1mo ago

What is it with websites restricting passwords to 8 - 16 characters? Is there some technical limitation to their system??

It's infuriating to create a "strong password" with letters, numbers, upper and lowercase, symbols, and non-repeating text... but it has to be only 8 to 16 characters long.

That's not a "strong" password, random characters or not.

Is there a limitation that somehow prevents these sites from allowing more than 16 characters?

I'm talking government websites, not just forums. It seems crazy to me.

83 comments

It's a massive red flag. It implies that they are actually storing the password instead of a (preferably salted) hash and that they have no idea what good security practices are. Storing a hash leads to same size strings, no matter the length on the password.
- And there's no reason a database can't store a very long hash as well. Storage is cheap for this kind of thing.
- That's why I only store and compare the first 8 characters.
  
  Why not store the whole thing?
- They shouldn't be using salted hashes since a decade or more. Best is to use a memory hard password hash function like argon
  
  Can you expand on this? My experience with Argon is looking up a Wikipedia page in response to this comment, but it looks like it uses a salt as an input?
It's informative. It informs you that you shouldn't use the site, if possible. Because it's also suggestive of poor security practices in general.
- Yeah, imagine my shock and disappointment when encountering such limitations signing up for credit monitoring (by one of the big 3). It's not enough that my employer has a breach, no. But also finding out that one of the big players has some ridiculous 12 character alphanumeric password restriction. Absolute dogshit.
  
  A random 12-character password should take years to crack. But they're probably also storing it as plaintext, so no need to crack, just breach the DB (which is probably also insecure).
There are valid reasons to limit password length. For example when a hashing function is used that requires a lot of processing power and the amount of power required to calculate the hash is related to the length. In that (very common) case, a denial of service attack vector is exposed. By simply spamming insane long passwords into a login form for example, the servers calculating the hash get easily overloaded. Even with rate limiting, only a small number of attacking nodes can be used to pull down a site.
So a maximum number of characters for a password is a valid thing to do. HOWEVER the maximum length for this purpose is usually set at something like 2048 or 4096 characters.
There is no excuse for a max password length of 16, that's just terrible.
- There is no excuse for a max password length of 16, that’s just terrible.
  I get your point above, and the reason I hate short passwords is that I use passphrases. They are not only easier to type in, but long passphrases of 4+ words (plus a few extra characters and a number) are considerably more secure than the "best" 16-character password made up of random characters.
  Per your problem above, is this why some sites send you a 2FA code before asking for your password? To avoid that potential DOS attack?
  
  Yes in your specific scenario, you are righr. But if you even the playing field, apples to apples. If you have 4 words of each 4 letters plus random char at the ebd, lets say equating to 20 characters in total, a random 20 character password is better. Words/phrases are now commonly added to bruteforce attacks unlike before. Use an good password plus a 2fa that isnt sms or email for best protection, or dump passwords if you can for hardware keys.
- Sixteen is the minimum where I work. We upped it at the end of last year. Fortunately, we also fixed our password policy to expire annually. It used to be every three months, which leads to recycling.
  
  NIST recommended to never have passwords expire since like 3 decades. You gotta get rid of that. It makes your org less secure.
  Probably best to just fire whoever set that up. They're clueless
  
  There's always recycling. Or changing that final character from a 1 to a 2, etc. The human brain just cant handle the complexity otherwise.
- You could put a timeout on the hash function so that it can't be abused that way, but then... why not just make a limit so it can't anyway.
ive mostly noticed this on old systems.. where the field length for password was decided by an intern 30 years ago.
- This is it right here. The new system has to talk to the old database which has a character limit for that field. Untold amounts of money and effort would be required to update the back end.
  
  Too real, I know of a company that is changing a number from 8 to 9 digits and it's estimated to cost around 230m to complete. Insanity.
  
  Passwords should be hashed to a fixed length. Character limit implies clear text passwords are stored.
I had the opposite issue… how the heck do I need a username over 1000 characters? lol
- Because the requirement was to allow user names, the dev asked what the limit should be, the PM said, “I don’t care, make it 1000”, and so the dev did it.
  Source: I’ve been working in software far too fucking long.
  
