Slashdot Mirror
Passwords Are the Weakest Link In Online Security
Orome1 writes "It's not surprising to find that 79% of consumers use risky password construction practices, such as including personal information and words. The recent Gawker breach and a detailed analysis of breached passwords show undeniably that passwords continue to be the Achilles' heel of the average Internet user. This insecure trend sadly doesn't shift as 26% of users reuse the same password for important accounts such as email, banking or shopping and social networking sites while 29% had their own email or social network account hacked, and over half (52%) know someone who has had a similar problem."
Users are the weakest link.
Mod me down, my New Earth Global Warmingist friends!
There's lots of buzz going around about the Gawker breach and discussions on how good/bad the passwords were. I looked at the websites that Gawker owned and most of them are tech websites, frequented by people that have some knowledge of security and computer systems.
I would assume that much of the readership is like myself. They know that access to their Gawker account is the most sacred and guarded of personal intrusions, and would thus treat security as the utmost important thing. My Gawker password was the ultimate in high security. It was a 280 character alpha-numeric password containing my social security number, all of my credit card numbers, my date of birth, my address, every password to every other website I use, plus all of my wife's data. That way I know that anyone who tried to crack my Gawker password could never do it, and all my information would be safe.
Wait, no, I got that backwards. Sorry, I used "cock" as the password for Gawker... probably. You see, if I were to log into Gawker, I would assume that the password was about as secure as writing it on the bathroom wall. In addition, I know my browser would remember whatever stupid password I typed and I wouldn't have to remember it for more than 30 seconds. Furthermore, if someone hacked it, and posted a stupid comment as "bullcrapgawkeruser222" I would likely neither notice nor care. If I did care, I would create "bullcrapgawkeruser223" with a password like "cockk".
Even more likely, if I ever commented more than once on any Gawker owned site, I probably just created a new account because I forgot I had an old one.
So, can we stop doing ultra-security analysis on what is probably a bogus set? Next I'm going to see an analysis on how insecure Masterlock combination locks are because the users don't use uppercase letters and punctuation.
And I would say that it's even worse when you can't type your question. Too many people know my mother's maiden name, my first car, my high school -- and I assume much of this information can be had publicly as well. If I were to imagine trying to get this information on someone, I'd just call them or their family, pretend to be some High School Reunion Committee, and say "We are celebrating the class of 1987 at Shrub High" and they'd probably go "Oh no, I'm graduated in 1992 at Rose Garden High". Then reply "Oh really? I guess you're the wrong Joe Blow, I'm sorry for your trouble, thanks bye."
Multiple attack vectors over one secure password, ridiculous. I think GMail at least does the semi-sane thing and instead of security questions, uses a phone number to verify you if you would ever lose your password.
And that's what is needed, identity verification if the password fails. Not a cheap way to do that in an automated and very dumb way.
There was, also for years, really dumb advice such as to never write a password down. That is unrealistic given the number of passwords someone needs to know today and leads to using the same password again and again. Now, you don't have to write it unencrypted, you could use Rot13 or, even better, some other code of your devising -- but it's better than keeping all this in your head in this day and age.
You obviously not had to deal with the average user. I run a web site that has accounts and many non-tech users and many people can't even understand the concept of password let alone asking them to upload a public key. I regularly get complaints that our site isn't "user friendly" because the person can't manage to even remember their username... so anything that is even slightly more complicated or involves something that they don't deal with in every day life it's right out.
Having the Web browser handle passwords is one way to address this. For a new site, I make a password in KeePass, store it in that database, as well as have my Web browser store it. This way, I don't have to bother typing it in, it will be of a decent character length (20 chars), and of random characters, and a blackhat that gets that password won't have access anywhere else I go.
Since my KeePass database syncs with my phone, if I'm using another computer somewhere else, I still have access to sites I go to.
This isn't the best of all worlds solution, but it does work.
How many places do need a login? Websites, computers, programs, ...
If all websites would use openID, that would solve already a lot. However many places give me my login and then ask me to change that every month. At work every first day of the month I change all my passwords. That takes me about 20 minutes.
So I have several passwords depending on level
1. Generic websites. Lowest security level (e.g. Pa55word)
2. Work related. These will change every month and will include some sort of year/month where only that part changes (e.g. 10Work12 for this month)
3) Provider related pass word for email and connection (Resused semi-random 8 charcater password)
4) Personal password for local system and openID and banking(Reused semi-random 8 carcater password. Different from 3)
5) Secure password for encryption, ssh and the like (Loooong semi-password of at least 16 characters.)
So the moment I am forced to change passwords where I used first 3 or even 5, I will go back to less secure of 2.
The main problem is that each security person treats their security as if they are the only one and treat security with the standard error. Solving a social problem with a technical solution. It is very hard to explain people that changing passwords every month will LOWER the security.
It is the nature of people to find the way of least resistance and as long as security people do not understand that, nothing will change.
I sometimes feel that it is not about security, but about reliability. Reliability is moved from the IT department to people who do not understand security, because they 'did something' and now it is not their issue anymore. That is why they also look only to the security of 'their' system and not at security as a whole.
Don't fight for your country, if your country does not fight for you.
Password Composer http://www.xs4all.nl/~jlpoutre/BoT/Javascript/PasswordComposer/ is what I use.
For example http://www.slashdot.org/ and my master password of buba yields a right(md5sum("slashdot.org:buba"),8) yields fc56e979
They have a static web form, a bash script, and a greasemonkey script. I have also written a delphi app that runs in Linux, Windows, Mac that I keep on my memory stick. So all I have to do is remember one master password, for example "buba". And with that master password every site gets a unique password that is hard to crack. I decided about four years back that if anyone ever hacks one password of mine or can fool me into revealing a password to them, that is all they get one password.
The ironic thing is the only site that I use a regular password that I came up with, that is related to me, that can be broken by a dictionary attack, is the one for my slashdot account. Still the same password I came up with in 1999 or 2000. I assume no one else would want to hijack my opinions.
vi +
No, the weakest link is the flawed authentication mechanics that requires you to use passwords in the first place. Bad password are just the natural result of that. If you want to fix the problem, you have to fix the way users authenticated themselves, not just chose a better password.
I have a mobile phone (two, actually). I also live in a hole in the ground (not quite literally, but close) that's a cell shadow with intermittent coverage at best, and zero signal a lot of the time. Your authentication scheme won't work there, and will also be spotty in my office, which is smack in the center of a building.
You people are determined to ruin my version of reality.
If libertarians are so opposed to effective government, why don't they all move to Somalia?