Maybe you could also mention StrongPhrase.net when suggesting ways to generate random passphrases?
Seems pretty uncontroversial to mark this as approved and accepting PRs 
I donât know about recommending a specific password generator necessarily, but I think that warrants its own separate discussion.
4 Likes
Not really sure what is the benefit. Just use your password manager for this.
1 Like
I should make a new topic asking about StrongPhrase, but here are some of the strengths it has compared to most other password generators
Today was my first time seeing StrongPhrase.net, and the concept is great.
With that being said, I disagree with allowing a third-party to provide or even recommend passphrases. For example, clearnet users could have their IP address logged with a list of passphrases they were shown or copied from the page.
Even as a Tor user, I do not believe that is a good practice.
1 Like
Password managers have option to capitalize words, add numbers and special characters, so even a 3-word pass can be a bit stronger. But yes, 4 words should be recommended as a minimum. Problem is, there are services that limit passwords to 20 characters, and itâs almost impossible to have good passphrase in that case.
And I see passkeys are mentioned 8 times in that text, and 2FA only once, which tells me someone has a horse in this race 
The benefit would be for certain things like your phone password that you canât really use your password manager for. For those cases having a minimum number of words is important. I think 4 words is too low though, the EFF recommends at least 6 words.
1 Like
why canât you use your password manager for that?
I completely agree, the recommendation should be 6-10 words from the EFF Large Wordlist. This is better in line with what other experts recommend.
3 Likes
Obviously you canât get your password manager open if your phone is locked, and whatever device your password manager is on would need to be running that would be attacked and they would get your encryption password that way. Better to memorize it, thatâs the whole point of diceware anyway, to have a memorable passphrase.
3 Likes
Depend on what keyboard you use, you might be able to use emojis or UTf-8 characters, which could produce strong password without making it too long
Yeah and you can create those easily with your passeord manager⌠honestly not understanding your point.
For creating them? Sure, but maybe the password manager doesnât give a good minimum or maybe it doesnât have a minimum length at all. Also not every password manager even has a diceware generator.
Isnât using more than 7 words from the EFF large wordlist providing more entropy than an average person could reasonably need for a passphrase? Unless the attacker is a state agency, 90 bits (7 words) of entropy is fine for pretty much anything, while 128 bits (10 words) is for if you want to future-proof. It depends on the person, but personally for me, a 10-word âEFF passphraseâ is quite hard to remember and cumbersome to type. Especially when some of the words included are pretty rare and unconventional.
This is partially why I also think that StrongPhrase does serve a purpose. It is true that password managers are capable of creating passphrases for you, and StrongPhraseâs system is less efficient compared to the EFF wordlist. However, even a 12-13 word StrongPhrase passphrase is much easier to remember because the words are more common and easier to type because it flows like a regular sentence would.
Arguably yes, which is why it makes sense to give a recommended range. Even with the most extreme threat models you reach a point of diminishing returns where the entropy of the passphrase exceeds that of the algorithms being used.
For example, Android uses AES-256 for encryption so there is absolutely no reason to have a passphrase with >256 bits of entropy since you AES-256 will only accept a 256 bit key which is derived from the password using a KDF (key derivation function).
Personally I use a 10 word passphrase (~129 bits entropy) which will remain theoretically secure for the foreseeable future. I would argue more than that is basically completely pointless.
5 Likes
For iOS and Pixel I was under the impression a numeric PIN was sufficient due to the nature of Aseries/Tensor. Based on this thread that is not true so Iâll be reading closely. Glad this is being discussed, and I may need to revisit this.
Your understanding is correct, the vast majority of people are perfectly safe using a 6-8 digit PIN on devices with the Titan M2 or iPhones since the 12 (I think). Itâs just that then you are depending on those secure elements not being compromised in the future. They are holding up extremely well so far though so itâs really not something I would worry about unless you have an extreme threat model.
Further reading:
https://xcancel.com/GrapheneOS/status/1791837791887729143#m
2 Likes
Recommending a strong password is good, but the choice of any specific number of words is arbitrary. Here alone in this post suggestions range from 4 to 10 words (52 bits to 129 bits). Itâs meaningless without an appropriate threat model. Secure against what? Against basic mass password cracking 4 words may be enough, but 6 words may be on the low end against law enforcement (the threat model in the article), while 10 words is arguably overkill for even the highest threat models.
The study referenced by the article is about improving digital forensics and is not cyber security advice. It âexplores rule-based optimization strategies to enhance the effectiveness of password cracking while minimizing resource consumption.â It states the following:
Results indicate that while three-word passwords provide improved memorability and usability, they remain vulnerable when common word combinations are used, with up to 77.5% of passwords cracked using a 30% common-word dictionary subset.
Note that:
- It is already known that 3-word passphrases can be cracked. The study is about doing so more efficiently.
- The study is not done by law enforcement and the methods are not limited to law enforcement.
- The choice of 3-word passphrases is based on the UK National Cyber Security Centreâs (NCSC) three-word password guideline. It DOES NOT state that 4 or more words are secure.
- The passwords are NOT RANDOM. The passwords cracked were either part of previously breached passwords (eg Rockyou) or part of a user survey where participants were asked to create a password composed out of three dictionary words of at least eight characters each.
You disagree with the existence of websites like Strong Phrase and Get a Passphrase?!
That doesnât make sense to me. Youâre implying that those generators are not safe. That is a bold claim.
I would also argue that using your password managerâs generator is also using a third party. Anything short of you coming up with your own passphrase is using a third party. And we all know how terrible of an idea it is to come up with passphrases on your own.
Moreover, Strong Phrase doesnât have word separators (a recommendation I made) or special characters. Most websites require that you include special characters in your password. We now know thatâs not the best of recommendation, as the length of a password is more important than the variety of its characters. And yet many websites still wonât allow passwords longer than 16 to 20 characters, but that is beside the point.
My point is, it is fair to assume that anyone in this community who uses passphrase uses special characters, because it is a requirement from most websites. Doing so, makes their passphrases different from the one they generated.
But even if they were identical, I donât think it is unsafe to use websites like Strong Phrase.