Re:Terry Pratchett's unfinished novels destroyed by steamroller
by bitexploder on Wednesday August 30, 2017 @08:35AM
(#15131558)
That platter was remarkably intact. You can send broken drives like this to a company. They can recover most of the data. NSA could do it too. If you don't overwrite several times with a random pattern it is recoverable.
Also that drive is old and I am suspicious that is the only copy, or that the novels we're there anyway. Seems likely Pratchett destroyed them already and this was symbolic.
Re:Terry Pratchett's unfinished novels destroyed by steamroller
by Houshalter on Wednesday August 30, 2017 @07:00AM
(#15131117)
By the time we get the technology to do that, we could probably build an AI pratchett that generates great works perfectly within his style and quality. Like that black mirror episode but slightly less creepy.
Re:Terry Pratchett's unfinished novels destroyed by steamroller
by mattkevan on Wednesday August 30, 2017 @07:24AM
(#15131206)
Ooh, I wonder what would happen if someone fed all of Pratchett's books into a RNN?
Would it generate new Discworld stories, or would it rip a portal into the Dungeon Dimensions releasing the eldritch horrors that dwell within?
Re:Terry Pratchett's unfinished novels destroyed by steamroller
by gnode on Wednesday August 30, 2017 @08:24AM
(#15131486)
It would produce superficially Pratchett-sounding prose with no coherent plot.
Re:Terry Pratchett's unfinished novels destroyed by steamroller
by LeifCarrotson on Wednesday August 30, 2017 @12:13PM
(#15133372)
Presumably, the RNN would be deep enough to recognize that "a coherent plot" is a component of the work on which it's being trained and which it must generate. But not so deep that it generates eldritch horrors.
Re:Terry Pratchett's unfinished novels destroyed by steamroller
by OtterCoder on Wednesday August 30, 2017 @07:00PM
(#15136843)
That would be a fantastically deep network. Current AI can sometimes keep track of what's going on until the end of the paragraph. A novel's worth of threads to track is simply too much for the naive toss-it-in-a-wood-chipper approach.
It might not be a stretch to say that future generations would have the technology to recover the data, if past trends are any indication:
http://www.smithsonianmag.com/history/ancient-scrolls-blacke...
As far as physical destruction goes, this isn't even that extreme.