# Privacy vs. Security: Complexity at the Crossroads

URL: https://securitytable.ai/privacy-vs-security-complexity-at-the-crossroads/
Published: 2024-06-18
Duration: 2148 seconds
Season: 2
Episode: 18
Hosts: Chris Romeo, Izar Tarandach, Matt Coles

## Show notes and chapters

In this episode of the Security Table, Chris, Izar, and Matt delve into the evolving landscape of cybersecurity. The episode has a humorous start involving t-shirts and Frogger as a metaphor for the cybersecurity journey, the conversation shifts to the significant topic of cybersecurity being at a crossroads as suggested by a CSO Online article.  
  
They explore the concept of moving from a product-centric to an architectural-centric approach in cybersecurity, discussing the design and integration of inherent capabilities rather than relying on add-on products. The hosts look into the complexities of security and privacy, analyzing their intersections, the challenges of privacy threat modeling, and the importance of understanding the broader ecosystem in which data interacts. The episode concludes with a lively discussion on the evolving nature of security and privacy regulations, the impact of complexity, and the need for continuous threat modeling.  
  
Article mentioned in this episode:  
[Cybersecurity at a crossroads: Time to shift to an architectural approach](https://www.csoonline.com/article/2126804/cybersecurity-is-at-a-crossroads-its-time-to-shift-to-an-architectural-approach.html)  

Follow AI Security Table:

➜ Home: [https://securitytable.ai/](https://securitytable.ai/)

➜ X: [https://x.com/AISecurityTable](https://x.com/AISecurityTable)

➜ LinkedIn: [https://www.linkedin.com/company/ai-security-table/](https://www.linkedin.com/company/ai-security-table/)

➜ YouTube: [https://www.youtube.com/@AISecurityTable](https://www.youtube.com/@AISecurityTable)

## Transcript

**Matt Coles (0:09)**: I need to frame.

**Chris Romeo (0:10)**: Please get into frame.

**Matt Coles (0:12)**: Please get in the frame.

**Chris Romeo (0:16)**: Uh, it's okay if you disagree with me. You're right. Okay.

**Matt Coles (0:22)**: Oh, you have a spelling error on your shirt, right?

**Izar Tarandach (0:24)**: A what?

**Matt Coles (0:25)**: You have a spelling error on your shirt, you know.

**Izar Tarandach (0:27)**: Do I?

**Matt Coles (0:29)**: You do.

**Izar Tarandach (0:29)**: Well, it's not mine. This one's commercial.

**Matt Coles (0:32)**: Well, then you bought a bad one because it's okay if you disagree with me.

**Izar Tarandach (0:38)**: What's the spelling mistake? Can't force you to— that's so cool. That's cool.

**Chris Romeo (0:52)**: You literally have a shirt with a bug or a fly.

**Izar Tarandach (0:59)**: I was actually just asking if you guys were paying attention.

**Chris Romeo (1:01)**: Apparently we were.

**Matt Coles (1:03)**: So Monday morning.

**Izar Tarandach (1:05)**: Nothing gets by Matt.

**Chris Romeo (1:07)**: That's right. So hey, I just realized we had started our podcast about a minute ago.

**Izar Tarandach (1:13)**: Oh, you guys are there.

**Chris Romeo (1:14)**: Yeah, so people are here. So welcome to the security podcast where everybody wears a black shirt. Also known as the Security Table. So my name is Chris Romeo, joined by Izar Tarandash, Matt Coles, and we're here to talk about cybersecurity at a crossroads. Sounds like such an important headline. We're at a crossroads, everyone. After almost 30 years in this profession, I finally reached a crossroads.

**Izar Tarandach (1:42)**: It's the other way around. I see us jumping from crossroad to crossroad. It feels like a traveling salesman problem or something.

**Matt Coles (1:48)**: We keep kind of like frog.

**Izar Tarandach (1:51)**: It's like Frogger. Yeah.

**Chris Romeo (1:56)**: Take that step forward, step back, step forward, step back, side, sideways.

**Izar Tarandach (2:02)**: And then it gets hit by a truck.

**Chris Romeo (2:06)**: You get by the truck. This is a cybersecurity is a metaphor. No, no, no. Frogger is a metaphor for cybersecurity. Wow. Well, there's a t-shirt. Put that on.

