Slashdot Mirror
Smarter People Don't Have Better Passwords, Study Finds (bleepingcomputer.com)
An anonymous reader shares a report: A study carried out at a college in the Philippines shows that students with better grades use bad passwords in the same proportion as students with bad ones. The study's focused around a new rule added to the National Institute of Standards and Technology (NIST) guideline for choosing secure passwords -- added in its 2017 edition. The NIST recommendation was that websites check if a user's supplied password was compromised before by verifying if the password is also listed in previous public breaches. If the password is included in previous breaches, the website is to consider the password insecure because all of these exposed passwords have most likely been added to even the most basic password-guessing brute-forcing tools.
Why does this still work? I would think we would have adjusted things years ago so that once a wrong password is tried like, oh, I don't know, say 50 times the account is locked. Or don't allow more than one attempt per second. Something along those lines.
This assumes that higher GPA means smarter. While this may generally be the case, this is far from a foregone conclusion. Smartness or intelligence is a complex subject, and the measurement of intelligence is not something that is trivial and universally accepted. A different study that has access to other measures of intelligence – such as standardized aptitude tests – to combine with GPA may yield further insightful result.
no, study only proved the students don't give a shit about the security of their school's email accounts. maybe smart people give even less of a shit about things that are extraneous in life than "normal people"....
The college that conducted the study is in the Philippines. The experiments were run against the college's student email accounts... which does raise a few easily-dismissed ethical concerns, but I digress...
There's really no reason to assume the USA would be involved at all, other than the reference to NIST, which isn't too surprising. Many places refer to NIST standards, just to avoid a certain standardization problem.
You do not have a moral or legal right to do absolutely anything you want.
Use a password manager, and you never need to remember what rules were in use where.
You do not have a moral or legal right to do absolutely anything you want.
I wouldn't expect intelligence to factor into strength of passwords.
I agree with you up to here.
Instead, I would expect password strength to correlate to paranoia - people who think it unlikely someone will try to use their account will use a somewhat weak and easy to remember password...
While I don't specifically disagree with you here, perhaps a better correlation can be found by looking at cognitive burden. That is, while some people likely use the paranoia factor to motivate them to use/remember long and complex passwords, I suspect that most people think along the lines of, "I am just not willing to burden my brain with yet another long and complex password for blah blah blah."
That is not to say that cognitive burden is the only determinant, since things like organizational policy (e.g., in a school or business) might set and enforce minimum complexity with which the user must cope. Rather, in the absence of a forced minimum, users will employ the simpleest password they can comfortably get away with. Where comfortable is different for each individual.
they looked at grades, which is a dubious measurement of intelligence at best.
verifying if the password is also listed in previous public breaches
So does NIST recommend maintaining an offline archive of every breach ever or are they recommending you transmit the password in cleartext to a 3rd party?
... than the 'other' people. Smart people tend to think for themselves, to ignore common beliefs and behaviors. Smart people are like cats who are difficult to herd. If the gospel among computer users is to have an obscure password, smart people will question that and may do so only for special accounts.
The 'other' people, OTOH, tend to do as they are told, to follow the rules, to behave themselves. If they are told to use safe passwords, and they can remember that rule, they will make some effort to do so. Those 'other' people are like dogs- they will do as told if they understand and remember the rules. We all like dogs, but not everyone likes those smartass cats.
...omphaloskepsis often...
A similar phenomena would be "security fatigue" -- the sense that it's either all pointless, or that as security measures grow more complicated, the costs exceed the benefits for more and more situations.
Strange things are afoot at the Circle-K.
While I don't specifically disagree with you here, perhaps a better correlation can be found by looking at cognitive burden
I think this is probably a better take on it than I had. I agree that cognitive load is a large factor on what I personally end up using for password strength, after the fiftieth password you are just like "screw this, using password pattern 1".
things like organizational policy (e.g., in a school or business) might set and enforce minimum complexity
The funny thing about this to me is that it greatly increases cognitive load of passwords, making the password a little stronger because it enforces say a special character, but across the board makes it VASTLY more likely I will choose the same password across multiple sites because otherwise I cannot remember what I chose and don't want to have to think of a new complex password.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Let me get this straight. So the NIST is saying that when a new user creates an account on a site, that site should immediately shuffle a copy of that password off to another site where it can be compared tona list of passwords on that site.
That sounds a little shitty. When I sign up for an account somewhere, the password I create and give them shouldn't be passed around to other entities. It sounds like a great opportunity for somebody building a password dictionary to log a copy of everything that's being sent to that site.
Did they run a brute force attack against their own email system, or does the college store passwords in plain text?
The real "Libtards" are the Libertarians!
> Read about rainbow tables
Good advice. You should take that advice. Maybe even try using one.
Let's look at your claim regarding the length of the password. Back in the early 1990s, MD5 was the recommended algorithm. It had a short 128-bit hash. That's roughly the same entropy as an 18-20 character password. As long as two passwords were both at least 20 characters, a longer password wasn't better because they'd both get reduced to a 128-bit hash anyway. By the late 1990s weaknesses had been found in MD5 and we started recommending SHA-1 instead. I personally distributed sample code showing how to convert your MD5 password hashes to SHA1, something that sounds impossible at first.
Then about 15 years ago MD5 was completely broken. Anyone with a clue moved to SHA1 or, later, SHA2. IF your web application is using an algorithm that has been broken for 15 years, AND your pass is at least 20 characters, longer than 20 isn't much more secure.
You might be thinking "there is a four character password with the same hash". No, there isn't, in all likelihood. There are very few 4-character passwords, and very many possible 128-bit hashes. For any given long password, there probably is no short password with the same hash.
