News for hackers. Stuff that matters.
Re:Let’s talk about usernames by kbenson on Monday February 12, 2018 @02:19AM (#16357089)
> then realized I could just generate my entire username and have been doing that since, when I don't care about identifying myself to others. My password manger saves it, so it really doesn't matter to me.

Even better, if they have your email and use it for password recovery, you can basically turn it into a two step authentication by not saving the password and using password recovery every new time you need to log in. Though, that can get annoying if their password recovery takes a while to send.

Re:Let’s talk about usernames by rando444 on Monday February 12, 2018 @02:52AM (#16357203)
Creating two-step authentication for yourself, but allowing attackers to have one-step authentication... interesting concept.
Re:Let’s talk about usernames by fbdjskajxb on Monday February 12, 2018 @03:16AM (#16357273)
If you are making up random usernames on a banking site, attackers aren’t going to know your username. So it is two-step.

This is not a great idea if you have a public profile connecting your username to your email because someone can hack your email.

But you not knowing your password doesn’t hurt your security as far as I can tell.

Re:Let’s talk about usernames by rando444 on Monday February 12, 2018 @07:14AM (#16358133)
Assuming an attacker can't know your information is not a good idea.

Your login information can be gained via keyloggers, network sniffing, phishing scams, malware, malicious employees, and all sorts of other methods..

This is why two-factor authentication is so important, to help prevent your account from being compromised in the event that your username and password is.

Re:Let’s talk about usernames by fbdjskajxb on Monday February 12, 2018 @02:01PM (#16360998)
The part I don’t get is how not knowing your password makes the situation worse. The password recovery mechanism exists whether or not you use it every time you log in.

The way I see it, not knowing your password removes some potential threats around managing that password incorrectly, at the cost of increasing the risk of losing access to your account if the recovery mechanism doesn’t work.

Re:Let’s talk about usernames by rando444 on Monday February 12, 2018 @06:40PM (#16363437)
It doesn't make the situation worse, you're the only person suggesting that.

It offers some extra security, but very little. It's the digital equivalent of locking your back door but not bothering with the front door.

Re:Let’s talk about usernames by fbdjskajxb on Monday February 12, 2018 @11:00PM (#16364637)
>It doesn't make the situation worse, you're the only person suggesting that.

The comment I originally responded to seemed to think so.

Re:Let’s talk about usernames by tjoff on Monday February 12, 2018 @07:51AM (#16358264)
If you are making up random usernames on a banking site, attackers aren’t going to know your username. So it is two-step.

The username isn't supposed to be a secret, social engineering will most likely be very easy.

Re:Let’s talk about usernames by fbdjskajxb on Monday February 12, 2018 @02:01PM (#16361010)
Agreed, but this is true whether or not you’re using parent’s plan of not knowing the password.
Re:Let’s talk about usernames by kbenson on Monday February 12, 2018 @03:39AM (#16357365)
As your sibling comment correctly inferred, this would hopefully be done after setting some sufficiently hard random password on the service in question. At least that's how I've seen it described by those here that have mentioned they do it.

A sufficiently large random unknown password is actually significantly less likely to be brute forced than the service itself is to be exploited.

Re:Let’s talk about usernames by hawski on Monday February 12, 2018 @04:32AM (#16357552)
I'm toying with an idea of a SMTP proxy service so you could have a SMTP server on your smartphone. I would like to use Let's Encrypt certificates so my service would be just a dumb pipe. Then you could register using random email for every service. SMTP would handle signal outage as message would be eventually received. Or one could set up secondary SMTP server in DNS. I would like to give an option of subdomains handling. So every email would be on different subdomain. That would make blocking easier.

So you would have addresses like:

[email protected]

It would only be for receiving.

I know that there are lots of options already: 20 minutes mail, [email protected] or just a catch-all. But this could at least provide somewhat end to end encryption.

