2

About a year ago, I tried using LastPass but found the user experience very lacking, and went back to my tried-and-true method of just having a bunch of different passwords (some password reuse, but I've never ran into problems with it) and my email has a unique, non-reused passphrase of 30+ characters and I reckon I could just password reset any service eventually hijacked.

I know this is not good, and I understand the consequences - this is not about that.

I recently decided to "hey, let's give password managers another shot" and this time went with Firefox Sync (& Master Password).

From what I've noticed, it encrypts the locally stored cryptoblob using a key, which is in itself encrypted using the master password. The passwords stored on their server is (for some reason) encrypted using my Firefox Sync password (which is, of course, stored in the cryptoblob).

Full disclosure: My Firefox Sync password is not as strong as my master password. If someone stole it, they could theoretically download an unencrypted copy of my password blob to a new device simply by logging into my account. However; this requires me to confirm the login from my e-mail. The password to my email is, at current date, the same passphrase I use for my firefox master password.

Now, to the question: Is there something inherently less secure in using a long, humanly memorable passphrase (my master password) for my e-mail (which is required to download my cryptoblob) than having it be a data-secure password stored in said cryptoblob?

3

It depends on how you generate your passphrase.

A passphrase derived from a story, or a song lyric, or a book or poem quote, is inherently insecure. Crackers are already using quotes from publicly available works, and common transformations of them (like taking the first letter of each word) to guess passwords.

A passphrase you made up yourself is better, but it is still inherently less secure than a fully randomized password. It will have sentence structure, probably will favor commonly chosen words, etc. How much less secure is very hard to quantify; there have been studies that figure out entropy of passphrases character-by-character. Still, it's a better than a publicly available phrase, and a lot better than a single-word password with or without transformations.

But you could also make a fully random passphrase, Diceware-style. With enough words, these can be made as secure as you like; a diceware passphrase with N words tends to be about as secure as randomized, mixed-case with digits and special characters password with 2N characters. With a random passphrase, there is nothing inherently insecure about it; you can use this method whenever the service allows enough characters for 6-8 words.

  • I think your 2N is off by quite a bit. A good random passphrase dictionary will have ~2000 common English (or whatever language) words in it. Assuming the attacker has the same dictionary, this is way more than the <100 characters commonly available for a random password (letters, numbers, symbols). I'll let someone else do the math, but if you believe XKCD got the math right then a 4 word phrase will have about 44-bits of entropy, which is more than enough for almost anyone (and I suspect significantly more than an 8 character random password). – Micah Zoltu Jan 24 '18 at 22:09
  • The XKCD strip you mention does get the math right for a 4-word phrase. But you're underestimating how quickly exponents add up. 100^8 is a one followed by 16 zeros. 2^44 is slightly less than a 2 followed by only 13 zeros. – Ben Jan 24 '18 at 23:02
  • 1
    To be more precise: with a 7776-word dictionary (standard Diceware), 4 words has entropy of about 51.6. A randomly chosen "all keyboard characters" password of length 8 (assuming 95 characters available) has entropy 52.6. – Ben Jan 24 '18 at 23:07
2

Yes, it can be exploited to gain access to your passwords easier than if you had a stronger password securing your Firefox Sync password. You're adding potential for an attacker who is able to figure your Firefox Sync password out.

The email you receive to allow yourself to download the crypto blob isn't encrypted. It can be read by others. This potentially (Emails can be encrypted between providers, too, not just between your provider and you. However, they often aren't.) includes the big governments, the government of your country, your ISP, any other service provider in-between your ISP and the servers of Firefox Sync, and their respective governments. But it also potentially includes everyone to whose wifi or Ethernet you connect while receiving the email and the people who control the malware which may be present on that network, for example in the router.

Note that some of the above can be mitigated if you receive your emails encrypted via SSL. However, SSL has a ton of problems.

Let's just assume that the private key of some arbitrary anti virus company is compromised. Those can just intercept SSL traffic if they come across it which makes for example the attack by malware on the router or by your ISP a lot more plausible.

Furthermore, your email account may have a strong password but exhibit other weaknesses. For example, there may be a password recovery question simply asking for your mother's maiden name which is really easy to find out if the attacker knows who you are. This renders your email account's password strength pretty much useless. In addition to that, there may be a recovery email address assigned to your email account. If the account of that recovery email address has a weak password of exhibits other weaknesses (e.g. recovery via mother's maiden name), that also means your email account is insecure.

  • Thank you! I hadn't really considered snatching the authorization email. It's been TLS encrypted end-to-end, but it's still worth thinking about. The question was really if "A strong passphrase you know is less secure than a crypto you dont stored with a master key", but your answer is still very much appreciated. – Wolfie Oct 11 '16 at 12:51
  • Sorry, I don't understand what you mean by 'if "A strong passphrase you know is less secure than a crypto you dont stored with a master key"'. – UTF-8 Oct 11 '16 at 13:02
  • Trust me - it's hard to explain on my end, too. The base question was if encrypted passwords where you use a master password to decrypt them are inherently better than just using the master password for login in the first place – Wolfie Oct 11 '16 at 13:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.