>>106626121
without getting too crazy into math basically "entropy" in cryptography refers to the randomness in the system, and subsequently how big the key space is (how many passwords there could possibly be and will it tend towards a given set of them)
>>106626165
entropy doesn't refer to the length of the password, it would involve the character set used, and is a property of the SYSTEM, not of the individual password/key. Even if it was a property of the individual passphrases in actual use, the "bits of entropy" doesn't change because it would be the character set in use, not that there's dictionary words in play.
also dictionary attack would get CorrectHorseBatteryStaple long before Tr0ub4dor&3 because being made up of common words means it's going to be tried first and Troubadour is not even a very common word and might not be included in initial dictionary attacks. Also no one does dictionary attack anymore anyway and anything important is protected with more complicated systems than "input your password one time to be logged on for the next 24 hours my dude"
in his example correcthorseetc is a "better" password but just by being longer. It's unintentionally a very good troll pic because even though everything in it is wrong, correcthorsebatterystaple is objectively a better password than troubadour420 or whatever just from being longer.
He's also wrong about the "cracking a stolen hash is faster but it's not what the average user should worry about" remark because that is EXACTLY how dictionaries are built for dictionary attack and how intercepted passwords once handled properly start being included in spam letters addressed directly to you saying they watched you watch porn, and is why it's strongly recommended to use different passwords for everything (and that if they aren't different enough you need a pw manager to remember them all, they're not different enough)
also this man isn't aware of the water cycle