[HN Gopher] 2FA or Not 2FA
       ___________________________________________________________________
        
       2FA or Not 2FA
        
       Author : sam_lowry_
       Score  : 57 points
       Date   : 2025-03-18 12:38 UTC (2 days ago)
        
 (HTM) web link (mikhailian.mova.org)
 (TXT) w3m dump (mikhailian.mova.org)
        
       | Aardwolf wrote:
       | Plus many want your phone number or some random app by them on
       | your phone for their 2FA (instead of e.g. TOTP that you contol),
       | less secure because they now can leak your phone number or do
       | something with an update to the app
       | 
       | BTW what's the sentiment on passkeys?
        
         | account-5 wrote:
         | In my opinion passkeys, whilst solving the password related
         | issues, introduce their own. The risk of losing access to your
         | accounts is greater if you tie everything to one device and
         | that's lost or stolen, and the "solution" to use more than one
         | device is not a solution, or feasible for everyone. There's
         | also the risk of vendor lock-in, which is definitely an aim of
         | the big providers like Apple, Google and Microsoft; which is a
         | bigger risk to those less tech savvy.
        
           | dcow wrote:
           | FWIW no consumer implementation of passkeys is device bound
           | for exactly this reason. The threat of a user losing a device
           | is much bigger and pressing than getting phished. All
           | passkeys are backed up in a sync fabric.
        
             | gclawes wrote:
             | You _can_ use a standalone YubiKey or similar dedicated
             | hardware tokens as a passkey, but they have limited slots
             | for discoverable credentials.
             | 
             | But yeah, anything resembling a phone passkey seems to have
             | a sync fabric behind it.
        
             | jampekka wrote:
             | Most consumer passkey implementations are lock-in.
             | https://arstechnica.com/security/2024/12/passkey-
             | technology-...
        
               | reginald78 wrote:
               | And one of the developers of passkeys threatened to use
               | the specified attestation anti-feature to blackball
               | Keepassxc's implementation when they made something not
               | locked in enough.
               | 
               | https://github.com/keepassxreboot/keepassxc/issues/10407
               | 
               | There have been some discussions to create an export
               | standard since then but I remain skeptical. Why was this
               | not part of the original spec but the ban hammer was?
               | Depending upon how this standard is implemented I can
               | easily see it preventing export to anything but Google,
               | Microsoft and Apple's implementations. And it still
               | leaves the attestation badness in place.
        
       | edent wrote:
       | Password re-use is the bigger issue.
       | 
       | No one can crack your super-strong multilingual password. But if
       | a service accidentally leaks it, then it doesn't matter.
       | 
       | Credential Stuffing is how 23andMe were hacked. People reused
       | password, they were leaked from another service, attackers tried
       | them on a variety of sites until they hit the jackpot.
       | 
       | Unique passwords prevent that attack. Can't remember a thousand
       | different passwords? Use a manager.
       | 
       | Don't want to use a manager? Switch on 2FA. Weak passwords and
       | password reuse ceases to be a problem.
       | 
       | Yes, as the article points out, it slightly reduces ease of
       | login. But that seems like a sensible trade off.
        
         | benoliver999 wrote:
         | I would advise against password re-use regardless of 2FA.
         | 
         | - Lots of flaky 2FA implementations out there where it's easy
         | to get in without it, if you have the password
         | 
         | - If a service doesn't offer 2FA you are now unable to use it
         | for fear of sharing your password (like this website)
         | 
         | I suppose logically if your email is 2FA, then someone can't do
         | 'forgot password', but man that feels super flaky.
        
           | freeone3000 wrote:
           | Gmail "magic link" login (which is the reset password flow,
           | but without needing the password) is the same security
           | profile as Google OAuth, while exposing less user data (name
           | and profile are nonoptional) -- and also equivalent to an
           | Android-managed passkey.
        
             | 9dev wrote:
             | With the price being an absolutely awful user experience.
             | Did the email arrive yet? Or wasn't it sent in the first
             | place? Oh wait, let's check spam. Send again, what do you
             | mean _try again in five minutes_ , I want to sign in _now_!
        
         | autoexec wrote:
         | > Yes, as the article points out, it slightly reduces ease of
         | login. But that seems like a sensible trade off.
         | 
         | In addition to making the login process more complicated 2FA
         | can also introduce privacy concerns. A third party
         | authenticator app can collect all kinds of data for it's own
         | reasons (For example MS's app will request location and camera
         | permissions) and that 3rd party could also track which services
         | you log into, when you access them, and how often you access
         | them.
         | 
         | 2FA can also cause you to be locked out of our accounts, either
         | temporarily or forever.
        
           | vel0city wrote:
           | Having a TOTP app request camera permission isn't nefarious.
           | In fact, I'd 100% expect it. Most of the time people import a
           | TOTP secret from a QR code.
           | 
           | Microsoft Authenticator can be configured by an admin to
           | provide geo-blocking for attempts, so once again not just
           | some arbitrary demand. It's a selling point of the product.
        
             | autoexec wrote:
             | It's not unreasonable for Microsoft Authenticator to
             | request those permissions, but just because they can
             | justify asking for them it doesn't nessesarily mean that
             | you want Microsoft collecting that data, or any data, about
             | what you're doing and when. One of the nice things about
             | hardware tokens is that they keep third parties out of your
             | business entirely.
        
