Hacker News Reader

Comment by mdeeks | Hacker News Reader

Can someone define what is considered a weak vs strong password now for WiFi? The only guides I found online are years old.

Is 10 characters considered weak for mixed case letters, numbers, plus punctuation now?

View on HN

To do this formally, you need to consider information entropy. This is all about how you generated your password. 10 characters of totally random mixed case, numbers and punctuation gives about 60 bits of entropy which is strong enough.

HOWEVER, that calculation only works if all 10 characters were generated uniformly and randomly. Humans are terrible at this. Now, maybe your trick for turning words into safe passwords is great, but there is no way to be sure. Sadly, remembering 10 random characters is hard.

Luckily, easy to remember and strong passwords are possible. The system I would recommend is diceware: www.diceware.com

That diceware system is complete Snowden-level paranoia. Close the curtains! Burn after reading! For everyday techie joe a passphrase + a memorized complex password is just enough. If you're on the internet asking for a strong password method and reading diceware.com you may have your priorities set wrong like an untrained spy.

https://duckduckgo.com/?q=pwgen+strong+10&ia=answer

I would love to see a comparison between where physically and which modifiers are used for each character are, and the strength of a password.

Is a password which is very easy/comfortable to type out physically any more/less strong than another of the same length?

I ask this because I often use a visual pattern on the keyboard for a password and I don't recall which characters they may be, but I recall the pattern in need to type out on a qwerty kb

Depends on how good the pattern is, however entropy is lower all likelihood because the layout of qwerty keyboard is standardised.

Most password crackong dictionary already include common keyboard patterns sich as "qwerfdsazxcv" or variations of it.

There was a nice comic/picture of this. I tend to follow it. Basically using 3-4 short words as a phrase instead of random characters. You can toss special characters inbetween/before/after. They are also much easier to remember. Password "FoolMeOnce!ShameOnMe" for example.

But you've gone and picked only one preexisting phrase, instead of independently-random words. That cripples the security of your password.

Making a phrase is okay, but you have to start with actually-random words.

Well, it was an example, but I agree. For everything that I can I use keepass with better autogenerated random passwords, but for things like home WiFi and others that I may have to type in manually I'll use a phrase like this. A more random phrase is certainly more secure.

For completeness sake, this is probably the comic you are referencing: https://xkcd.com/936/

    curl -s https://raw.githubusercontent.com/first20hours/google-10000-english/master/google-10000-english-no-swears.txt | shuf | head -n 4 | tr '\n' ' '; echo
    mine wear vacation mostly

log2(10^16) = 53 bits of entropy or 300 years if your attacker can do a million guesses per second (the link says 1000 keys per second, but that's on the CPU).

You could also use `cat /usr/share/dict/words` instead of the `curl`, which is a much larger word list, but you get impractical passwords like "globular cellulose's malnutrition's dangling".

Careful, shuf is not cryptographically safe by default! You need to pass --random-source=/dev/urandom to get a proper RNG.

https://www.gnu.org/software/coreutils/manual/html_node/Rand...

Why does shuf implement its own random number generator? Why isn't /dev/urandom the default?

https://sockpuppet.org/blog/2014/02/25/safely-generate-rando...

shuf is not a crypto tool, and the GNU coreutils are written to be cross-platform, even where /dev/urandom doesn't exist, or is unreliable. That's my guess, at least.

Nice, yes that was the one, thanks.

Wifi password cracking is only around 1000X slower than a SHA256 brute-force if I remember right. So your password needs to be secure enough that if a hash of it was leaked it would never be cracked.... So very strong.

WPA enterprise using certificates is usually much harder to crack since you need to interrogate server, you can't just brute force hash. This method only really applies to PSK mode (home networks and small businesses usually)

Weak = is in rainbow table that hashcat is using

That should be enough, but keep in mind that you might be entering the password on a lot of non-keyboards. 20 lowercase letters is faster and far more secure. Even 14-15 lowercase letters is soundly better.

If you consider your random keyspace with 26 * 2 chars + 10 numbers + 20ish special chars then to crack 10 letters you'll have to try an average of ((26 * 2 + 10 + 20) ^ 10) / 2 = 6.8724016e+18 keys. If you then assume around 3 million hashes per second it still takes around 72641 days to crack your password.

Edit: As another comment said, just make sure it's not easy to guess based on rainbow tables and whatnot

Did you mean 72641 days or years?

(And could / should we include somehow that "hashes/second" increases by factor of ~2(?) each year?)

I read in many places that the easiest way to generate a really strong password is to memorize a fairly long sentence and use the first letter of each word.

However definitely DONT use a quote or lyrics. Needs to be something unique.

Reply on news.ycombinator.com