I finished
the story previously known as 'Untitled Superhero Story'. It was going slowly for a couple weeks after my last update, but really fast after I got this netbook.
I named it "The I in Evil" for many reasons that made sense to me at the time.
(
Some things remain to be done. )
Comments 10
(The comment has been removed)
There's a little bit of your life in it, too. The inspiration for one part came from your own encounters with someone who had borderline personality disorder.
Reply
Because, really, the correct answer here is to show up at Green Skull's next shindig and shoot him repeatedly with a rifle. Kevlar vests are rated for pistol ammo. There are trauma plates made of RHA that'll stop .308, but it's the same stuff they put on tanks, and it's really heavy. Also, rifles with more KE than .308 are really common, especially if you're stupid rich. In short, since Green Skull isn't running around in a mecha, shooting him with a rifle will be enough to severely cramp his style.
Reply
Reply
I suppose something I have to keep in mind is that any police sniper would then be more dangerous to Green Skull than Firehawk is. I suppose Firehawk has more money than any individual officer and many departments would have, and he's trained longer and doesn't have to worry as much about judgements on his conduct.
Reply
It comes down to what the police have on hand. Most police departments run 7.62x51mm NATO, which is what those trauma plates are explicitly rated for, but some of them run .300 WinMag, and some of them have .50 BMGs, either of which will cut through a trauma plate like it's not there.
For story purposes, though, it's nicely easy to have the robots be well armored enough to deal with any shoulder fired rifle. Most police units are short in the RPG department.
I do really like the story, BTW. I'm just a detail oriented person, even about utterly fictional universes. :D
Reply
Reply
Reply
Reply
Reply
Abstract:
"We examine the problem of keyboard acoustic emanations. We present a novel attack taking as input a 10-minute sound recording of a user typing English text using a keyboard, and then recovering up to 96% of typed characters. There is no need for a labeled training recording. Moreover the recognizer bootstrapped this way can even recognize random text such as passwords: In our experiments, 90% of 5-character random passwords using only letters can be generated in fewer than 20 attempts by an adversary; 80% of 10-character passwords can be generated in fewer than 75 attempts. Our attack uses the statistical constraints of the underlying content, English language, to reconstruct text from sound recordings without any labeled training data. The attack uses a combination of standard machine learning and speech recognition techniques, including cepstrum features, Hidden Markov Models, linear classification, and feedback-based incremental learning."
Reply
Leave a comment