The 'ArcAngel' sounds like an interesting tool - like the Microsoft Word grammar checker. It sounds like a tool I'd like to use and experiment with.
But running this competition:
Basically, the prize is $1,000 for the best novel and $1,000 for the best screenplay - again, judged solely by the Arc Angel.
What the heck?
The algorithm they described simply scores words for emotional intensity. So fundamentally you could give it garbage, plus sprinkled words to weight the words for emotional intensity, and it would decide it is the 'best' screenplay or novel
Not only that, but they claim that the technology is 'patented'.
I suspect that they are being a tad generous in their claims. A quick google through patents using their names and keywords like 'Story analysis' lists no patents issued by the US Patent office. (Well, it gives plenty of patents .. but no ones that seem to match)
(Use
www.google.com/patents to search .. it is much better than the official US patent search!)
Of course, the dishonesty is just my suspicion - but given the lack of Patent # and the lack of matching keywords in patents, it would be their fault if people assume the worst! I'm honestly surprised ... you can get the US Patent Office to issue a patent for almost anything *cough* Amazon One-Click */cough* even if it isn't actually patentable.
It is too late for them to get a patent now - they have publicly described the essence of the algorithm already:
1. Score words for emotional intensity/wisdom/other category
2. Group the words into which characters they belong too. (Optional)
3. Graph the emotional intensity (or whatever category) for each character - and possibly overall as well.
4. Compare the graphs to what the platonic ideal of a novel looks like.
The funny thing is that I did something similar with the first feature screenplay I tried to write. I didn't have the dictionary lookup, but I tried to graph the intensity for each character .. so that it wouldn't be 'flat' all the way through.
I've got a lot of sympathy for this kind of tool - it would be an interesting add on to Final Draft. However the idea of it being used to judge the 'quality' instead of making a useful estimate of an attribute is simply bizarre. For example, I find I often use the 'read out your screenplay' feature while reviewing .. it is totally artificial but forces my brain to listen slowly which often picks out silly mistakes I'd otherwise miss. However, being automatic is particularly bad at many things. It can't pick out the difference between 'He winds the clock' and 'The winds are strong' for example.
The main issue with the algorithm they've described is that it would miss all emotion which is done in subtext. Take the remake of 'War of The Worlds' where Tom Cruises goes to make a sandwich for his daughter, only to have the daughter tell him that she can't have the sandwich because she's allergic to peanut butter. That's a great scene - the single line "I'm allergic to peanut butter" tells us EVERYTHING about Cruise's character's relationship with his daughter. It is a REAL kick in the guts for his character .. not to know something so vital about his own kid.
But for an automatic system to tell us that 'I'm allergic to peanut butter' won't invoke as much emotion in the viewer as 'You are a useless father' is nonsense.
Likewise, the most horrifying line in 'Terminator 2' is when John's foster mother refers to the dog and says "Wolfy's fine, honey. Where are you?". I'd like to see an automatic system that could figure that one out.
Mac