I joined this forum only a few days back. Likewise, I have accounts in several apps and e-commerce site. I was thinking whether my password was strong enough?
Just use a different password for each site.
As long as the password is RANDOM and over 8 characters, it is going to be pretty much impossible to crack.
I just spent several days trying to unlock a computer whose owner had died without leaving the password to anyone. Even though I had a pretty good idea of some words that he might have used, I gave up after trying several hundred million variations on those words, using a password-cracking program.
There are various techniques to do it and many of them don’t involve hacking your password at all. It is fairly difficult to hack an individual password in any reasonable amount of time if it has even modest encryption.
Most account breaches are done in much less sophisticated ways. One is just to steal an unencrypted list of them from a server. That has happened a bunch of times even to large companies and it is especially bad because people tend to reuse their passwords in lots of places. You can’t control the servers but you can make sure that you don’t use the same password(s) across many different sites.
There is also tapping into wireless networks in places like airports, bars and coffee shops. It is possible for security to be so weak that even semi-skilled hackers can intercept all your traffic including passwords. Don’t use those if you are paranoid or at least install an encryption solution of your own.
However, the most common ways that passwords are hacked is through simple phishing (being redirected to a site that asks you to give your password voluntarily), social engineering (someone from your “IT department” just calling and asking for it so that they can authorize some “repairs”) or just looking for Post-it notes that people leave lying around.
There is also the dumbest way of all. The top 20 passwords are dead simple and used by a surprising number of people. All someone has to do is run down the list and they will get into a significant percentage of computer, phone and other systems.
It is fine and even advisable to use moderately strong passwords that are unique to different sites and systems but you don’t have to go overboard. The vast majority of the real-world attacks can be defeated by some understanding of how the most common attacks happen and some common sense.
The thing is, a random string of 8 characters that is never re-used and never written down is impossible for normal human beings to remember. Oh, you can memorize a random 10 character string. But how many of them can you remember? And can you remember the new random 10 character string when you have to change the password 3 months later? And again 3 months later?
If you have 10 passwords, and each is a random string of 8 or more characters, and they are changed regularly, normal human beings cannot cope. They will be forced to either re-use passwords, or write them down, because the human brain just can’t cope.
Note that writing down your password is fine–as long as you keep the paper in your wallet, not taped to your monitor. A guy walking down a line of desks at night and looking for passwords taped to monitors is one thing. A mugger who steals your wallet isn’t going to care about your work email, and neither is a good samaritan who finds your purse on the bus.
If your security system relies on the users doing things that normal users can’t do, then your security system is MORE VULNERABLE, not less. The users will work around your precious security system, because otherwise they can’t do the work they need to do. Memorizing and forgetting random strings constantly is very difficult, and users won’t do it. So now where are you?
The problem is that XKCD’s technique is not as secure as he thinks.
There is cracking software that will create password candidates by combining words and characters in ways that humans are likely to do it initially. This software was used to crack some large percentage of the downloaded hashes stolen from Yahoo.
Random characters are the way to go. Just use a password manager.
Is this true? I have been under the impression that selecting six random words from a word-list using “diceware” would yield 77 bits of entropy, even given the attacker knows the method and word list from which the passphrase was chosen.
I definitely agree with using a password manager, and using randomly generated shorter passwords (especially since many sites limit password length to something less than a phrase), but you still need a passphrase to encrypt your password manager files, or PGP keys, etc…
Anyway, is the diceware approach really already broken?
The operative piece is that is the words and characters not be chosen “in ways that humans are likely to do it initially” – they explicitly need to be random, as in chosen by a dice roll/good random number generator.
The Diceware/XKCD technique is just noting that a sequence of four to six random common words is much easier for a human to remember than a password with the equivalent entropy in random characters.
(FWIW, my setup is a password manager with long random-character passwords for individual sites, and a diceware passphrase on the password manager).
Except if your password is four dictionary words it is breakable by a dictionary attack. First try all the dictionary words, then all combinations of two dictionary words, then three, then four. Yeah, that’s a lot. But it’s a much smaller space than if every character in your password was a random ascii character. It’s (number of words in the dictionary)^4, not (number of characters)^26.
Anyway, hackers aren’t breaking into your account by brute-forcing every combination of characters. I mean, if the NSA wanted to break into your account because you were a Russian spy, maybe they could do that. But they have better ways than that.
What exactly is the threat here? A guy in Russia who gets your credit card number and bank account information. He’s not getting that because he targeted you personally and guessed your password.
I think you mean (number of characters)^(password length) for the latter.
But any password can be broken by a brute force attack, such as the one you describe. The only way to defend against that is making your keyspace large enough that is infeasible.
The diceware word list is 7776 words long, meaning the keyspace for four-word passwords is of size 3.66e15. Assuming there are about 70 “reasonable” characters (upper and lower case alphabet + some punctuation) that is equivalent to about an eight-and-a-half random character password.
The difference is that eight random characters is about the limit of what a normal person can memorize, but four random words is almost trivial. Expand it out to six or eight words and it is not much harder for the diceware password, but a lot harder for the equivalent random password.
Very high profile people (politicians, celebrities, etc.) are likely targets for hackers trying to guess passwords. So a hacker will go to gmail and try to log in to their account with common passwords like passw0rd, 12345678, etc. Maybe they get lucky and happen on the right password.