**Izar Tarandach (2:17)**: That's a t-shirt. That set the—

**Chris Romeo (2:18)**: Yeah. All right. Well, so this article is from CSO Online back in, I think we said May of 2024, and the article is entitled Cybersecurity at Crossroads: Time to Shift to an Architectural Approach.

**Matt Coles (2:36)**: Okay. So that was an architectural approach.

**Chris Romeo (2:40)**: Yeah. Well, what is an architectural approach? Let's explore that idea first from our perspective. So Matt, what's an architectural approach?

**Matt Coles (2:48)**: Uh, well, I'll say the context I know is you build capabilities in by design at your architectural level, and that's what you rely on. Uh, and hopefully that's what they mean here. And I'm guessing that that is right. So the fundamental statement they make in this article is product-centric versus architectural-centric. And I think I just came to realize that that means building capabilities into your architecture rather than relying on add-on products. And so that's, that's interesting, right? We've always touted build-in, not bolt-on, but why are we talking here in May 2024 now talking about build-in versus bolt-on?

**Izar Tarandach (3:38)**: I know, I know, I know.

**Matt Coles (3:41)**: Is it not?

**Izar Tarandach (3:43)**: Because nobody's doing it because somebody's doing it.

**Chris Romeo (3:47)**: Yeah. But I think this is, I think they're taking attention. I think they're, I think, but I think they're talking about something else. I think, I think we're, I think you're kind of reading the things we love into, into, because look at the continuation of the, the subheadline. The need for greater scale intelligence and automation is driving massive change in security operations in the SIEM market. So I'm reading this as more of, they're talking about an architectural approach to the solution versus a point product challenge that we've had for a number of years about how you're going to have the architecture come together. Now, I guess we'll have to keep reading to see which path is the correct interpretation here.

**Matt Coles (4:36)**: Is there?

**Izar Tarandach (4:37)**: Yeah, so that's the thing, you know. So when we talk security, we talk security, usually product security, right? But there is this thing of the security of the whole environment where you're operating. You are, sorry for the lack of a better word, cybersecurity stuff. And, uh, yeah, I, I agree with, uh, with Chris that this is what they seem to be going towards. But at the same time, seriously, to, to come now and say, oh, we need to look at the whole as an architectural thing. I mean, we've been talking about Lego pieces and Jenga pieces for what, 30 years? And to just come now and realize, oh, we have to structure this thing into layers and we have to. Instrumentalize each one of them and we have to have, oh my God, if I hear that word again, that expression again, a single pane of glass that shows everything, right? Last I counted, I must have at least 4 single pane of glasses or something.

**Matt Coles (5:40)**: But, um, it's not that we have a, we have a jar, we have a jar for that, don't we?

**Chris Romeo (5:45)**: Yeah. Oh yeah, we do.

**Izar Tarandach (5:46)**: We do. Right. But I'm getting ahead of the, I'm getting ahead here. Let's, let's go layer by layer.

**Chris Romeo (5:54)**: Yeah, let's unpack a few things that they said here because there's a couple of things in this first section that it seems like it's part of the playbook of a successful cybersecurity article. You have to, you have to, you have to mention the attack surface and the evolving threat landscape, which is second paragraph. And then you also have to state that it's your firm belief that every one of these issues will grow substantially more complex over the next 2 to 3 years. You could say that since the beginning of time.

**Izar Tarandach (6:26)**: And you can say that about anything.

**Chris Romeo (6:29)**: Yeah. In the next 2 to 3 years, cars are going to become more complex.

**Matt Coles (6:33)**: Wow. Technology does not reverse complexity. It's just—

**Izar Tarandach (6:38)**: oh, wait, wait, wait. That's good. That's good. Well, that's almost Arthur C. Clarke level.

**Chris Romeo (6:46)**: Okay.

**Izar Tarandach (6:46)**: No, seriously. That's good.

**Chris Romeo (6:48)**: Matthew C. Clarke over here.

**Matt Coles (6:50)**: I mean, think of it this— I mean, it is real, right? We don't— technology becomes more complex because humans want to do more and we want to be more productive. So we build more complexity into the technology. The technology never reverses complexity.

