Cool concept!!

Created a repo with some licensing information (Pew Survey is very protective of its data, and I can't blame them).

Pew Forum hired a polling firm to call 35,556 people in the continental USA, meaning that I happen to have a very rich dataset. I manage to extract out 99 atheists who believe in God and save them into my own personal *.CSV file. I need 505 words for each atheist to produce a 50,000-word novel.

+1, sir!

Danae Marks (a White blue-collar laborer) entered a church, ready to deliver a speech to religious people. She saw religion as a progressive force in society, but one that has engaged in a horrible crime: interfering in the secular world and trying to sway the national government. This violation of the separation of church and state cannot stand!

Danae Marks used practical experience and common sense to make the claim that the universe is controlled by a magical force called 'Karma'. Danae Marks argued that Karma was a personal, physical being. She then held out a personal copy of the Bible. This 'Bible' came directly from Karma, but ignorant people have chosen to interpret the book literally. A literal interpretion is one that is doomed to failure. Only metaphors lead to success. Danae Marks smiled when she explained how Heaven and Hell are the same as the 'religious' people viewed it: good people go to Heaven and bad people go to Hell. Danae Marks then explained in depth how Karma is able to break scientific laws from time to time, causing 'miracles'. Danae Marks declared that Karma strongly supports clear ethical standards that all humans must obey.

The theists took notes.

I know this sound surprising, considering this is a novel-writing competition...but I don't actually like novels. I always believe in conveying messages in as few words as possible, rather than filling the page with worthless purple prose. I enjoy reading novel synopses, not the novel.

When writing this competition, I had two goals: 1) write a program that generates a novel, 2) write a program that generates a meaningful story.

Goal 2 is a complete success, in my opinion. Each sentence conveys information about what a real person said over the telephone. Theoretically, someone may want to read more of the novel to find out about more atheists, although boredom will eventually set in for the reader...

The problem I'm having is Goal 1. Each sentence in my template is too effective at conveying the data. Each data point only really gives me one sentence (and sometimes even less!), and while there's 99 different atheists, that only means 99 sentences.

My estimate is that each "chapter" is about 200 words long. This means far from getting a glorious 50,000 words novel, I am stuck with a ~19,800 word novella.

Now, there's a lot of data points that I am not using yet, and clearly using them will lead to more words to be stuffed into each chapter. But as text is efficient at conveying information, I'm going to have to exhaust a lot of data to reach my original quota of '505 words/chapter'. And each data point I add seems more and more tangential to the original purpose of the novel: to show the religious beliefs and practices of self-proclaimed atheists who believe in God. As a result, I may consider just abandoning Goal 1 entirely. It may be possible for me to get up to 505 words...but the cost may be a less meaningful story. And I think I prefer meaning more than mere word count.

There are also examples of people getting the Green complete tag despite not reaching the 50,000 word criteria too, so there's no shame in abandoning the pursuit for the Great White Novel.

There are also examples of people getting the Green complete tag despite not reaching the 50,000 word criteria too, so there's no shame in abandoning the pursuit for the Great White Novel.

My belief is that the spirit of the 50,000 words should be interpreted in light of what NaNoGenMo was in reaction to: lots of successful Twitter bots and the like, very few long-form generators. Writing a novella that says everything that you wanted to is, to my mind, well within the bounds of NaNoGenMo. I've personally viewed the length as aspirational: we, collectively, don't know how to get a machine to write a good novel, so we're slowly fumbling our way towards writing a coherent novel, gradually trying to make each attempt more interesting.

As a practical matter, if I was in your place, I'd step back from thinking about the length and turn instead to the problem of keeping the reader's interest for more chapters. Is there a way you can add more variability? But I'm not in your place. You get to be the one to declare it complete, on your terms.

If I may air another viewpoint re word count:

What attracted me to NaNoGenMo when I first heard of it was the absolutely absurd requirement of 50K words. How do you get a machine to fill that space in a "sensible" way? Or do you even try? If not, what do you accept for "not sensible" ways to fill it? And why?

To me, answering those questions is probably just as central to NaNoGenMo as the actual generating of text. If it was simply billed as a computer story-writing jam or something, with a reasonable word count goal, I probably wouldn't have bothered to look into it.

So, you could say I am rather strongly pro-taking-50K-words-seriously. (I'm also rather strongly pro-novel-means-NOVEL-gosh-darn-it, at least in odd-numbered years, but that's a slightly different matter.)

That said, I'm virtually in the same boat as you @tra38, and I don't have any good ideas.

My current plan is to generate two versions of my novel at the end: one at readable length, the other a 50K word version to get the Completed label even though 98% of it will be pure wharrgarbl.

novel means new, gosh-darn-it!

I actually did end up generating a novel of over 50,000 words, simply by duplicating the data (so instead of analyzing 99 atheists, I analyzed 99 atheists twice). This brings the final word count to 61,662, but I suspect most people are going to skim past the first chapter to look for any interesting combinations.

Full Text: https://github.com/tra38/The-Atheists-Who-Believe-In-God/blob/master/atheists.md
Repo: https://github.com/tra38/The-Atheists-Who-Believe-In-God

I'll also have to stop here, since I won't have much time to work onward on this project. If I had more time, I would try to create multiple templates, such as "atheist drives to work", "atheist writes a novel to promote his/her non-religion", etc.

As a proof of concept, it works and I really like the resulting output. It's much more interesting than reading a table, at least.

According to this graphic:

It's likely that the number of atheists who believe in God would likely decrease even further in the coming years. So this novel is a product of its unique dataset...there will soon not be enough god-fearing atheists to fill even a novella.