The login limits the password length to 20 characters but the change password has a larger or no limit.
Thanks for the report — I’ll look into it. I generally create only 20-character passwords so I hadn’t run into that. For those who are wondering, the minimum is 8 characters, which is the recommendation from NIST, but the change password text in WordPress recommends 12. And frankly, I see no reason not to go all the way to 20 if you’re generating a random password in 1Password or LastPass anyway.
Hmm, can you tell me where you ran into this (the on-site login dialog, the WordPress standard login screen, etc) and what happened? I’m having trouble reproducing it.
For reference, I created a 30-char password and had no trouble logging in with it. I didn’t check how many characters were in the password field on login as I had 1P auto-fill.
I changed my password to 30 characters but when I went to log in with LastPass, only 20 characters were allowed in the password field of the login screen. However, since Jolin (below) says that s/he had no problems doing exactly this, it must be an error on my part. Sorry.
Sometimes you’re on a machine/site without ready access to 1passqword and you have to type in the password, so there is an (edge) case for shorter passwords. I mean, not shorter than 12 characters, let’s not get silly!
No worries — I was able to type more than 20 in the spots I tried too, but you never know what quirks are hiding in the corners.
I’ve occasionally run into that in the past. I’ve solved it by simply not using machines where I have no access to 1Password! 
My iOS pain has always involved having to type in a gibberish password with an iOS virtual keyboard. Even if you’re limited to 12 characters, that’s potentially 24-30 keystrokes.
I solve it by using 1Password synched via Dropbox to all my devices. I’m practically a legacy user at this point, so I haven’t used a AgileBits account. I freely recognize that I’m in trouble if I’m working on a strange machine, the edge case you mention.
A related (sort-of) edge case: my ISP has moved to recommending 3-4 word pass phrases in lieu of random characters, going so far as to say there’s no need for the latter. I’m not entirely sold on that idea, but it does mean that even a 3 word passphrase can easily exceed a 20-character limit.
If you’re looking at standardizing, are there practical considerations, other than storage space, to limit the password field to 20 characters? Personally, I choose the longest length a site will allow (some allow 50 characters or more, and 1Password generates that in a heartbeat).
They are right and they are wrong. For a user of a password manager, they are wrong. 100% wrong. a 20 character random password is many orders of magnitude more secure than one that is much longer and composed of 3-4 words. Heck, a 12 character random password is more secure.
For people who do not use a password manager, your ISP is 100% right because a short set of words is much easier to remember, easier to disguise, and easier to recover (as in you might remember two of the words and have an idea of the third).
However, if a person is choosing the words, then it all goes out the window and you’re back to MONKEY1 level security. For example, I know someone who had the pass phrase “top hat and cane”, which hopefully everyone can see is a terrible password. Even worse, the person was a HUGE Fred Astaire fan, which anyone who met him for more than 4 seconds would know.
The passphrases have to be pretty long and use words that aren’t related in any obvious way. This is the classic xkcd “correct-horse-battery-staple” example:
The entropy is high enough, particularly if you go up into the 32-character range (that’s what @joe recommends in “Take Control of Your Passwords,” where he explains this at length.
But with a password manager, as @kreme says, I think 20 random characters is best, in part because you can’t possibly know or remember the password. And of course, always remember this other bit of wisdom from xkcd:
1 Like
I’d seen that first xkcd comic a few times, but not the second. Comedic gold…
Actually…that isn’t correct…and as a former computer security guy for the DoD this is one of the things that I did professionally.
Password crackers operate via several different methods…dictionary and/or rainbow tables are commonly used for shorter passwords and the dictionary tables include things like “Four score and seven years ago”…in order for this password to be easily cracked by either the whole phrase has to be in the dictionary and/or rainbow table in order for it to be cracked, merely having those six words in the table separately won’t work.
What you say about having 4 words…even if the 4 words are individually in the dictionary unless all possible combinations of 4 words are also in the dictionary as a phrase for an individual guess they’re fine as illustrated below…sure, you could write your hacking program to use a dictionary and then all possible word combinations…but then you’re really just using brute force methods at that point.