**Chris Romeo (7:06)**: So I'm trying to think of an example. Has there ever been an example where something became less complex?

**Izar Tarandach (7:17)**: Oh, just because I'm going to read. Heat pumps versus air conditioning.

**Matt Coles (7:21)**: Air conditioning?

**Izar Tarandach (7:22)**: Yeah. Heat pumps are simpler. You do a hole in the ground, you stick a rod into it, heat does the physics things that heat does, and magic.

**Matt Coles (7:33)**: That's a good example, actually. Now, having said that, you have a lot more software. To control that. The heat pump itself may be simpler, but the control systems are more complex because you need to get a little balance.

**Chris Romeo (7:45)**: Ah, Matt for the win.

**Izar Tarandach (7:47)**: Yep. Nope. Totally true.

**Chris Romeo (7:49)**: Totally right. Complexity wins.

**Izar Tarandach (7:51)**: I'll give that one to you.

**Chris Romeo (7:52)**: Yep. And I think the, the obvious parallel or factor that we can draw from this is complexity has always been the enemy of security. And you could say complexity is the enemy of privacy as well. Because the more complex something gets, the more challenging and least— less likely you are to be able to secure it and protect the privacy of the data contained within it. So complexity is actually an anti-pattern.

**Izar Tarandach (8:24)**: But that, that, that point there, actually, I, I think it's pretty interesting because it's not that complexity hits privacy the same way that it hits security. I think that in the security bit, you're totally right. Like, There's that quote that I love going back to, that things should be as simple as they can, but no simpler than that. But for privacy, I think that the hit on privacy is only to the extent that we need a certain amount of security to protect privacy. But privacy itself as an idea, as a thing, I'm not quite sure the complexity hits it the same way. 'Cause it's not like because something is more complex, you're going to be needing more data or you might be, or, or, or getting more data or distributing that data in different ways or like that, that's part of the complexity.

**Matt Coles (9:19)**: That's the point actually. That's the point you just made, I, I think is important. And I, as I'm learning about privacy and certainly we're not the experts here, we, we definitely want to tap into other, some other folks, friends of ours to, uh, about that, but security, security can be a point solution, right? You can attach security to, uh, you know, take a design and, and fix a security issue, right? Secure communications, right? TLS. Good example, right? We can apply a point solution to a problem, making more secure. When it comes to privacy, though, privacy is all about— first off, privacy is heavily subjective. And it is also related to data coming in and data coming out. And so if we look at it, just to sidetrack here, right, obviously, because this is not related to the CSO article, CSO article. When we look at a data flow diagram, we can look at a data flow diagram and because we can look at it as an element, by element or by interaction, we can solve security issues by elements or interaction. But when you look at privacy, for the most part, you're looking at the collection of elements and the collection of interactions. And so you really need that, you know, you really need the flow and not just the elements or the interaction. And so it is much more, it's more complex by design. It's more complex in its base characteristics because it's all about where and how and why you collect data and where and how and why you store that data and manipulate it and making that connection.

**Chris Romeo (11:02)**: You almost said it's complex by design.

**Matt Coles (11:04)**: I did say complex by design.

**Chris Romeo (11:06)**: When you backed out of it, but I caught it. I was like, hey, that's a nice—

**Matt Coles (11:08)**: That's about where I was going.

**Izar Tarandach (11:11)**: But actually that puts us into an interesting challenge still going with the offshoot here, that when we are threat modeling and we are looking at a diagram, we usually know when to stop the level zero. We know what sort of granularity we need for the blocks that we are going to put in there. But I think with the little experience that I have had so far and looking to learn more, that if you are threat modeling for privacy, You need to even step back from the level zero because now you have a whole context happening around those blocks and, and their connections that is not immediately apparent from the diagram. That, that, that there is a story on top of the system that is much more influential on the privacy thing than on the security thing.

