turns into a major edit.
When I do this in programming the compiler tells me where things have gone wrong and I can find the errors more easily. It also doesn’t have to look pretty, it just has to do its job.
When writing a paper the aesthetics of the order of the flow of ideas is important. I’m not particularly good at getting this right off the bat, heck, I don’t consider my papers to be particularly easy to read even after I have spent a lot of time tweaking them for ease of readability and hoping to make them more accessible.
With shorter papers this process takes less effort because the possible permutations of ideas available on snippets are that much smaller.
I’m starting to think that paper writing has the following scaling equation
Where 
I’m looking forward to the days when I can tell my computer to write a paper semi-automatically and hopefully the computer will take care of all of this.
in the meantime I’ll go back to burying myself in my equations and sort the paper out.
For most average people, the exponent gamma is actually smaller than one – and I know people whose gamma is negative.
Hi Lubos:
If the exponent is less than one then they’re padding the paper by using triple spacing and adding fluff.
If it is negative, then producing an infinity long text takes no effort. Presumable because of the black hole information issue, this infinitely long text will also have no content
Exactly, David. Concerning the first, less dramatic interval, the limiting behavior for an exponent converging to 0 is the logarithmic dependence.
I am sure you had to encounter this scaling, too. You need to have a long text for testing so you write rubbish, select everything, copy/paste (or copy and paste N times), then again select evreything, copy/paste, and so forth. The number of required cycles scales like the logarithm of the resulting text.
Some people are capable of doing the same thing but in their final text, there are no long repetitions. They must use copy/scramble/paste instead of a simple copy/paste although it takes the same short time.
Lenny Susskind and a collaborator has proven that the log time is the limit saturated by black holes and matrix models. Well, it’s almost the same problem.