(no subject)

[kjorteo]

Current status on the editing/proofreading/revising/etc. phase of my novel:

(Note that I am employing a "give each chapter a complete overhaul and then a few quick passes, and then a quick pass to every other chapter so far, and then move on" strategy to keep everything relatively current and attempt to avoid the Forth Bridge problem.)

(Ch. 1-4 overhauls and passes done way back before I had the idea to keep track of the dates on which I finished anything)
Ch. 1 pass:	2012-06-10
Ch. 2 pass:	2012-06-14
Ch. 3 pass:	2012-06-19
Ch. 4 pass:	2012-06-23
Ch. 5 overhaul:	2012-08-15
Ch. 5 pass:	2012-08-28
Ch. 5 pass:	2012-09-12
Ch. 1 pass:	2012-09-18
Ch. 2 pass:	2012-09-24
Ch. 3 pass:	2012-10-01
Ch. 4 pass:	2012-10-08
Ch. 5 pass:	2012-10-18
Ch. 6 overhaul:
Ch. 6 pass:
Ch. 6 pass:
Ch. 1 pass:
Ch. 2 pass:
Ch. 3 pass:
Ch. 4 pass:
Ch. 5 pass:
Ch. 6 pass:
Ch. 7 overhaul:
Ch. 7 pass:
Ch. 7 pass:
Ch. 1 pass:
Ch. 2 pass:
Ch. 3 pass:
Ch. 4 pass:
Ch. 5 pass:
Ch. 6 pass:
Ch. 7 pass:
Ch. 8 overhaul:
Ch. 8 pass:
Ch. 8 pass:
Ch. 1 pass:
Ch. 2 pass:
Ch. 3 pass:
Ch. 4 pass:
Ch. 5 pass:
Ch. 6 pass:
Ch. 7 pass:
Ch. 8 pass:

AAAAAAAA THIS IS NEVER GOING TO BE DONE :C

Comments

[premchaia_pre4]

… I cannot imagine why you would be using a pass order that's asymptotically quadratic. Why? o.O

[kjorteo]

Like I said, it's the Forth Bridge problem. Every time I neglect any one chapter for too long, it starts to deteriorate in comparison to the level the other ones are suddenly at--obvious skill evolution like http://www.zapcomic.com/2003/07/20030725/ to http://www.zapcomic.com/2011/03/519-left-out/ is okay in sequential media like webcomics, but a singular novel is supposed to read start to finish like the whole thing was written by the same author, and that's proving harder than I thought it would be! So my strategy is "overhaul chapter x, then quickly go over every single chapter before it to make sure they're all still okay, then move on to chapter x+1...."

[premchaia_pre4]

… okay, sure, but why not overhaul all the chapters, then do light passes over all the chapters, or… ?

I mean, not having written any novels, I'm sure there's something I'm missing about that process, but.

[kjorteo]

Note how long the overhaul took compared to the light passes. If I overhauled every chapter in order and ignored the others, then they would all be in need of overhauls again by the time I finished, in an infinite loop. In fact, that already happened once; this current process is my strategy for how I hope to make my second time through this whole thing the one that finally ends it.

[premchaia_pre4]

Well, if we model the stylistic drift as linear over total writing time spent, and we have that “overhaul all non-first chapters” results in sufficient drift to convert the first chapter from an up-to-date state to a requires-overhaul state, then you're hosed, right? If the drift is monotonically increasing in delta-time between one part of the work and another, and there's a maximum amount of acceptable drift, and your working speed is bounded (or else the relevant subjective time is locked to working speed anyway), then there will exist a size of work beyond which you cannot force the drift below the acceptable threshold.

So I suppose it depends on the mechanics of “light passes” versus “overhauls”. If the former are enough more efficient at pulling stylistic elements together, but cannot be performed if there is a sufficient difference to start with, then I suppose that makes sense; you're linking a chapter into the new refresh chain with each overhaul, and the overhauls are there to move from a globally old state that had no refresh chain into a globally new state that does.

I mean, that would explain it. Is that what you're doing?

[kjorteo]

I cannot believe I (mostly) understood that (I think.)

Basically, you're right in both counts. In theory, yes, something absurd like a 30-chapter set would take so long that even light passes exclusively with no overhauls at all would probably still risk that problem. However, I think an average-sized novel (it's something like 70,000 words at five chapters in, and knowing what I know about the last three, I therefore project roughly 120K overall) is small enough that this system still works as described in your second paragraph (yes, that is exactly what I am at least trying to do.) Or, at the very least, I seem to have made it this far, anyway....