**Matt Coles (12:03)**: So, so forget the, forget the CS Oracle. Let, let's, let's just, let's just take this as a thread. This, uh, This is, this is perfect. So, 'cause now you're, now you're hitting an actual, this is, this is a problem of mine, uh, or pet peeve maybe. So first off, for those who are, uh, who may not be in the know, what Isarge is describing as a level 0 is also referred to as a context diagram. But there's a subcontext, or there's a, there's a broader context that gets hidden when you do data flow diagrams. And, and, you know, First off, the first level of granularity that we lose in context diagram is we lose architecture. We lose a lot of the deployment characteristics, right? The network and some, and the environment in that, you know, around that system. And so when we set a context, we say, oh, here's, we have their software and our software talks to a DHCP server and our software talks to an LDAP server and our software talks to the database. Never mind that it passes out of our building and off often to the internet, or it goes off and it runs into this, into the SEM or to this data collector or whatever. And we lose a lot of that information that is important for this purpose. And when it comes to privacy, because we're modeling the system, we're not modeling the processes around the system, meaning the system lives within a broader context and it's interacting Yes, it interacts with users on the outside, but those users are customers. Customer data then comes into our system and goes out into that database and gets processed in a certain way and then comes back to our, to our system that we're modeling. We lose a lot of that context unless we, unless we actively make the point to call it out. And so when we traditionally do threat modeling, we're losing a lot of that data by design. By design, right, in the process to reduce complexity of the process.

**Chris Romeo (13:58)**: I want to come back around to this argument that Izar was making about there's a difference in complexity between security and privacy. When, and so if I start with the complexity from a security perspective, I'm measuring the complexity of the feature, the thing that I'm building. The security itself isn't complex. It's how complex the thing that I, the thing that I'm building is more difficult to secure because it's complex. And so I would say the same applies from a privacy perspective. The more complicated the feature that I have that's interacting with the data, the more challenging it is to be able to protect the privacy of that data. So the complexity is tied to the feature, not to the security. I mean, security can certainly be complex, don't get me wrong, and privacy engineering disciplines could be complex as well, but I'm really focused more on the middle, the thing that we're trying to secure, the thing that we're trying to keep private. The thing itself is where the complexity is dangerous.

**Izar Tarandach (15:08)**: I was talking with Kim the other day, and I asked her to explain to me Really, what are the challenges with privacy threat modeling, right? As a discipline. And we got into this back and forth where she would explain to me the privacy approach, the privacy issue. And I would try to play math and look how to apply formal methods to that and make the thing like almost math-like and a problem that you can actually attack with the tools that we already have or something approximate to that. And every time that I tried to do that, Kim would step back with, but wait, what about this? And what about that? And just adding levels of complexity into the process itself, not into the thing called privacy. That kept being, probably mischaracterizing it, but in my simple head, it kept being a simple problem. But then it had layers and it had interactions and it had offshoots and it had the second order results that I just couldn't attack with the tools that I had. And that's where the complexity lies for me.

**Chris Romeo (16:23)**: Uh, Izar is using Kim, Kim is Dr. Kim Woods, just in case anybody doesn't know who Kim is.

**Izar Tarandach (16:28)**: By now you guys should know who she is.

**Chris Romeo (16:30)**: You should know, you should know, but I want to make sure everybody does because she's brilliant. And, uh, is the one who teaches us about all these privacy things.

**Matt Coles (16:40)**: I think based on what you were just describing, there's our, and this sort of solidifies a concept or an idea I was just thinking, uh, to relate here. It's, it's kind of like, um, kind of like what's a good, I, I'm losing the analogy here, but if you have levels of something, and you start, if you start from, if you start from here and you work your way down and we're talking, I'm talking about decomposition and right, reducing complexity and whatnot. If you start from here, you work your way down, at some point you'll reach the floor. But with privacy, you start up here and now as you work your way down, do you ever reach the floor? And so if your level to begin with is already, already higher, if your high watermark is already twice or 3 times or 4 times as high as where you start from a security standpoint, then you may never reach the floor or you have to do a lot more effort to reach the floor.

**Chris Romeo (17:37)**: So why, why is that though? Why is the, why is that mark so high to start with and how you're describing it?