             | tracker1 wrote:
             | Yeah, especially in that MS Authenticator is usually
             | configured for Push mode as opposed to TOTP.
             | 
             | That said, I prefer simple/general TOTP implementations
             | that I can just use with my password manager for my own
             | convenience. But the permissions being asked for are
             | completely reasonable for the actual use of these
             | applications.
        
       | senectus1 wrote:
       | lol
       | 
       | >One of the passwords that I know by heart is a famous classic
       | quote clumsily translated in a mix of French and Dutch. It is
       | long, it can not be brute-forced because of its length and I am
       | pretty sure it is not present in any of the rainbow tables. I
       | never spell it out, let alone write it down, but it is in my
       | muscle memory as I haven't changed it for years.
       | 
       | One keylogger OR phishing attempt and his super complex and
       | secure password is ruined. with no 2FA to protect him.
        
         | akimbostrawman wrote:
         | If a system used for logging in has a keylogger its game over
         | anyways since it could just copy 2FA (excluding hardware token)
         | or cookies.
        
           | dharmab wrote:
           | 2FA doesn't necessarily require entering a code. A lot of 2FA
           | schemes use a push notification where the user confirms the
           | login on a separate device.
           | 
           | Session cookies are a problem, some services provide a way to
           | invalidate all session cookies, but it's often buried deeply
           | in settings and users aren't aware to find the button after a
           | compromise.
        
           | echoangle wrote:
           | How would a keylogger steal cookies? There are hardware
           | keyloggers that only record typed text.
        
             | TeMPOraL wrote:
             | For almost everyone on the planet, being spied on by a
             | hardware keylogger is about as likely to happen as getting
             | eaten by a shark.
             | 
             | Security is important, but not the be-all, end-all. People
             | really should get their priorities and threat models
             | straight, and start counting in denial of service and data
             | loss due to security system being incompatible with human
             | beings as a cost.
        
             | akimbostrawman wrote:
             | Most modern malware are software suits with multiple data
             | extraction feature. Getting a hardware keylogger is
             | extraordinary unlikely and at that point why not also
             | consider video surveillance or evil maid attacks.
        
         | BlackFly wrote:
         | Malware can sniff one password in a similar way that malware
         | can read your password manager's vault. So yeah, one malware
         | and all your unique website passwords might need to be rotated
         | after a recovery anyways.
         | 
         | If the 2fa was a yubikey or some other unique made-for-purpose
         | device, the accounts wouldn't need to be recovered, but TFA is
         | talking about microsoft authenticator so it could very well be
         | on the compromised device.
         | 
         | The most irritating thing about security is people treating it
         | as one dimensional: more or less secure. 2FA changes security
         | in a complex way.
        
         | Y_Y wrote:
         | I assumed everyone salts their memorised passwords. Like you
         | add something that's a function of the thing you're logging
         | into, in a way that's not atrociously obvious. I think (hope)
         | that anything beyond "superpassword1+facebook" is going to be
         | adequate to stop a successful phish cascading to other
         | accounts.
        
           | thaumasiotes wrote:
           | > I assumed everyone salts their memorised passwords. Like
           | you add something that's a function of the thing you're
           | logging into, in a way that's not atrociously obvious.
           | 
           | You're thinking of something other than salting. A salt isn't
           | secret. The point is to make sure that the same password on
           | two services, or on two users of the same service, doesn't
           | have the same hash each time.
        
             | Y_Y wrote:
             | You're dead right. It was an improper use of a specific
             | technical term and that sort of thing is very annoying.
             | Thank you for your service.
        
           | autoexec wrote:
           | That only matters if you reuse a memorized password. If I
           | have my hacker news password memorized but never use it for
           | anything else there's no need to worry about it.
        
       | _dain_ wrote:
       | _> Now, my Github access depends on the second factor, which I
       | have chosen to be Microsoft Authenticator running on my phone. I
       | genuinely do not know what will happen if my phone breaks down,
       | so I downloaded TOTP codes from Github and even tried one to see
       | if it works, and so far it does, but now I have one less TOTP
       | code to use in case something happens. Moreover, since Github is
       | now a special case for my password management routine, I am
       | afraid I may loose those TOTP codes and be totally locked out of
       | my account._
       | 
       | You don't need a phone for this. You can put the secret key into
       | your password manager and it can generate the TOTP code whenever
       | you need it. KeepassXC and 1Password support it.
       | 
       | It stretches the definition of "two-factor" but I don't care;
       | like the author I'm more concerned about phone theft and losing
       | access to everything.
        
         | dandersch wrote:
         | >It stretches the definition of "two-factor"
         | 
         | I always considered the time aspect to be the more important
         | "factor" of TOTP in practice. E.g., if somebody managed to peek
         | over your shoulder or film you while typing in your password.
         | With TOTP, they have under a minute to capitalize on that
         | information. I still have that base covered with my password
         | manager handling the secret key, so I never saw the appeal of
         | tying TOTP to a phone.
        
       | foreigner wrote:
       | This assumes that services are handling your password responsibly
       | and not e.g. storing it in clear text in a public S3 bucket,
       | which in practice happens _all the time_.
        
         | BlackFly wrote:
         | Actually, the author is assuming that you will generate a
         | password like `Password123!` for an obviously fly-by-night
         | company and use a password manager for other websites of medium
         | trust, the author states as much. My reading of his suggestion
         | is that the memorized passwords are used for things like ssh or
         | possibly logins on laptops/pcs. Some people have a good
         | instinct for such things.
         | 
         | This is quite reasonable.
         | 
         | - Useless passwords for useless websites that needlessly
         | require accounts. - Autogenerated passwords for websites of
         | infrequent use that you don't need to trust much. - Memorized
         | passwords for logins of high importance that you need to trust.
         | 
         | Since we only have so much capacity to memorize a password, the
         | idea of reusing a password for the few high importance logins
         | you have can be quite reasonable.
        
