And you should not prioritize security over the goal of the product.
The conversation is a discussion of relative value and tradeoffs. Does increasing security make the tool as a whole worse? Sometimes - the answer is yes.
I have a nice set of front windows, but that means a risk of someone breaking through them. I accept that risk for the windows - the extra light and visibility is well worth it, and the windows are not the only way in. Compare to short sessions.
E.g. my bank makes me type my password and sends 2fa codes when initiating/approving wire transfers… even when I just logged in a minute ago. If I’m doing 2 wire transfers in a row, it doesn’t care, it still has me fully reauthenticate for every wire transfer.
But I’m fine with that because moving money is something that I’m willing to accept however many roadblocks are thrown at me.
But do I want that happening when I go to post a tweet? Absolutely not.
In other words, let’s adopt the concept of refreshing authentication upon particularly sensitive user actions and have lax requirements in other cases.
We don’t need to think of sessions as “logged in” or “logged out”. It’s possible to design a system where you’re always logged in forever, but you need to reauthenticate based on certain rules or actions given the context of the app and risk/threat.
The "session" that can do things like change 2fa/billing/contact-info (decidedly not-normal things) should last for exactly as long as it takes you to complete that form, and should require your pass/2fa again to touch.
This is currently Google's approach, and I find it much more sane.
Really? I'd change banks over that. If I log into my e-banking website, the main activity I'm going to do is pay bills. I would absolutely not tolerate having to jump through hoops to do it.
In the spirit of adaptive security, someone moving $100k will probably be fine dealing with a couple extra password / 2fa prompts... but I agree it would be annoying to deal with this for day-to-day (consumer) bill pay. A bank could throw fewer roadblocks when paying a $500 invoice vs. a $50,000 invoice
The workflow I described also is part of a dual-approval model where a finance person sets up the transfer and it's approved by a 2nd person (who's presented with a bunch of authentication/password/2fa prompts, almost to a fault).
But again, I'm fine with it because it's large amounts of money in corporate bank accounts. But yea, agreed it would be annoying and should be toned down in a consumer use case.
* logs out after 10 minutes of inactivity - so doing anything that involved switching between accounting app and it is annoying * not allowing more than one tab open at once - that's just stupid in its entirety.
Optimize the application to run the best for all the users first and then adjust the security implementation as necessary. Otherwise, you could DoS yourself by trying to be too secure.
I think it depends what kind of buisness you are running and what a security breach means for you or your users. If it is a hobby forum, well yes, UX matters more. But if you screwed up security for anything with big money related - you probably want to prioritize security first and not after you lost some billions.
Security is just another non functional requirement (mostly) of a product.
To obtain good enough security, defence in depth is still a good principle to follow. It means you are not putting all your eggs in one basket. It often means that each individual control does not have to be perfect or massively over engineered.
I would argue that you are arguing to prioritize security over the goal of the product. Right here and right now - you are literally doing it.
> To obtain good enough security, defence in depth is still a good principle to follow.
I don't disagree! I just think that each "defense" needs to actually be considered on the whole, not as just another bonus to security. Short sessions SUUUUUUUUUCK. They make your product shitty. Users hate them. They don't add a ton of security.
Are there products that should still have them? Sure. Probably lots of products in VERY specific places. Should they be the default everywhere? Sure as hell not.
Okay. I will compare using house to short sessions.
Short sessions is like having house with every doors having a lock and I need to use the keys to get into different room, if I stay in one room for too long, including the shitter. I also need to use the keys to open windows and the oven. And developer going "well, you shat yourself ? That's your fault, should've had keys on you at all times".
That's what short sessions are. Delusional security clowns ignoring usability. It's less than security theatre, it's security circus.
Requiring re-auth to pay some money or delete something important is reasonable stance.
Requiring re-auth few times a day just to browse data in the app is not,
And then, since the top comment was encouraging a wider view, there's even wider view: business needs. Truth is, short sessions make me viscerally hate the product, creates a desire to avoid using it, and becomes a factor in decision to switch to a competing offering when such possibility arises. Or to not switch - one big reason I'm still using my current bank account instead of another one I had to have to get better mortgage rate, is because the bank operating that other account has short session times and associated pseudo-security annoyances.
I think in some (very narrow) cases, short session times and aggressive reauth do add depth and can be an effective part of a security program.
But, all too often, defense in depth is used to mean:
* Vendors In Depth, whereby every security tool under the sun has to be deployed (or at least purchased) to achieve “security”. Or, worse, the Noah’s Ark model where you buy two of everything.
* Uncoordinated and/or seemingly random layering of controls that either don’t add to the overall confidentiality, integrity, or availability of the system being protected. Or, worse, are positively counterproductive and actively reduce the real-world security of the system.
Vendors will always love the phrase 'defense-in-depth' more than they care about assessing whether additional layers of tooling and controls actually provide more defense, because a vague appeal to defense-in-depth is a great way to justify purchasing more security software.
It would be naive to think this doesn't affect the volume of research produced promoting and emphasizing the importance of defense-in-depth either, or how frequently papers on basically any kind of attack end with something that means 'this is why a defense-in-depth approach is needed in XYZ area' even when defense-in-depth is at best incidental to the substance of the paper.
You can trivially tack that on to the end of pretty much any paper about any exploit, which raises the question of how meaningful an observation it really is.
In some sense, the problem is that the goal is zero controls breaking, but of course, that also provides no information on security. Intuitively, it would seem that parameterizing security (in a diagnostically useful way) would also require a number of quite different measures. For instance, it matters who (how sophisticated and motivated) your attackers are. And some of the parameterization would be based on use - every time you force someone to type a fixed password, that password is more exposed. Combining all these quantities would just be Bayes-based.
I think an exercise like this could be instructive. For instance, short sessions mean more authentications and both usability and failure modes. But they also mean that an attacker would need to wait until the next session, or have a smaller chance of hitting a session. Again, these are quantifiable, at least in rough terms. And this kind of analysis does expose some additional questions: eg that short sessions make no sense without guaranteeing the security of the client system (ideally by reinstallation from known-good reference).
At least in my experience, "security people"
But the way risk is managed in the industry (multiplying likelihood and impact) is completely incoherent and voodoo. See the book "How to Measure Anything in Cybersecurity Risk" [1] for a good explanation of why it doesn't work and better ways to do it.
Which is a long way of saying, no, security doesn't use quantitative techniques mostly, but it would be possible if people understood how to measure and manage risk properly.
[1] https://onlinelibrary.wiley.com/doi/book/10.1002/97811198923...
Nonsense, defence in depth is a core security principle. You should not rely on a single control to protect you.