**Matt Coles (17:43)**: I think it, yeah, I think it, I think it's because, I think it's because of that notion of, of by element versus by system. And so when I, when you look at security, you can say I have eliminated Element A talks to element B over a network connection, right? Whereas with privacy, I have to know I have, I have a, I have system, gotta keep in frame. I have a system and that system talks to a system, but that system could be very complex in and of itself, right? Look, think about the, the, think about the context. When we, when we're modeling a system, we look at that system context, that, that thing is a box. When we model privacy, we need to look at the ecosystem that it sits in.

**Chris Romeo (18:25)**: Yeah, maybe I'm doing it wrong then, because I don't think of privacy, I don't think of data at the system level when I'm going through a threat model. I think about what data is passing through that element and what data is being stored in that element and what are the properties of protection for that particular data. So maybe I'm doing it wrong, but I tend to not think of, I don't think of privacy at the, I don't step back to the 50,000-foot view looking out the window of the airplane trying to figure out where all the data is. I'm down in the weeds looking at the data that I can certainly combine it all at the end, but I want to be down at the element level thinking about the individual pieces of data.

**Matt Coles (19:09)**: I think the problem with that is, you know, you lose, again, you lose a lot of context when you do that, right? And I'm not saying it's the wrong thing to do. I'm just saying that I think we may do ourselves a disservice by doing it at that level. Because when you look at a piece of data that comes into your front end, well, you can, if you don't know, if you know that that's data that is privacy-related data, right? So whether it's PII or PHI or whatever it is, right? Some sort of data that needs to be protected from a privacy standpoint. But you look at it as, as data coming into your front end, to your web application, and you're going to do something with it. You can secure it there and you can protect its privacy there. But what you don't know is, was it, was it collected? Was it delivered by the user with the expectation that it would be privately protected? I think, yeah, you don't know what it should be making that collection. You don't know what its collection parameters were. You don't know how it, how it made it to you, right? There's, there's stuff that happens outside of that scope, which makes this privacy challenge complex.

**Chris Romeo (20:15)**: I wanna ask you a question though, in response to that, because when I'm talking about privacy, thinking about data at the element level, the scenario you just described, how do you even know what data's in the system then? I know what data is in because I'm dealing with it at the, I'm, I'm collect, I'm figuring it out at the element level and then rolling it up. In your scenario, how do you even know what data is being processed without looking deeper?

**Izar Tarandach (20:38)**: And that's exactly the complexity that I was talking about, right? You just pointed exactly at it. You are looking at a lifecycle of the data, the data inside the system, how it moves, where it gets stored, how it gets transformed. The thing is that when I was looking into it with Kim, all of a sudden, the data had a different lifecycle. As what's my expectation towards that data? For how long is it going to live? Where it came from? Where is it going to? Does it have the same privacy value? And that's something that I'm just inventing here, that I think. Does it have the same privacy value in this system as it has in that system, right? So an example that we keep going back to, the famous, Somebody erased all my Apple devices hack of a couple of years ago. Somebody called to Amazon, said, I am such and such. And they said, we can only give you the last 4 numbers of your credit card. And they got it. And then they called Apple and Apple asked, okay, to prove that you are who you are, what are the last 4 digits of your credit card? Right? So that value just, just, it's the same data. It has different values in both sides of the equation. And there's no rule, there's no threat model that we can put in place that's going to absolutely cover that, that case.

**Matt Coles (21:59)**: You know, one other thing I, I wanna highlight here, actually, I, again, this is actually, this has been pretty thought-provoking just sitting here talking to you both about this. I think the other, the other piece I think that is important to highlight is we always go down. Decomposition. So we always start in the context and work our way down and decompose and decompose. When I'm talking about having that—

**Chris Romeo (22:26)**: I was just thinking about having that bodily decomposition metaphor there that you were like decomposing, decomposing.

**Izar Tarandach (22:35)**: I was just thinking of Monty Python and the decomposing composers. Okay, go.

**Matt Coles (22:41)**: So we all, but we, we tend to go down, right? We start with, we start with the context and we work our way down. When I was talking about the ecosystem, we need to understand that ecosystem view, that doesn't mean we need to start with the ecosystem. We could start with a level 0 and go level -1.