           | wobfan wrote:
           | I'm just asking myself, why not use the Password generator +
           | manager _always_ if you have it installed either way. I 've
           | also used some of my digit-only 8-char passwords for some
           | websites where I'm just indifferent about people logging into
           | my account, but usually I just use the PW manager. It take
           | maybe 1-2 more clicks, but more importantly, it saves me from
           | the website saying "ohhh noooo please add a special
           | character", and then "ohhhh sorry but theres no upper case
           | character", blah blah.
           | 
           | By always using the PW manager I have a clear and standard
           | route of registering accounts that is not a lot more work, is
           | way more safe by default, and also can save time if at some
           | point in 2 years you want to log in again, because of some
           | random event. Sure, email reset would be possible, but that
           | takes time again.
           | 
           | Another counter-argument against the article in general, at
           | least in my opinion: while 2FA adds a time consuming step to
           | the login, it happens rarely. I use a lot of services and
           | usually always enable 2FA if it has even a single bit of
           | personal or critical data. But as soon as I'm logged in, the
           | access tokens or refresh tokens are valid for such a long
           | time that I rarely have to do the 2FA challenge again.
        
             | bolobo wrote:
             | I use a laptop, desktop PC, phone, and 2 tablets at home.
             | Another PC and laptop and tablet when I visit my parents.
             | Not all of them are mine, and it is _very_ annoying to have
             | to login to a website on them. You have to go through the
             | unlock flow on your own device (long and complicated
             | password) to access the password, and then copy the site-
             | specific password (usually long and complicated) to the new
             | device.
             | 
             | It is a giant pain. I can understand why people wouldn't
             | want to go through it.
        
               | TeMPOraL wrote:
               | Also, which password manager I should use anyway?
               | 
               | As far as I can tell, there are SaaS ones, broken ones,
               | no longer maintained ones, and the ones that don't work
               | on multiple platforms. There's not one password manager
               | I've heard of that didn't exhibit one or more of the
               | above "features".
               | 
               | "Perfect is the enemy of good", but the effort around
               | making informed choice makes not using password managers
               | seem better.
        
               | _dain_ wrote:
               | KeepassXC is FOSS, runs locally, is actively maintained,
               | and is multi-platform.
               | 
               | https://keepassxc.org/
               | 
               | I've used it for years with no complaints, it's
               | wonderful.
        
             | arkh wrote:
             | > It take maybe 1-2 more clicks
             | 
             | 1-2 clicks here, a couple there, and a click heavy UI.
             | Welcome to the clickodrome, where your patience is tested
             | to its limits.
             | 
             | Why do people click on everything without reading? Because
             | you trained them to.
        
           | 9dev wrote:
           | That makes the questionable assumption you'll always be
           | accounting for all attack vectors someone might come up with
           | on services that you don't think of as dangerous. It only
           | takes a single, contrived way to daisy-chain a way into
           | another system using that innocuous messaging platform you
           | signed up for eight years ago that can now be coerced to send
           | spoofed emails in your name.
           | 
           | A password manager with randomly generated passwords and 2FA
           | is the only sane response to millions of automated attacks.
        
       | Tractor8626 wrote:
       | Author doesn't understand problem space at all.
       | 
       | 1) Weak passwords are not ok even on throw-away accounts. Just
       | because you have no use for it, doesn't mean nobody has. Sending
       | spam, or impersonating you or some other creative use.
       | 
       | 2) Nobody is going to bruteforce your password. We don't use md5
       | anymore. You password will get stolen. By phishing, malware,
       | social engineering, password reuse etc.
        
         | einr wrote:
         | By your own argument, if no one is going to bruteforce your
         | password, what then is the issue with a weak password?
        
           | Jnr wrote:
           | Password lists are full of weak passwords. You don't brute-
           | force the password, you use a password list instead.
        
             | notpushkin wrote:
             | Potato, potato. Does anybody really say "bruteforce" _not_
             | meaning a dictionary attack?
        
               | janalsncm wrote:
               | The author of the article, apparently.
        
           | giantrobot wrote:
           | Credentials stuffing. Attackers can spam a site with logins
           | with common passwords. Too few sites implement good
           | mitigations against this because it's easy to block/lock
           | legitimate users that typoed a password.
        
             | tracker1 wrote:
             | This is why I isolate authentication to a separate
             | application. I also implement max attempts per N minutes
             | for IP and User. Most users once authenticated are good for
             | the work day. Auth going down doesn't (generally) affect
             | the work.
        
         | arkh wrote:
         | > Just because you have no use for it, doesn't mean nobody has.
         | 
         | Lot of websites you'll visit once per decade (maybe) still ask
         | for account. Or things like the software you get to manage your
         | gaming peripherals which nowadays all ask for an account for no
         | reason.
         | 
         | Those accounts getting hacked? I don't care. So they all get a
         | shitty birthday password if they accept it. If they prefer to
         | use some stupid "X uppercase, Y lowercase, Z numbers, some
         | special characters" I'll make a new account next time because
         | I'm not using a real email. Or just stop there.
        
