Interesting Links for 16-03-2018

[andrewducker]

"I Watched All 629 Episodes of The Simpsons in a Month. Here’s What I Learned."

(tags: simpsons society TV review women Intelligence )

Ask a computer to do something and it'll cheat like hell

(tags: ai computers problems intelligence cheating funny )

Technology Meant to Make Bitcoin Money Again Is Now Live

(tags: bitcoin cryptography money )

Live in Phoenix, Arizona? You can now fall asleep in the back seat of a driverless taxi

(tags: taxi automation usa )

If your parents keep getting your name wrong it's because they love the dog just as much

(tags: memory names family )

Two months of daily GTA causes no significant changes in behavior

(tags: behaviour games psychology )

Planting GMOs kills so many bugs that it helps non-GMO crops

(tags: genetics farming insects )

A look at a comic that touched on everything that makes Batman great

(tags: batman comics )

Decoding the Signals In The Prime Minister's Speech on the Russian Poisoning of Sergei Skripal

(tags: Russia UK diplomacy viaFanf )

Gibraltar warns it could rescind citizens rights if Spain uses veto on Brexit deal

(tags: Europe Spain Gibraltar )

No progress in finding solution to Irish Brexit border problem

(tags: Ireland UK Europe NorthernIreland parliament borders )

After Rihanna instructed her 60,000,000 Instagram followers to delete Snapchat, the company's shares plunged- causing the it to lose about $600,000,000.00 in market cap in a matter of hours.

(tags: money celebrity socialmedia )

Cyberattacks Put Russian Fingers on the Switch of American Power Plants

(tags: power Russia security hacking usa )

Comments

[simont]

Ask a computer to do something and it'll cheat like hell

Another particularly annoying example of this is if you try to apply a fuzzer to some code that's not well set up to be fuzzable.

If you're using, say, AFL as your fuzzer, it likes command-line programs that it can feed stuff to on standard input and gradually tune for inputs that have interesting effects. So if you want to fuzz, say, a pile of library code that accepts input in the form of in-memory data structures, probably the first thing you do is write an AFL adapter, in the form of some sort of wrapper program that takes stuff on standard input, parses it into instances of those data structures, and calls the library routines on those in turn.

If you do this, you will inevitably spend the first week combing out the bugs AFL found in the new parsing code, and it's even money whether your patience will last long enough for AFL to start taking an interest in the code you actually wanted to test.

Two months of daily GTA causes no significant changes in behavior

What, not even being better at GTA? :-)

[andrewducker]

Same is true for most testing code, in my experience. I wrote some code yesterday, and then some test code to mock out the various dependencies and test each of the paths. Test code was about three times the size of the actual code.