I came up with the idea to generate passwords for internet sites by the following shema:



So the actual password is a hash with a reasonable lenght of characters (depends on the hashfunction, sha512 = 64 chars)

lets say my masterpassword is "ilovesecurity", the hostname is "stackexchange" and the tld is "com". So the plainpass would be "com.ilovesecurity.stackexchange" and the hash (the password I will use on the website):


Is this "method" secure and or practicable? If yes, what hashfunction should I use? If no, what can I improve? How can I eliminate the cons?

My pros and cons so far:


  • secure?
  • no password manager needed
  • "portable"


  • Need to generate the hash (by tool or online hasher)
  • Different password policys on websites
  • Secure? Not really anymore. Practical? Not nearly as practical or secure as using a password manager. Besides, what will you do for length and symbol requirements? – Alexander O'Mara Jun 3 '16 at 5:47
  • I will use a hashfunction that generates hashes with symbol. Sha512 was just an example – licklake Jun 3 '16 at 5:54
  • Ok, but some badly designed sites disallow them, while others require them. – Alexander O'Mara Jun 3 '16 at 5:55
  • These badly designed webpages don't have "critical" contents. So I don't realy need super secure passwords for them anyways. – licklake Jun 3 '16 at 6:01
  • 1
    So you wouldn't be able to ever change passwords with this approach? That is a major fluke. – Julie Pelletier Jun 3 '16 at 6:21

It's only as secure as the master password. Everything else relies on the the assumption, that the algorithm is not disclosed to an attacker aka security through obscurity. Oh, and you just disclosed your algorithm to the internet.

| improve this answer | |
  • Well its not. Wordlistatacks are not effective against this method. Also security through obscurity ist always a problem, password managers have that too. The "algorithm" is just an example of my idea to get your thoughts on it, i won't use that productive. – licklake Jun 3 '16 at 6:51
  • Actually, word list attacks would be effective against this method. Are you under the misconception that a "word list" is a literal list of single dictionary words that is simply tried one by one as-is? A "word list" also contains things that are not words but are likely (or actual past) passwords and passphrases. In addition, each word is tried not just by itself, but also mangled by various rules; one of which will definitely be mixing it up with the website name, in various orders, and possibly even interspersing the characters. – Ben Jun 16 '16 at 13:46
  • Also, you may be misunderstand the concept of "security through obscurity" when you claim password managers "have that too." A good password manager relies on a strong master password which you keep secret, similar to an encryption key, and modern, strong cryptographic techniques to keep your data secure. "Security through obscurity" means that if the attacker knows your method, you're doomed. Keeping a password secret is not considered "security through obscurity." Anyone can know your method there; they just can't know your secret password. – Ben Jun 16 '16 at 13:49

There are many password "managers" that works in that way. For example Masterpassword app is really close to your idea.

The algorithm is open source so I disagree with @RonWayn. As long as you cannot reverse the algorithm and find the master password with a given used password this should be secure enough.

| improve this answer | |

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