However…once you get over about 18 to 20 characters…tables really don’t work any more as the size of them is too large to be easily sorted through…and while one might argue that it’s really 25 characters before this takes effect the bottom line is that once you get past a certain length then the only feasible password cracking method is simple brute force…i.e., you try every possible combination until you find the password and the longer the password the longer it takes to crack.
There are numerous password strength web pages on the web…I personally recommend using Steve Gibson’s (of Security Now podcast and SpinRite fame) page at https://www.grc.com/haystack.htm as a good example…you put in a potential password and it tells you both the size of your password universe and the length of time it will take to crack it in a brute force scenario with varying cracking horsepower scenarios. Steve admits right on his site that Haystacks isn’t a password strength meter but rather a size of the universe meter…but you can plug the same example passwords below into any strength meter and get pretty much the same results.
Your statement that a 20 character random password is many orders of magnitude more secure than 3-4 words is incorrect.
Using the Haystacks page…
The 20 random character password f6bnvCdxToTFvfYeNB7Kf is cracked in 2.28 trillion centuries using the massive cracking array of 100 trillion guesses per second.
The 26 character 4 random word password correcthoursebatterystaple is cracked in 20.36 trillion centuries…or roughly ten times longer than your random 20 character one.
Your suggested 12 character one…like C3hmqU6TxHE% is cracked in a mere 1.74 centuries.
You’re correct…using a password manager is a superb idea…but you don’t really need to have random passwords. Once you get long enough…and long enough is an ever increasing time as computers get faster and drives to store the tables get larger…then password cracking is down to just plain, simple, old brute force…and the only thing that really counts there is length.
I always include the 4 basic food groups for passwords, upper, lower, symbols, and numbers…but they really don’t make much difference…using CorrectHorseBatterySt84%# which is also 20 characters increases the cracking time to 89 trillion trillion centuries…but even without anything but lower case characters I’m thinking that 20 trillion centuries is long enough. Even with the reasonable premise that the average cracking time is half of the longest 10 trillion centuries is still long enough.
The problem with completely random password…and it’s not a trivial problem…is that sometimes you’ll actually have to type the darned thing for whatever reason…iOS timed out on your phone and TouchID needs reactivating, you forgot your phone and need to read your email or get into your DropBox from your grandmother’s iPad…whatever…and typing CorrectHorseBatteryStaple1234$# is a whole lot easier than typing dNvnkCq;B7.FjomBYPFejtFawwVf9mx is…even if you’ve got your phone you still need to read and properly type the random one and you might be using an iPhone keyboard instead of a computer keyboard which makes it even harder.
Even if a human picks the words instead of using something like Diceware or random pages from the dictionary…4 words are still fine. Unless the bad guy knows the names of your 3 children and cat…and is willing to modify his table to include the 4 of them as a single phrase in all possible combinations…AmyCarolRobertTiger is still 19 characters and takes1.3 billion centuries. 1Password gives that last one green status and about 80% of the bar is filled…and adding 2 random digits and symbols doesn’t make it any harder to remember or type.
Correct…”Four score and seven years ago” is a bad choice but “Goat tiger owl eagle shark cat” is perfectly fine even though all of the latter are animals…unless that complete phrase…and all other possible combinations of those words…is included in the dictionary. If they’re not the bad guys are on to brute force and length is really the only thing that matters there.
But tables of, say, 10,000 common words in all of their combinations? I believe that this is what hackers have been doing since the CorrectHorseBatteryStaple comic, precisely because it is better than brute force and because enough people started using this method.
I found this great explanation why Steve Gibson’s page miscalculates (though not deliberately) the amount of entropy in a password like CorrectHorseBatteryStaple:
https://treskal.com/kha/blog/2011/08/26/how-much-entropy-in-that-password/
1 Like
Even with 10,000 common words the size of the table might quickly get out of hand…even at only using 4 words…. Me…I would pick the words randomly from a complete dictionary which according to google has somewhere on the order of 750K words.
I tried to figure out the permutations of 4 out of 10,000…but couldn’t find a large enough calculator on the web…but I was able to find a factorial calculator and use those numbers to approximate it. For 4 out of 1000 it’s about 10 to the 12th combinations and for 4 out of 10,000 it’s about 10 to the 17th combinations. I assumed a max length of 30 for each entry and my very rough calculations got me to about 300,000 TB of drive space required to store the table. Even with some sort of data compression…I guessed a maximum compression of about 100 and this is still 3,000 TB of drive space needed…and at that size the access time obviously becomes an issue even if the hacker can afford 3,000 TB of drive space.