           | wvh wrote:
           | That is your perspective. Not that of the site owner, or the
           | internet at large, victim to any abuse somebody unkind can
           | unleash.
           | 
           | Security is a bit like traffic. If you're alone in the world,
           | you do you. But you are not alone, you have a responsibility
           | to others, be it passers by, fellow travellers or those loved
           | ones depending on you making it back alive.
        
             | ss64 wrote:
             | If a new account has that much power to abuse the system,
             | then your problem is not the 2FA security. They don't need
             | to crack your account, a bad actor could just create a new
             | account for themselves.
        
             | Macha wrote:
             | > That is your perspective. Not that of the site owner, or
             | the internet at large, victim to any abuse somebody unkind
             | can unleash.
             | 
             | Frankly, in a lot of these cases the site owner (e.g.
             | Razer) has already decided to put their interest ahead of
             | mine by requiring accounts to e.g. configure peripherals
             | locally so they have can harvest sign ups for their
             | marketing lists or tell investors they have XXX MAUs. I
             | don't care if my password choice inconveniences them in
             | turn.
        
             | janalsncm wrote:
             | Sounds like you're describing a situation where every user
             | must have a secure account in order avoid a problem. In
             | that case, it's not just the technically-minded folks here
             | who need to have secure passwords. The site itself should
             | enforce password security, not individuals.
        
           | greatgib wrote:
           | Same thing for example for public tv broadcast streams. Often
           | they want you to create an account with an email to watch
           | streams, for totally zero use for you but they hope to be
           | able to target you with ad and co.
           | 
           | Same thing, I create a random account with random creds each
           | time I want to use it. And there will be zero impact for me
           | if it leaks...
        
         | TeMPOraL wrote:
         | > _1) Weak passwords are not ok even on throw-away accounts.
         | Just because you have no use for it, doesn 't mean nobody has.
         | Sending spam, or impersonating you or some other creative use._
         | 
         | Why should that be _my_ problem? It reeks of the same bait-and-
         | switch that banks are doing, with calling failures of their lax
         | KYC /security process "identity theft", calling themselves the
         | victim, and making the actual victim responsible for it.
        
         | croes wrote:
         | Depends on the purpose of the account.
         | 
         | For instance this requires an account
         | 
         | https://news.ycombinator.com/item?id=43245361
        
         | autoexec wrote:
         | > Weak passwords are not ok even on throw-away accounts.
         | 
         | They can be okay for throw-away accounts, it just depends on
         | the circumstance.
         | 
         | > Nobody is going to bruteforce your password.
         | 
         | I can assure that there are still people brute forcing
         | passwords. I see it happening all the time, especially for SSH
         | accounts. While you are correct that phishing and password
         | reuse are problems, they are also not totally solved by using
         | 2FA.
        
         | tigereyeTO wrote:
         | Author also confuses backup codes with TOTPs.
        
       | jve wrote:
       | > Once I added 2FA to my Github account, it became less secure
       | for me. Because security is not only about being protected from
       | intrusion, but also about being able to securely access data at
       | any time and in any circumstances.
       | 
       | Why is he mixing stuff he shouldn't? I didn't find a reasonable
       | explanation why 2FA makes things less secure. Not being able to
       | access your account is more secure and not less secure.
       | 
       | This article has to do with having a backup plan to access your
       | account and not with security. Or maybe "security" as in "peace
       | of mind".
       | 
       | As for Microsoft Authenticator, the data is being backed up and
       | can be restored. Except Microsoft accounts, they require
       | reconfiguration and... is useless to have it backed up and gives
       | false sense of "security" (peace of mind).
        
         | onli wrote:
         | The 2FA adds the risk factor of being locked out. Looking at it
         | that way I'd say it is valid to call that an insecurity.
         | 
         | You are right though that this isbabouy having a backup plan.
         | For example, the TOTP app I use supports exporting the logins
         | into a file, which van be saved elsewhere.
        
       | TimJRobinson wrote:
       | I thought he was going to mention the stupidity of sites like
       | Twitter that when you add SMS as a 2FA option you can now use
       | that to bypass the password and so are vulnerable to sim
       | hijacking, which given how incompetent phone company employees
       | are makes your security weaker.
       | 
       | Always use an authenticator app or physical key, most sites that
       | do SMS 2FA will then allow hackers to use it to bypass knowing
       | your password.
        
       | fmajid wrote:
       | This is compounded by the fact most 2FA implementations are
       | security theater not phishing-resistant, the sole exception being
       | FIDO2/Webauthn (e.g. Yubikeys), and Passkeys, but passkeys are
       | really single-factor authentication. Both FIDO and passkeys have
       | serious usability challenges, though.
       | 
       | What's worse, the most common scheme, SMS-based authentication
       | can lead to denial of service, e.g. you are roaming and do not
       | have access to texts, or have your account SIM-jacked as this
       | seems to be very popular way to steal people's cryptocurrencies.
        
         | _dain_ wrote:
         | _> you are roaming and do not have access to texts_
         | 
         | I truly despise this. It effectively disenfranchises people for
         | living outside of areas with good mobile coverage. Banks or
         | utility payments or parking meters(!) or whatever should not be
         | gated behind cellphone reception. Nevermind people who can't
         | use a phone at all ...
        
         | bradley13 wrote:
         | This. In theory, passkeys should be great. Sort of an SSH-for-
         | the-masses.
         | 
         | Unfortunately, the industry has mangled the implementation,
         | making them basically useless.
        