[premchaia_pre4]

Basically I was getting confused as to how such a model could actually work. It would have to involve nonlinear productivity, because in the linear case, “overhauling every chapter in sequence implies having to overhaul the first chapter again” implies “lightly revising every chapter in sequence implies having to lightly revise the first chapter again”. I was also getting confused as to how the light ones could be more efficient, because they'd take more context-switch overhead.

But: now that I'm awake again I think that was a red herring. Rather than being more efficient, it may be that the light ones are qualitatively different enough that they can be considered to introduce zero (or vastly less) backward force on your writing style, due to not thinking hard enough about them to hit the LTM write barrier enough to significantly advance the relevant subjective clock. This would be consistent with my experience with neurological memory. If so, the boundary between light and heavy attacks (I keep imagining them like in Dustforce now) is basically the boundary between short-term and long-term memory access patterns. That would also explain why a series of light attacks either could not be used or would be massively less efficient at moving the chapters into the globally new state, because that involves wider-spanning editing windows that don't fit into short-term memory. (It may be possible to reëncode those processes in a way that they can be done with multiple light passes by encoding partially-edited-state information into the text, externalizing the long-term element in a way that isn't subject to neurological write-induced drift, but that seems like it would be more awkward than it's worth. Note that that sort of thing is what is used (usually badly) by a lot of human organizations to distribute work between members.)

It also implies that you have to be very careful during later-chapter overhauls about not introducing inconsistencies that affect sufficiently wide spans of text as a whole to require doing another heavy attack on an earlier chapter, because that could set back the overall process pretty severely.

So that's a neat consequence of that architecture that it yields these nonlinear sequences.

[premchaia_pre4]

(It may be possible to reëncode those processes in a way that they can be done with multiple light passes by encoding partially-edited-state information into the text, externalizing the long-term element in a way that isn't subject to neurological write-induced drift, …)

Oh, and note that if this is globally so, that would make it possible, if difficult, for you to handle the 30-chapter case by doing that. The first part of it would involve more or less exporting the more macroscopically-oriented parts of your writing ability into the partial book (well, or into a separate adjunct form) in a way that won't drift, and then having the text partially write itself in combination with you channeling relevant bits of it as necessary. But it would probably require an awful lot of passes.

[kjorteo]

... Yes.

[premchaia_pre4]

I use Anghel's Pained Look! Or something. @.@

You don't have to think as deeply about light passes (that's what makes them “light”), so your skills don't change enough over the course of those to matter. Right? c.c

[kjorteo]

I hope so.

[premchaia_pre4]

That is such an appropriate face! I do hope it goes well, for what that's worth. But I probably shouldn't mention that XX XXX XXX XXXX XXX XXXX XXXXXXX er, no, I really shouldn't mention it. Never mind!

[kjorteo]

XXX is good, too.

[premchaia_pre4]

*flapflapflapflapflap*

*flapflap*

*flapflapflap*

[crassadon.livejournal.com]

"AAAAAAAA THIS IS NEVER GOING TO BE DONE :C"

Well, good! . . .or. . how does this work again. . ?

[kjorteo]

Um, I kind of want to finish it so I can look into getting it published or something ... this is my novel, it's kind of a big deal. :(

[archadia.livejournal.com]

OMG I feel like a total dorkus I still have those sitting in my inbox and need to read them!

[kjorteo]

Here, have the new version, then. :D

[archadia.livejournal.com]

eeeeeeeeee!

[davidn]

All right, so you still have a fair distance to go, but... missing out even placeholders for the first four chapters makes it look much longer! And I'd be interested to see what it looks like on an actual calendar, because as Premchaia said (in more detail), not every task takes the same amount of time...

[kjorteo]

Oh, is that what she said! All right.

[premchaia_pre4]

Oh, and an approximate eyeball extrapolation suggests that if things keep going as they are, it'll take about another year to finish. That's assuming you don't go “… well, maybe another overhaul of the first chapter” and start trailing off into the dark void shockwave.

[kjorteo]

Yes, I noticed that as well ... that's the problem, I think. That's a very sobering realization, to put it that way.

[premchaia_pre4]

In software there's a related phenomenon known as second-system effect. I have the feeling just about everything I do is affected by something similar, and that's why I haven't ever really released anything serious yet, argh.

It is supposedly useful to remember that All Art Eventually Abandoned Is, but I imagine you're already remembering it right now! so it's probably not very helpful for me to say. o.o