on 6/9/2014 7:04 PM David Krings said the following:
On 6/9/2014 10:44 AM, Jerry B. Altzman wrote:
on 6/7/2014 10:38 AM Gary Mort said the following:
A plea to anyone setting up a website where you will have users log on. Make
your default password rule something simple, like any 4 charectors.  A

password complexity system should allow for multiple tiers of rules with configurable default rule that is set, by default :-), to something simple. Tune those tiers and defaults based on your website need, not by blindly
implementing the preachings of the high priests of security.
That I agree with. Don't put Fort Knox security on a site that contains nothing secret. Then again, no matter how good security is, if it is really delicate info don't put it on the web at all.
It's all about your risk model.

http://bit.ly/1xxLQXJ (Link is SFW.)
Better yet: don't make users create accounts if they don't have to. Let them log in with FB, LinkedIn, Twitter, or Google accounts instead. The chances are
the user already HAS one of those.

I wouldn't count on people having this. Some places ask me to sign in with my FB account. I don't have one and the idea of expecting me to have one is rather obnoxious. I also doubt if it is wise to outsource security to a third party.

Sorry, I respectfully disagree. Of the several I mentioned, you claimed to only have one. You can offer the 'create your own account', but users should be encouraged to use some other account and use something like OAuth to provide user authentication.

The notion of "I don't have FB, therefore nobody should force FB auth" is equivalent to saying "we must absolutely positively backwards support IE6". This is 2014, sorry, if you don't want any social media accounts, that's your prerogative, but the vast majority of everyone else does.

And offer more options for the second factor. For example, I do not have a smartphone (yes, saves a lot of money every month). So unless you can figure out how to send an SMS to my landline forget it. In 2014 it should be possible to dial my phone and use voice recognition to confirm a pass phrase.
In fact, Sprint will do text-to-voice if it detects a voiceline (or at least it used to). But once again, we shouldn't aim towards supporting IE6 forever. We're also not optimizing the user experience for those using lynx...
Remember that you are not the world.

accounts most likely only need complex passwords[based on potential damage
of a compromised account...if a site manager can give out refunds and
credits for an e-commerce site, obviously you want to add extra security!]
Yes, for these things, you almost certainly want a second layer of
authentication atop the ones above. For these, little crypto keyfobs are
great. If the potential financial loss is large, the client should not balk at
the relatively small cost.

I agree, but in best US fashion the industry miserably fails at agreeing on a standard here. Then again, with any of these fobs you are authenticating the fob, not the person holding the fob. For that you'd need biometrics which is yet another can of worms.
Indeed: you are assuming that the user has both something-you-know and something-you-have. Biometrics isn't foolproof either, vis http://bbc.in/1oQshE4 (link is SFW).

More and more people just use "I forgot my password", and deal with it that way. Either you've exchanged the password for a security question, or just
access to a user's email.
That's because passwords suck! As do password managers which end up being the single point of failure (I do use them anyway). As mentioned above, it is sad that after over 50 years of client/server computing there is nothing better than and as accepted as user names and passwords.
User authentication is hard. Let's go shopping!

David

//jbaltz

--
jerry b. altzman | jba...@altzman.com | www.jbaltz.com | twitter:@lorvax
thank you for contributing to the heat death of the universe.

_______________________________________________
New York PHP User Group Community Talk Mailing List
http://lists.nyphp.org/mailman/listinfo/talk

http://www.nyphp.org/show-participation

Reply via email to