           | Spivak wrote:
           | I mean they're not totally useless, in the current
           | implementation they just can't replace passwords. I have a
           | bunch of passkeys in my Bitwarden and they function as a "log
           | in bypassing the 2FA screen" button. I get to skip the "we
           | sent a code to your phone/email/butthole" flow.
        
           | dcow wrote:
           | Funny how that always happens. Passkeys were supposed to be
           | great and what you ended up with is platform players abusing
           | their position to push lock in to their own passkey solutions
           | over fair access to arbitrary 3rd party providers.
           | 
           | When they do work smoothly they aren't useless though.
        
             | gclawes wrote:
             | Android has been surprisingly awesome with cross-platform
             | passkeys (i.e. google lets you use BitWarden/1Password/etc
             | pretty transparently).
             | 
             | Anything Apple related is another story...
             | 
             | There's some good movement in the linux desktop space I'm
             | excited about: https://github.com/linux-credentials/linux-
             | webauthn-platform...
        
               | ziml77 wrote:
               | I'm using iOS and Passkeys from 1Password work there too.
               | No problems with any browser on macOS either.
               | 
               | On iOS I can even see a list of password & passkey
               | providers, so I can ensure that the Apple one is toggled
               | off (as well as MS Authenticator which can also provide
               | those it seems)
        
         | autoexec wrote:
         | Passkeys can't always protect an account against phishing
         | either (https://cyberpress.org/passkeys-with-aitm-phishing-
         | attacks/) they just make it more difficult for scammers.
        
           | vel0city wrote:
           | The article agrees passkeys can't be phished. They
           | acknowledge they just push users to go through an unnecessary
           | account recovery and then phish those other credentials.
        
       | echoangle wrote:
       | > Because security is not only about being protected from
       | intrusion, but also about being able to securely access data at
       | any time and in any circumstances.
       | 
       | Is it though? Wouldn't that be some kind of safety thing? I think
       | having no way to access the data would be ultimate security,
       | that's why there's always a tradeoff between security and
       | usability.
        
         | kalaksi wrote:
         | It is. Confidentiality, integrity, and availability.
         | 
         | Edit: see e.g.
         | https://en.wikipedia.org/wiki/Information_security
        
           | echoangle wrote:
           | Interesting. Makes sense if you think about it though,
           | otherwise protecting from ransomware wouldn't fall under
           | security.
        
       | karel-3d wrote:
       | > Lately some services started requiring 2FA. One of them is
       | Github. Once I added 2FA to my Github account, it became less
       | secure for me. Because security is not only about being protected
       | from intrusion, but also about being able to securely access data
       | at any time and in any circumstances.
       | 
       | >
       | 
       | > Now, my Github access depends on the second factor, which I
       | have chosen to be Microsoft Authenticator running on my phone. I
       | genuinely do not know what will happen if my phone breaks down,
       | so I downloaded TOTP codes from Github and even tried one to see
       | if it works, and so far it does, but now I have one less TOTP
       | code to use in case something happens. Moreover, since Github is
       | now a special case for my password management routine, I am
       | afraid I may loose those TOTP codes and be totally locked out of
       | my account
       | 
       | What? That's not what security means. Sure, you traded
       | convenience for security.
       | 
       | Why is this upvoted at all?
        
         | TeMPOraL wrote:
         | Because the current obsession with security is making systems
         | _unsafe_ and unreliable for users, all while trying to get
         | those users more dependent on said systems in their daily
         | lives.
         | 
         | What is the biggest risk factor with Google account _for me_?
         | Is it an attacker guessing my password, impersonating me,
         | stealing my photos? No. It 's _me breaking my phone_ , and no
         | longer being able to pass 2FA[0][1]. The second biggest risk?
         | Me typing a wrong comment or YouTube, or doing any other minor
         | transgression against ToS of some Google service, and losing
         | access to everything in one go, with no recourse[2].
         | 
         | Note that literally _nothing_ in meatspace ever requires as
         | careful management over years to decades, as 2FA does. There is
         | _always_ a recovery procedure. You may need to stop by the
         | court or a notary, but no matter the fuckup, you can _always_
         | recover access - to everything _except_ modern Internet
         | services.
         | 
         | --
         | 
         | [0] - Yes, I have the security codes I generated _15+ years
         | ago_ ; I have them on paper, _somewhere_. Like most people, I
         | suck at keeping small paper documents accessible and available
         | over years.
         | 
         | [1] - Also yes, I did break my phone, and I survived this
         | without data loss only _because_ I had a complex setup around
         | Pebble and Tasker, that allowed me to operate the phone with
         | non-functioning screen remotely to the point I could mirror the
         | display to the computer and continue from then. Most people in
         | most situations can 't do that.
         | 
         | [2] - Other than complaining on HN and hoping some Googler will
         | advocate internally for me - which, as far as I know, they're
         | _explicitly not allowed to do_ , and it's a career-risking
         | move.
        
         | ozim wrote:
         | Well OP is kind of correct because in security importance is
         | given to CIA Triad: Confidentiality, Integrity, Availability.
         | 
         | Availability is important part of any system, if you lock out
         | people, system will be secure but it stops serving its purpose.
         | 
         | Where OP fails is he seems not to understand password leaks and
         | how much password+e-mail+phone number+other info lists are out
         | there.
         | 
         | There is credential stuffing and password spraying running
         | around from leaked passwords, so you can try dozen users on FB
         | and just hit jackpot with one or couple, you don't have to
         | brute force each one of them as there are countermeasures for
         | that.
         | 
         | Even if you have weak password with a trick it is trivial to
         | find out your trick when your password leaked from 5-10
         | services where you used the same e-mail.
        