**Izar Tarandach (22:58)**: Yeah. That's what I said, step back. Yeah.

**Chris Romeo (23:00)**: Right.

**Matt Coles (23:00)**: Well, right, exactly. So, so rather than thinking about, 'cause you asked Chris, well, what, how do we know what data is in the, if we look at the ecosystem, how do we know what data is there? Well, we don't. We have partial knowledge. We have partial information about the ecosystem and that's okay. Until we start doing the level 1, level 0, level 1, level 2. And then we put that back into context with the ecosystem. The ecosystem is our new level 0, right? And so really we're starting with level 1 probably. And then, and then we work our way down and then we work our way back up and eventually we have this full stack.

**Chris Romeo (23:35)**: I think he may have just invented something. Like, wouldn't it be funny if I made it in the room?

**Izar Tarandach (23:41)**: What do you even do when the privacy value of a piece of data changes over time?

**Matt Coles (23:48)**: The same thing we do with security over time. I mean, it's a state, it becomes a state machine that we don't, we don't do state machines, but we should.

**Chris Romeo (23:55)**: Threat modeling is a continuous journey, as I learned in a document called the Threat Modeling Manifesto.

**Matt Coles (24:01)**: And we have to, we don't need, we, we haven't introduced zero trust yet. We haven't brought up the zero trust word yet.

**Izar Tarandach (24:07)**: If you had no trust, you guys don't have to talk about zero trust.

**Matt Coles (24:10)**: Your system state changes over time. And so threat model, when you threat model and use a data flow diagram, you're giving a static window into that, into that state. But if you use state machines—

**Izar Tarandach (24:22)**: No, no, no, wait, wait, wait, wait. I get it that like over time, okay, security changes and threats change and all that good stuff. Actually, I just had a conversation with Adam last week about that, but that was really interesting. but, privacy. If the privacy value changes over time and I have data that was in my system that was not considered private for some, some reason before, that cat is not going back into the barn just because now I'm, I'm, I'm modeling different way with different values. That data is already there and used by the system and perhaps even unprotected because yesterday didn't have privacy value.

**Chris Romeo (25:02)**: Yeah, but I mean, you're asking for time travel as a security control or privacy control. Always.

**Izar Tarandach (25:07)**: Always.

**Chris Romeo (25:07)**: That's always is not, you, you can't, we can't travel, you know, we can't get in a DeLorean, run it up to 88 miles an hour and run backwards and re-architect our privacy requirements because they changed 20 years in the future. Right. What, like you said, once that, once it's out of the barn, it's out of the barn. But that was a, that was a decision made. At the beginning of the system, there's no—

**Izar Tarandach (25:31)**: With the information that you had at the time, really.

**Matt Coles (25:32)**: And your only, your only option is to rework. I mean, that's what rework is for, right? Or refactoring, right? And so—

**Chris Romeo (25:38)**: But his point, Izar's point is you can't get, once it's out though, you can't put private data that's been released back in the barn and say, it's private now.

**Matt Coles (25:46)**: Yeah, that's, that's true.

**Izar Tarandach (25:48)**: I mean, you have to do some sort of corrective—

**Matt Coles (25:48)**: Yeah, that's what CDRs do. You have to do some corrective action. You do have to have some sort of, I mean, we as, as system designers and security engineers and privacy people, right? Regulations decide that we are going to do, we're going to have to do refactoring and rework and some remedial action there. And that's fine. That's at the ecosystem level. One other thing, I think on your point of privacy, you know, data becoming private, if the privacy value changes and you're thinking about data as a classifying data as private or non-private or requiring privacy protection. I want to— sort of a mind-blown discussion there is not only the data wasn't private because the privacy regulations didn't exist at the time you built the system, but because privacy can be subjective, a user can make claim to their data or not over time. And so data that you did collect under privacy rules may then be released, or data that you don't collect under privacy rules may become opted in at some point to privacy. So it isn't just that your, the system, the, the regulation or, or the requirements change because of outside factors, but inside factors as well. Mm-hmm.

**Izar Tarandach (27:01)**: I'm going to act like I understood that and I'm going to nod very slowly. Wait, I'm thinking in a loud voice. Okay.