The more common password cracking is done by stealing a user database from a website. For example, hackers somehow gain access to the SDMB servers and download the user database. The passwords are in the database, but they’re encrypted. So if my password is ‘reddog’, it would be stored as something like ‘ida81201’ in the database.
Once the hackers have the encrypted passwords, they try all password combinations to see if they encrypt to the same thing. So to guess what my password is, they try encrypting various values to see if they match. Something like this:
aaaaaaa encrypts to 182kda99
aaaaaab encrypts to 1kdi91221
…
But they start with common passwords, dictionary words, combination of words, etc. My password of ‘reddog’ would eventually be figured out when they get around to this:
redcat encrypts to di1009289
redball encrypts to eido101204
reddog encrypts to da81201 <<<< match
But they don’t care about breaking into my SDMB account. What they really want is access to valuable accounts like email, bank, credit card, etc. Many people use the same password across sites. They have my email from the user database they stole, so they use that and the cracked password to try to log into banks, credit cards, email, etc.
The advantage of using harder to crack passwords is that it takes so much time that the hackers eventually give up. But any password can be cracked in time. To be extra safe, make sure you don’t share passwords with any of the important sites. It doesn’t matter a whole lot if you have the same password among various forums like this. But you want your important accounts to each have their own, unique passwords. That way if a password is cracked, the hackers will have limited or no access to your important accounts.
This is a really important point - I tried and failed to get this across to a fellow attendee at a security conference a couple of weeks ago - he was convinced that all he really needed was one really clever password that could be used everywhere*.
If someone gets hold of your facebook credentials, they’re going to try the same credentials on other sites, including sites like Amazon, eBay and PayPal, where they can spend your money, or other social sites where they can glean additional personal information that can be used for full identity theft such as applying for credit in your name, etc. It’s really important not to use the same password in multiple places that each have different fragments of your identity recorded in them.
(*His definition of ‘clever’ wasn’t too clever either - he was talking about making it out of identifiable personal information such as the name of his first pet, plus the birthdates of his two kids)
Re humans can’t cope: absolutely. These days you absolutely MUST use something like a vault to remember them all.
Writing it down and keeping it in the wallet would, in my case, result in my needing to carry a second purse - I have a LOT of passwords for both work and personal use. For someone who only has a few, that option might work well.
A diceware approach (I gather that’s the one where you use 4-5 real words) may be easier to remember but again, if you’ve got a lot of logons, you’re looking at having to write them down or use the same phrase all over the place.
Personally, I use a password vault, with a master password based on a phrase that has meaning only to my spouse and myself, with a mix of upper/lower/numeric/special. So we only have to type that password (which would take a billion centuries to crack). The one frustration with the vault is that some passwords I have to use multiple times a day do NOT allow copy-and-paste, but force me to manually type in the gobbledygook password.
Ramblings below, as I got interested in the topic and started playing around.
I’m not a security expert so I had no idea how much difference it makes having “mackdonna shoehorn butterhorse” or even “m@ckd0nn@ sH03horn b1tt3rhors3” versus “$^!)$JXBVUSD)(#2U$IFY&3ADS” or “asdkjhf gdkvud sjdhfaegflh hfd”, what with the tools to try brute-force attacks.
So I plugged these into GRC's | Password Haystacks: How Well Hidden is Your Needle? and got the following estimates (for the offline fast attack scenario):
For the one with pure gobbledygook (26 characters): 2.08 thousand trillion trillion centuries.
Adding some more gibberish to that to make it 30: 6.90 hundred trillion trillion trillion centuries
For the first one (a phrase from a Dope thread, 30 characters): 4.32 hundred million trillion trillion centuries
For the leetified Dope thread phrase: 6.90 hundred trillion trillion trillion centuries
The 30-character all-alpha garbage phrase: 4.32 hundred million trillion trillion centuries
So to me, it looks like purely a combination of the mix of characters (alpha only, upper/lower, numeric, special) and the length. The benefit of using a string of words (leetified as appropriate) versus something really random is that it’s a whole lot easier to type. It was interesting to play around with the time differences with shorter words just with substituting numbers / switching case. I plugged in gibberish of the length I frequently use and it would take 6 days in a massive cracking scenario.
What I’m wondering though, is how useful such cracking is beyond using the common lazy-person passwords. If the hackers have gotten the hashed password file e.g. MamaZappa at Straightdope Dot Com with hashed password 123789vkcdsfgiufgdgdfg, and want to spend the time finding a password that translates to that hashed value… then in a week or three they figure out that “mackdonna shoehorn butterhorse” was the source value and they start logging in everywhere and ruining my life. Seems like a WHOLE lotta work for a single password. They’re better off trying “password”, “123456” or “mamazappa” which might give them a hit in a few seconds. If that fails they go on to try the next person.
Or maybe they do that to get the low-hanging fruit, and meanwhile have a separate process generating all sorts of random passwords and see which ones are found in the millions of hacked hashed passwords they’ve stolen.
But in xkcd he’s not claiming (number of characters)^26. He shows each word as having 11 “bits of entropy”, (thus presuming a dictionary of 2048 “common words”,) and then biquadrates that to get 2^44.
If he were calculating by number of letters, then it would be 26^(number of characters), in his example 26^25, or 2.3*(10^35), which would have well over a hundred bits.