         | _dain_ wrote:
         | _> What? That's not what security means. Sure, you traded
         | convenience for security._
         | 
         | Your Google/Microsoft/FB/etc accounts are used to sign in or
         | access information about other extremely important things in
         | your life, like your bank, payroll, travel companies, health
         | records, taxes, insurance, everything really. If you get locked
         | out of those SSO "identity providers", you get transitively
         | locked out of everything else. Which means you could end up
         | without access to money, unable to pay rent, unable to travel,
         | contact friends for help, etc.
         | 
         | In other words, your _physical_ security would be threatened,
         | as in you might literally become homeless and die of exposure.
        
         | megous wrote:
         | Handling recovery codes is part of security. There are many
         | gotchas even with a typical recommendation to print them and
         | store them physically.
        
       | latexr wrote:
       | > Because security is not only about being protected from
       | intrusion, but also about being able to securely access data at
       | any time and in any circumstances.
       | 
       | This felt like the author bending over backwards to justify their
       | choice. They find 2FA _less convenient_ and conflate it with
       | being less secure. It's not the same thing.
       | 
       | It's OK to say "not all my accounts are equally important and I
       | need to access some of them in situations where 2FA and complex
       | passwords aren't worth it". It's not OK to sell the idea that 2FA
       | does not generally offer security.
       | 
       | This reminded me of the "SEO expert" a few years back which was
       | trying to convince everyone, with wrong information, to not use
       | HTTPS (which, I realise only after writing this, the author's
       | website also doesn't use).
        
         | guappa wrote:
         | If we define data that can't be accessed by anyone as secure,
         | throwing disks in a furnace is how you achieve security.
         | 
         | I don't think this definition is very helpful though. So I
         | prefer the one where the entities that need to have access
         | still can access.
        
           | latexr wrote:
           | > I don't think this definition is very helpful though.
           | 
           | Because it's a straw man, and straw men aren't helpful for
           | discussion. No one is suggesting making data wholly
           | inaccessible.
           | 
           | Data that you cannot access "at any time and in any
           | circumstances" (author's words) can still be secure. A fairer
           | analogy would have been storing disks in a locked safe in
           | your home. It's not as _convenient_ to access it, but it is
           | secure. Should you do that for all your data? No, but neither
           | have I advocated for that. I very clearly stated that I think
           | it's OK to have different levels of protection for different
           | types of data.
        
         | mouse_ wrote:
         | If I can't access money that I secured, I didn't secure it
         | properly.
        
           | latexr wrote:
           | Which has nothing to do with my argument. See the reply to
           | the other user, which expands.
           | 
           | https://news.ycombinator.com/item?id=43421721
        
         | Spooky23 wrote:
         | Security is about risk management. If the value of what you're
         | protecting is low, or the consequences of not accessing
         | something is high, than the MFA control may not add value.
         | 
         | The problem is in general people are really bad at assessing
         | risk. You tend to see extremes.
        
       | seethishat wrote:
       | For the author... you may export/backup the TOTP secret from the
       | TOTP app (MS Authenticator in this example) and save that
       | someplace secure. You may use any TOTP app with that secret to
       | generate codes to log in.
       | 
       | I typically do this as a safe guard in case the device that is
       | running the TOTP app breaks or fails in some way.
       | 
       | I'm old enough and my passwords are random enough that I don't
       | worry about them being guessed. However, I do worry about web
       | security in general and the fact that companies are basically
       | black boxes with loads of young inexperienced developers. They
       | may be storing our passwords as plaintext strings and have them
       | in a public S3 bucket. But, if you want one of my YubiKeys, you'd
       | have to find me and physically steal it from me and hope I don't
       | notice.
       | 
       | So there are some useful things with regard to 2FA.
        
       | bradley13 wrote:
       | He's not wrong. I use a password manager that stores passwords
       | only locally, not in the cloud. I can generate arbitrarily
       | complex passwords and never have to worry about how hard they are
       | to type or remember.
       | 
       | 2FA makes my workflow significantly more difficult. As a result,
       | for non-critical sites, I have started allowing the browser to
       | store my passwords, thus relying on the 2FA-authenticator for
       | security. This result is likely _less_ secure overall, since the
       | browser 's password storage could realistically be compromised.
       | 
       | That said, I do have to acknowledge the point in another comment
       | that phishing may be the bigger threat. Log into a fake site with
       | a password but no 2FA, and you are toast.
        
         | postalrat wrote:
         | What is the difference between a secure password synced with a
         | password manager and a cookie that doesn't expire synced with a
         | browser or extension? I don't see much difference so why bother
         | even having a password?
        
           | megous wrote:
           | The difference is that cookies are not even encrypted and you
           | can get all of them by `sqlite3 .dump` on `cookies.sqlite`
           | file, while password manager likely offers more protection,
           | since that's the point of it. :D
           | 
           | Also SID are ephemeral and expire, while passwords are not
           | supposed to be.
           | 
           | Session IDs may be tied by the backend to a client IP
           | address, or some such, but likely that's not done very often,
           | and may not be that useful.
        
           | 9dev wrote:
           | What if your cookie is compromised and you want to ensure the
           | attacker cannot log into your account anymore?
        