Re:Let’s talk about usernames by mseebach on Monday February 12, 2018 @05:26AM (#16357775)
The idea of having easy to use single-use email addresses is good, but I don't think it's practical to run an SMTP server on a smartphone (you need the relevant ports open and forwarded through various NATs) and email anonymisers are already banned from signup on many sites, so *.emailproxy.net would quickly be, too.
Re:Let’s talk about usernames by chii on Tuesday February 13, 2018 @01:46AM (#16365218)
it'd be cool if you could use an ip as the mail domain, and simply having an (easy to use) app running locally allowed that IP to receive email.
Re:Let’s talk about usernames by cuckcuckspruce on Monday February 12, 2018 @10:37AM (#16359253)
If you're serious about this, consider running UUCP as the mail transport. It can be tunneled through SSH easily (or have TLS applied to it with something like stunnel), allows either end to initiate a transfer of data (assuming open ports), handles dynamic network addresses easily, and will likely be a much smaller drain on mobile phone batteries.

Plus it can allow sending and receiving of email and files, if you so choose.

Re:Let’s talk about usernames by hawski on Monday February 12, 2018 @11:38AM (#16359771)
I will look in to it.

I'm serious but my todo list is quite long.

I was thinking about dedicated application on phone for it instead of regular mail client. Mainly to provide easy interface to manage big amount of accounts, banning hosts etc. Also then the app could just use service like Google's Firebase Cloud Messaging. That would wake my app and then it would get a message. I hope that it is fast enough that a sending SMTP server would not timeout.

Re:Let’s talk about usernames by kbenson on Monday February 12, 2018 @12:52PM (#16360377)
Backup MX servers do this every day. One of the commonuses of differing MX priorities is to provide a backup MX server to accept mail on the behalf of your mail server if it is overloaded or unavailable. This has been used in the past to provide reliable mail transport to mail servers that are not always connected.

One downside has traditionally been that backup MX servers were generally much less stringent in their connection level spam filtering/blocking (since the downstream server is generally responsible for that, and they may be a backup server for multiple downstream servers), so it became common for spammers to send directly to lower priority mail servers to take advantage of this and bypass a lot of that active filtering at the eventual destination. Expect a lot of spam to queue up.

In your case, you actually would control the backup MX and the eventual destination (if it indeed is a separate SMTP server), so that's less of a problem. You could just put a pretty harsh timer on the queued mail, and throw it away after 24 or 48 hours. Then again, you could probably do all this almost identically by replacing the SMTP server run on the client device with an IMAP client, and just have delivery end at your server.

Re:Let’s talk about usernames by cyberferret on Monday February 12, 2018 @04:01AM (#16357443)
It seems that the EnvKey app/service does this. You only register with an email address, and if you need to log in to the app, it will email you a one time password to use.

Nowhere near as inconvenient as it sounds - It is the sort of service that you would rarely log in to. Mainly when setting up a new server or adding a third party service to your list of environment variables on your server.

I can see that for a service that you would use several times a day (Twitter etc.) then it would be a major PITA.

Re:Let’s talk about usernames by kbenson on Monday February 12, 2018 @04:11AM (#16357477)
> I can see that for a service that you would use several times a day (Twitter etc.) then it would be a major PITA.

Well, even then, it depends on how long sessions last in combination with your browser cookie policy. How often do you authenticate your HN account?

Re:Let’s talk about usernames by cyberferret on Monday February 12, 2018 @04:16AM (#16357494)
True. Long running sessions would help. There are those that would argue it is a security risk though.

Even EnvKey that I mentioned above has a session cookie of some sort - I can usually use the app for several days after logging in, even if I close the app - but after that I am prompted to instigate the email with my unique login key.

Re:Let’s talk about usernames by kbenson on Monday February 12, 2018 @04:30AM (#16357542)
It would be nice if we eventually got to a point where control of whether a password was even allowed, how long your session cookies lasted, and the ability to list and invalidate all existing sessions was as common and expected as a password reset system.
Re:Let’s talk about usernames by foxhop on Monday February 12, 2018 @10:15AM (#16359082)
Although I disagree that it's two step authentication in the general use of the term, I actually built this type of authentication flow into Remarkbox (https://www.remarkbox.com)

It works really well for most users although it does have some quirks.

Re:Let’s talk about usernames by kbenson on Monday February 12, 2018 @12:57PM (#16360415)
How is it not, even in the general use of the term? Instead of site username and site password plus a separate token to a previously agreed upon authenticated service (whether phone or email account), it's site username and site account email (hopefully hidden), and a token sent to a previously agreed upon authenticated service.

If your account name and associated email is known, it's not really better than a username and password (except that it's delegated to what should be one of your strongest accounts that you protect more diligently), but if the email is not generally known for that account name then it's extra identifying information that must also be known to access the service account.