SHA-1 is a 160-bit hash. It's even less likely that a short password of say 36 bits entropy is going to have the same 160-bit hash as a longer password. ALL possible 36-bit passwords combined only cover 1/2^124 of the outputs. In other words, the odds against getting a match, even trying ALL of the short passwords, are far less than the odds that you will win the lottery without even playing, by finding a winning ticket.
SHA-2 came out in 2001. There are no rainbow tables for SHA2, because the key space is too large. So if your application has been *properly* updated in the last 10-15 years, rainbow tables simply do not apply.
I advocate passphrases for their reduced burden. Increased complexity will statistically cause a variety of other behaviors. Any low entropy (patternlike) behavior can be emulated, and we spent 10-20 years teaching poorly, increasing the complexity tax for tiny entropy gains.
superman is weak. We get that.
Superman1! is just as weak. Fuck everything that has ever suggested otherwise.
$up3rm@n is almost as bad.
zxcvbnm is not strong. Neither is qrafzvwtsgxb. We see what you did there. Obviously, anything using those patterns is a form of low entropy, with days numbered.
rrrybgdtsmmmmlibad is a nursery rhyme and adds zero (mental recall) burden. Does your kid like Bob the Builder and Spongebob? cwfiwliap. These even dominate the brain tax required by correcthorsebatterystaple, which is frankly a four-letter password in a bigger alphabet, begging for pwning.
It's possible that futuristic tech can be powered by mass dumps (probably lifted from google) of our written text (such that iwtbotiwtwot from dickens or tbontbtitq from shakespeare become weak) but phrases are likely to lean on personal interests, contemporary events/pop culture. Leverage a single unfolding point of mental recall for your brain, not a pile of them. Which are adhering to a heap of rules that will only yield brief, first-derivative resilience.
Bottom line, with the way things stand today, a password manager is the only viable option for anybody that has even the slightest concern about security.
What happened to a gool ol' plaintext file with logins and passwords, stored in an encrypted container?
Every end has half a stick.
He was working in the Manhatten project making the first atom bomb. Place was teeming with top physicists absent minded professors and was run by the Army that had safes allocated to all top scientists. After a long and interesting story about how he got into safe cracking, he mentioned: He was challenged to crack the safe of Niel Bohr or Oppenheimer. He did it in less than two minutes. Asked how, he replied, "Physicists always use 3141, 1414, 1783, or 2245 as the code". They are PI, sqrt(2), sqrt(3), sqrt(5)
sed -e 's/Chuck Norris/Rajnikant/g' joke > fact
Grades are also measures of conformance and memory which are both actually crutches that can reduce the degree of developed intelligence.
It is interesting that the only intelligence required here is the intelligence to know that memorable passwords are a security risk in general and that proper password security requires the use of a well-protected password vault and automatic password generator so that no compromised site will ever reveal the password you've used on any other site.
Yes, I get your point.
- Parent poster points about rainbow table (tables that point hashes back to strings that can generate the same hash).
- You point that a well designed (=non borked design) hashing function should give two different hash for two dissimilar short passwords. Thus you would need a giant rainbow table that gives a password for *every single possible 160bit hash* (that's ~10^48 entries, i.e.: within an order or two from the number of atoms on earth). Fat chance.
BUT!
Even if the hash->password direction is *hard*.
password->hash direction is easy.
All the algorithm you mention (MD5, SHA1, SHA2 and let's throw SHA3/KECCAK in the mix too) are all extremely *fast hasing functions* (They are ultra fast, and have very low ressource requirement by design).
Meaning you can take a GPU running a special compute shader/OpenCL/Cuda code that can process millions of them in a second.
So you could scan through ALL the common password (based on frequent leaked passwords and/or on frequent paterns, etc. and their substitutions) within a reasonable time until you find a match.
As the summary points out, we humans are bad at picking-up password, we definitely use less than 2^ ${whatever bits used by current popular hash} different passwords.
Even if you use salt (so your hash doesn't match any other precedent hash in any rainbow table), and even if you use the latest *hashing* function (SHA3 - well okay, it's a sponge function, but basically works the same), it's definitely within the reach of a reasonable budget to loan GPU compute time on the cloud and brute force the passwords.
So if a database containing SHA{n}-hashed (and optionally salted) passwords get leaked, you can consider that all except the most unusual passwords can be brute-forced.
So in short DO NOT USE HASHES. USE KEY DERIVATION FUNCTIONS.
Things like bcrypt, scrypt or the current competition winner argon2, are on purpose designed to be slow and resource intensive.
(By iterating multiple rounds, by require significant memory, etc.)
For you, it doesn't change much if loging in take now a third of a second - you only log once, after all, it won't kill you to wait for 300ms just once at the beginning of your session.
But for potential brute-forcers, not being able to quickly go through million of tests is suddenly a huge blocker.
So in short :
do NOT use SHA2 for your password database.
use bcrypt/scrypt/argon2 instead.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
> do NOT use SHA2 for your password database.
use bcrypt/scrypt/argon2 instead.
Brypt is indeed designed as a *password* hashing function, so it's better for passwords than sha-2 is. I think sha-2 is also acceptable.
> Thus you would need a giant rainbow table that gives a password for *every single possible 160bit hash
That's actually the difference between a rainbow table and a simple lookup table. The rainbow table can be as big or small as you want. Larger tables allow faster "unhashing". While BUILDING the rainbow table, you have to compute all* the hashes, but not save the results.
> So you could scan through ALL the common password (based on frequent leaked passwords
Definitely don't use a common password. Using "password" as your password will suck no matter what else you do. Therefore it's a mistake to say:
Doing X won't help if the passwords suck. Therefore don't do X.
Because we know weak passwords will be weak no matter what you do, it probably makes sense to try to make the system as secure as possible +for users who use good passwords+.