       | apples_oranges wrote:
       | Off topic: This is a website without HTTPS
        
         | red369 wrote:
         | Interested in your thoughts (and anyone else's) - is this OK
         | for a website just serving content for consumption only,
         | without logins, forms, etc?
         | 
         | Would people rather that no-one can MITM and see which articles
         | they read on the site? Or is it about conditioning people do
         | demand HTTPS so that it's there when they need it?
         | 
         | BTW: I feel I might be missing the real issues or
         | unintentionally straw-manned them.
        
       | jesprenj wrote:
       | I think that 2FA is widely required not to protect users of
       | online services but to protect the services themselves and/or
       | their administrators from a burden of annoying work. Users that
       | create accounts with weak passwords because they don't care about
       | their accounts present a problem for the company behind the
       | service, because their accounts will eventually become used as
       | puppets of spammers etc. By requiring a very complicated login
       | procedure, spammers won't try to hack those unused accounts and
       | exploit them, resulting in less email being sent to abuse@.
       | 
       | And I think that's the most probable reason 2FA is _required_
       | today on most online services.
        
         | dcow wrote:
         | Nailed it. 2FA is about businesses doing risk management.
        
       | donohoe wrote:
       | It's hard to take this advice from someone who's site doesn't
       | even support HTTPS.
        
         | jampekka wrote:
         | HTTPS isn't very critical for a static site. Only the Sith, and
         | security zealots, deal in absolutes.
        
           | SahAssar wrote:
           | HTTPS protects the integrity and confidentiality (to a
           | certain point, see ESNI) of what you visit. Those are
           | important even if you don't submit any additional data to the
           | server.
           | 
           | Disclaimer: I'm not a sith or a security zealot.
        
       | xpil wrote:
       | I've noticed another issue: 2FA software fragmentation. Over the
       | past 30 years, I've accumulated more than 600 online accounts
       | (many of them dormant), with about 25-30% offering 2FA. As a
       | result, I now have three separate apps for generating codes. I
       | could consolidate them into one, but, frankly, I'm too lazy. So
       | instead, I often find myself scrambling between different 2FA
       | apps and devices just to locate the right account.
        
       | perlgeek wrote:
       | There are really two aspects two 2FA:
       | 
       | 1: you as a user might want to protect your account
       | 
       | 2: A website provider wants to secure their own business
       | 
       | If it's about 1, then the argument "I don't reuse passwords" hold
       | some water. It doesn't protect you from somebody getting access
       | to a plaintext copy of your requests (for example on the load
       | balancer that terminates the TLS), but that's only a small part
       | of potential attack surfaces.
       | 
       | But from the perspective the website owner, they might suspect
       | that many of their users reuse passwords between sites. 2FA is a
       | great defense against that, and also against brute force
       | attempts.
        
       | PinguTS wrote:
       | There is another issue with all those growing 2FA/MFA protected
       | accounts: managing your unexpected death.
       | 
       | How do you teach your beloved ones to access your accounts. And
       | they need to remember what to do. For some accounts may it
       | doesn't matter. For others that means to be able to end the
       | subscription. Because not all subscriptions are associated with a
       | credit card, which just expires.
       | 
       | But if you allow to contact a company by a third party to cancel
       | and or change things then this becomes the go to for social
       | engineering.
        
         | croes wrote:
         | Isn't the device to access the password manager mostly the same
         | device used for 2FA?
         | 
         | I hope most people use a password manager.
        
           | dewey wrote:
           | Most people are just storing them in the browser but that's
           | not stopping people from coming up with easy to guess
           | passwords.
        
         | pests wrote:
         | Services like FB have the concept of a legacy contact, an
         | account that can manage your page after death. Also the concept
         | of your profile switching to a memorial page, with your legacy
         | contact doing moderation of posts.
         | 
         | I did the math years ago and even back then, thousands of users
         | would pass daily and now with most of the world population on
         | FB it probably comes in handy.
        
         | bobbiechen wrote:
         | In the US, you have a legal right to access accounts of the
         | deceased through RUFADAA. It's not as simple as logging in
         | (usually requires a court order) but it's possible. I wrote
         | about it and related issues here:
         | https://digitalseams.com/blog/what-happens-to-your-online-ac...
         | 
         | (and recently discussed on HN:
         | https://news.ycombinator.com/item?id=42991112)
        
       | 1970-01-01 wrote:
       | OLD NEWS:
       | 
       | Something you have.
       | 
       | Something you know.
       | 
       | Something you are.
       | 
       | Pick two.
       | 
       | NEW: Dismiss anything that is a hybridization. That is 1.5FA
       | 
       | It's 2025. 0-days are rampant. If you're not at 2.0FA, you're at
       | a high risk of getting popped.
        
       | alwayslikethis wrote:
       | I wish sites offered a way to opt out of 2FA if your password has
       | enough entropy (128 bit random string). These are not getting
       | cracked anytime soon. I store my TOTPs in keepassxc with the
       | other passwords anyway. The keepassxc database is the "something
       | you have" and its password is "something you know", and the
       | random string is a testament to that. Also stop forcing SMS 2FA
       | please. I don't want to need to have a phone.
        
       | evolve2k wrote:
       | > I never spell it out, let alone write it down, but it is in my
       | muscle memory as I haven't changed it for years. There is no way
       | someone on the internet can break into my ssh account or gmail
       | account protected by such a password.
       | 
       | Oh dear. The issue isn't the brute force, it's that the online
       | services leak and get cracked. And in an instant a single script
       | takes the newly discovered username password combo and starts
       | hammering it into the top 10000 websites, all within moments of
       | the leak data becoming available.
       | 
       | Your super secret favourite phrase is worth crap once leaked
       | alongside your email address.
       | 
       | Further don't choose Microsoft for your Auth app, Go with an open
       | source option, maybe one that encrypts and syncs so you have
       | multiple devices just in case.
        