My point wasn’t really in the actual numbers for cracking time or entropy that haystacks comes up with but with their general magnitude…and it clearly shows that even with a miscalculation in entropy that length is what matters and it’s a really long time…even 16 characters it’s about 100 million centuries.
I’m not enough of a math expert to say that Steve miscalculated the entropy…and I’ve seen that comic before…but even it’s premise has some arguable flaws…for instance I’ve seen an entropy calculation page (not that one) that says if the password is “I have three dogs” the e in the word have doesn’t significantly increase the entropy. This might (or might not…I’m not a mathematician) be actually correct…but it really doesn’t matter as I understand it. Once you get down to brute force you’re no longer worried about each individual character in the potential password…you’re worried about the complete guess. You put the guess “aaaaaaaaaaaaaaaaaaaaaaaaaaaaa" in the front end and the password either breaks or doesn’t. If it doesn’t…you try “aaaaaaaaaaaaaaaaaaaaaaaaaaaab” and then “aaaaaaaaaaaaaaaaaaaaaaaaaaaac" and so on through all the complete permutations of the 95 character long set that is typically used. Once you’ve gotten all the way to the end of the 29 character long guesses (as in this example), you increment the length to 30 and start over.
There is no such thing as an uncrackable password…at least for all practical purposes given an unlimited time and cracking array scenario. All you need to do is make it long enough so that none of the easy cracking methods work…once you eliminate attacks based on dictionary (including phrases famous speeches, and the like), rainbow tables (which are similar to but not the same as a dictionary) and everything else then the only thing left is brute force and length is the sole key there. I don’t care if my password is ultimately crackable…just whether it’s crackable before the heat death of the sun…which is about 5 billion years or 50 million centuries from now.
Realistically…for anybody that the three letter agencies aren’t trying to hack…as long as it’s good for multiple centuries in the massive cracking array scenario…it’s probably good enough. My 1Password master password is more than 17 (that goes back to the way Windows NTLM and NTLMv2 passwords are hashed and stored) and is good for more than a trillion centuries…so if Steve is incorrect and it’s actually only good for a million centuries I’m OK with that.
A completely random password of the same length is much harder to type correctly on either an iOS or MBP keyboard…ask me how I know this…I used to have a completely random one and it took me weeks to before it became muscle memory and even then it wasn’t perfect. Heck…even my current one I type wrong about 1 out of 6 times the first try.
I generally use zxcvbn for password strength calculations — there’s a version of it at:
https://www.bennish.net/password-strength-checker/
It calculates the entropy of Y20WsEXrMYtlGZ74v#@1 at 104 and correct-horse-battery-staple at 63. It doesn’t attempt to calculate time beyond suggesting that both are “centuries.”
I don’t have a feel for how the calculators compare technically, though I had seen some criticisms of Steve Gibson’s calculator in the past, which is why I’ve stuck with zxcvbn.
Avoiding random passwords for typing reasons is important, but in my experience, only for a small number of important passwords like iCloud, Google, Netflix, and Dropbox.
Oh, I completely agree with you. I just wanted to point out that smart people trying to break a password do get smarter about it, and that Steve Gibson’s page isn’t necessarily correct in its assumptions, as it ignores the fact that hackers don’t necessarily brute force guess.
But none of that helps if the site you’re using has stupid password requirements.
I literally just had to reset the password with my credit card processing gateway – a site that should take security seriously – and they kept refusing my new password. They have so many password requirements it took me a while to figure out that my new password was my default 16 characters and they have a max password length of 15!
(This site also mandates you change your password every 60 days. Since I only use it about twice a year when there’s a problem, I always have to go through the password reset system, which is horrible.)
Seth,
This is really helpful. Thank you so much for making it so clear. The next time I change my passwords I will go with 4 or 5 words.
Betty
Betty Fellows
Platinum Plus Results
Phone: 650-364-4134 || Email: bettyfellows@yahoo.com