  That's some malicious compliance right there! 😂
No, there is no valid reason to limit web passwords to lengths as short as 8 or 16 characters. If someone has built such a system with a technical limit that short, then what they have built is (from a security perspective) garbage.
Thankfully, NIST finally dropped their terrible password guidelines of the past in favor of sensible ones. Perhaps this will lead to fewer bad decisions being made in web development circles.
A few relevant sections:
https://pages.nist.gov/800-63-4/sp800-63b.html#usability-considerations-by-authenticator-type
https://pages.nist.gov/800-63-4/sp800-63b.html#length
https://pages.nist.gov/800-63-4/sp800-63b.html#passwordver
Obligatory xkcd:
https://xkcd.com/936/
(To be clear, this comic's approach to passphrases is sound advice.)
- Yeah, I usually limit passwords to 256 characters, because that's way longer than anyone needs and still short enough to not worry about overloading something.
  
  What are you worried about overloading?
It’s usually shoddy (or intended?) coding that only allows a 16 byte length for the password. One character equals one byte of memory so my guess is they only allocated 16 bytes of space for the password. The irony is NIST 2025 recommendations argue for AT LEAST 15 characters for passwords.
- One character equals one byte of memory so my guess is they only allocated 16 bytes of space for the password.
  This is true for storing text in general but passwords aren't supposed to be stored as text, they should be hashed. The size of the hash will depend on the hashing algorithm. In other words, if there's a database limitation for the size of a password, it probably means they're storing the password plaintext 💀
  More likely than not it's just some poorly designed validation
  
  Yea, you are correct. The database size should be based on the hashing algorithm.
- What's funny is a character isn't necessarily a byte now. It could be 1, 2, 3, or 4 bytes. Or only 2 or 4 bytes if we include utf-16 and 32. Character encodings are fun!
  
  Guess how many systems 'with full unicode support' are broken by using emojis in your username or password...
Irá usually bad backend design, bad frontend design, all made by people who are only vaguely aware of security, and how it works.
It's the same bunch that brought us "change your password every two weeks" and other insane anti security designs. They make it worse without even realizing it.
Do hope that your passwords aren't stored in plain text!
Yeah that's cause they're using ancient systems that probably store the password in plaintext. If you absolutely must use it make sure that password is only used on that specific site. I would strongly recommend looking for other ways before you do though.
- If you absolutely must use it make sure that password is only used on that specific site. I would strongly recommend looking for other ways before you do though.
  Not only did I use a completely unique and random password, but all the "security questions" (they asked for quite a few), were just random words for things like "Your pet's name".
  In addition to the shitty password requirement, 2FA is also done by phone or text. Like, come on!
  
  You should treat security questions like passwords and use strong alphanumeric passwords as the answers. Just make sure to store them in your password manager in such a way that you can remember which one goes with which question!
  
  At least provide email as an option, which might at least be TLS encrypted. If you're going to screw up security, at least make it something I can somewhat secure.
Probably running on code from before 2000.
There’s no good reason today and in the future, period.
There are “experts” who still claim these, but they are based on a very dated recommendation from at least 15 - 20 years ago at this point. To some, such non-sensical requirements (by the fact that we should be storing passwords as hashes today) have become doctrine, rather than any fact based in reality.
And some users have been conditioned into thinking that these are good security practices as well, because governments and banks still make use of them, and these are the very organizations that should be the best-in-class when it comes to security. Some of these users become CEOs or product designers with more say than their IT and security experts in the company. The rest is history from there.
Piggybacking on this because i do IT for a major US bank that requires passwords to be a set number of characters for its employees. I don't want to be specific, but it's in the range OP posted of 8-16.
- Are they still running Windows 2003?
  
  Windows 11 and believe me updates cause no shortage of issues.
- I'm sorry. Is there nothing you can do to fix it?
- Be the change you want to see
  
  Like an individual dev can decide to make random changes to bank systems lol
It means it’s stored unencrypted.
- You don't want your password to be encrypted, you want it to be hashed.
  
  I was going to say hashing is a kind of encryption but I can tell you're an engineer, so I'll simply link my feeble cite and go back to my corner quietly.
There was one point in time when Intel's website only allowed up to 14 characters and disallowed certain special characters. If I had to guess why, fear of inadequate error checking and fear of sql injection.
This is an interesting question. Honestly, I've never signed up with any government website.
This begs the question though, what country?
I'm from the USA and have little to no reason to trust signing up on a government website.
- This begs the question though, what country?
  Canada. We do a lot through our government systems: license plate renewal, address changes, taxes, benefits, etc.
  
  Here in the USA, I just prefer to show up in person.
  Hell, just 2 days ago I went in to renew my mother's auto tag. She wasn't even with me.
  Easy peasy.
IBM mainframe in development history. And they never went back to add hashing.

83 comments