         | ajd555 wrote:
         | I could not agree more with this comment. OP entirely misses
         | the point of 2FA. I sleep so much better at night knowing that
         | I have different passwords for every account, and 2FA where
         | possible. One should not write about 2FA when one uses the same
         | "uncrackable" password everywhere...
        
           | autoexec wrote:
           | Maybe I missed it (it's early and I haven't even had coffee
           | yet) but where did the author say they resued the same
           | password over multiple sites?
        
         | femtozer wrote:
         | Not sure I understand -- passwords are generally hashed in
         | databases. Even if leaked, an attacker would still need to
         | brute-force the hash to retrieve the actual password, wouldn't
         | they?
        
           | evolve2k wrote:
           | You'd think so. But over and over plain text leaks of
           | passwords is the practical reality of the modern internet. A
           | disgruntled staff member, poor tech practices or someone
           | working out a way to get in and get access.
           | 
           | The https://haveibeenpwned.com/ project regularly shares new
           | breached datasets. Reusing passwords across websites without
           | MFA is just not not not recommended in 2025.
        
           | parliament32 wrote:
           | "Generally", sure. How do you guarantee every service you've
           | ever signed up for uses proper salting and hashing though?
           | All it takes is one for your entire security model to go down
           | the drain.
        
         | autoexec wrote:
         | > Oh dear. The issue isn't the brute force, it's that the
         | online services leak and get cracked. And in an instant a
         | single script takes the newly discovered username password
         | combo and starts hammering it into the top 10000 websites, all
         | within moments of the leak data becoming available.
         | 
         | This is only ever a problem if your password is reused. Don't
         | reuse passwords and if some website is hacked and they were
         | storing your password in plaintext you just have to reset your
         | password (the same way everyone else does, 2FA or not)
        
           | c22 wrote:
           | Also don't re-use email addresses.
        
           | robertlagrant wrote:
           | > This is only ever a problem if your password is reused
           | 
           | That is the context of the reply, although I think they
           | misread the article.
        
         | Justsignedup wrote:
         | the weird thing is:
         | 
         | - if you use 1password (an example), then you're generating a
         | bunch of random and unique passwords for every site - questions
         | to verify you as a 2fa tends to be less secure since you tend
         | to make simple answers for those. And they're not convenient to
         | enter into 2fa apps. - 2fa apps are typically great ways to
         | guarantee one bit of randomness into the process
        
           | tracker1 wrote:
           | I use the same app for most 2fa and the passwords themselves
           | (Bitwarden). It makes the 2FA slightly weaker being in the
           | same app, but infinitely more useful. It does bug me that
           | they (Bitwarden) as a service want me to use 2FA for first
           | logins, which makes it harder to access. My master passphrase
           | is long, unique and only on their app/site.
        
         | tracker1 wrote:
         | While true... this is less of an issue if the breached database
         | includes strongly encrypted passwords with individual salts. At
         | least half of them are going to be part of existing breaches,
         | but you aren't going to bother with the rest as it can/will
         | take an exponential amount of time if they are treated
         | properly, leaving top's password safe(ish).
        
           | tomjakubowski wrote:
           | Passwords can leak in many ways other than database breaches.
           | Malicious front-end code and accidental logging that goes to
           | a public place like an S3 bucket are two examples.
        
           | 9dev wrote:
           | So I am supposed to trust that the random forum I have to
           | sign up for to view the solution of a question securely
           | hashes the password I send them?
           | 
           | That's pretty much like handing you car keys to a random
           | person on the street and be confident they will take it to
           | the bank and put it in a locker.
        
       | megous wrote:
       | > I never spell it out, let alone write it down, but it is in my
       | muscle memory as I haven't changed it for years. There is no way
       | someone on the internet can break into my ssh account or gmail
       | account protected by such a password.
       | 
       | SSH... no, they can't.
       | 
       | Gmail... yes, you're basically giving your password in plaintext
       | to someone on the internet.
       | 
       | I dislike 2FA for similar reasons, but I would rigorously
       | segregate remembered passwords between those shared with some
       | service, and those used locally.
        
       | gibibit wrote:
       | I agree with the article. Maybe businesses are trying to protect
       | themselves, but as a user, mandatory 2FA reduces the level of
       | security I can achieve for myself.
       | 
       | Because security is not just confidentiality, it's also
       | availability: the "Security CIA Triad" is Confidentiality,
       | Integrity, and Availability.
       | 
       | If I can lose access (availability) to my online account by
       | losing some physical item (e.g. lost cell phone), or if some
       | third party can prevent me from accessing my 2FA (e.g. banned
       | from my email provider by DMCA takedown request), then I have my
       | availability, and hence overall security is at risk.
       | 
       | Additionally, requiring a phone number for online services means
       | that the confidentiality of my identity is reduced. It becomes
       | impossible to be anonymous. For instance, you can't use Signal
       | messenger without a phone number, so there's a chance your
       | identity can be leaked.
        
       | gpi wrote:
       | Can't seem to access the site over https.
        
       ___________________________________________________________________
       (page generated 2025-03-20 23:01 UTC)