**Chris Romeo (27:08)**: No, I'm, I'm tracking. I, I, it makes sense.

**Matt Coles (27:12)**: It's, well, because, because privacy, because privacy has a, because privacy has an opt-in or opt-out.

**Chris Romeo (27:17)**: Right. We generally— And that could change in the future.

**Matt Coles (27:19)**: Right. Here in the US, we generally do opt in, sorry, opt out of things, whereas in, you know, much of Europe, it's an opt-in decision. But the state of value, state of data may change within the system context, or it may change because GDPR 2 comes along and now everything has to change, right? Or depending on where you sell things, right? If you sell product into, or if you're storing data for Chinese users or California users, but not Mexico users or, or, uh, New Mexico users, right? Where state privacy changes.

**Chris Romeo (27:55)**: So I think the takeaway here is you should have protected all the data that your system ever has, could have, or will have in the future.

**Izar Tarandach (28:08)**: Might have.

**Chris Romeo (28:09)**: Might have. Should have been completely locked down from 1970 forward, protected by formal methods, uh, which I didn't even get a chance to, to address that one. But formal methods in my mind breed complex— more complexity than they're worth.

**Matt Coles (28:28)**: They shouldn't, they shouldn't. But yes, we had—

**Izar Tarandach (28:31)**: we shouldn't, we shouldn't, but we tend to do badly.

**Matt Coles (28:37)**: Look at MISRA, right? If you look at MISRA as a as a coding standard. It is not a coding standard in terms of, you know, certain language constructs, but it has an important statement about the complexity of algorithms, complexity of functions. The number of statements in a function needs to be reduced so that it's not complex on a per-function basis. You can do manual, you can do formal methods and manual code review.

**Izar Tarandach (29:05)**: That's something that I never understood. How do you measure complexity by number of lines of code?

**Matt Coles (29:11)**: Cyclomatic complexity, I think, is the measure.

**Izar Tarandach (29:13)**: Yeah, but, uh, you know, that, that will see people write like long functions just to avoid complexity.

**Matt Coles (29:22)**: This probably matters more in C than it does in, in other languages, right? Where in C you have so many different ways that you can do pointer, uh, you can do pointer manipulation in C. And therefore you have lots of, like, what can you do in a for loop? And how much do you have, how much can you do in the for loop rather than building functions that are running other lines of code, right? You can do stuff in a for loop in C that you can't do anywhere else.

**Chris Romeo (29:50)**: And so remember, with complexity comes job security.

**Izar Tarandach (29:56)**: Ah, good point. Very good point.

**Matt Coles (29:58)**: So here's a question. How come, how come we never had an opt-in or opt-out feature for security, but we do for privacy?

**Izar Tarandach (30:08)**: Oh, we do have an opt-out version. Do we? Yeah. Accept the risk.

**Matt Coles (30:14)**: Where do I get that slider? Where do I get that slider on?

**Izar Tarandach (30:18)**: It's not a slider. It's something that I keep being shouted at.

**Chris Romeo (30:22)**: Yeah. Yeah.

**Izar Tarandach (30:22)**: Yeah.

**Chris Romeo (30:23)**: It's going to be okay.

**Izar Tarandach (30:23)**: We accept the risk.

**Chris Romeo (30:24)**: Yeah. Yeah. Yeah. That's a, that's a, Matt, that's a very compelling question. Why do we not have an opt-out, opt-in? We have opt-out, opt-in. We don't really have opt-in. Opt-in, like you said, is a European construct. Opt-out from a privacy perspective is more of the US world, US or market or whatever. But yeah, why don't we give people the ability? I mean, now you're starting to talk about loosening. They're coming all the way back around full circle. Like Matt's, Matt's asking for loosening. I'm not asking for loosening.

**Matt Coles (30:56)**: Hold on. Just, I'm not asking for loosening. I'm just asking, why did we never think about this?

**Izar Tarandach (31:01)**: No, no, no, no. Wait, wait, wait, wait, wait. So opting in into security is actually building our security program. Opting out is accepting the risk. What's the big deal?

**Matt Coles (31:12)**: As a user, can you say don't use TLS?

**Izar Tarandach (31:14)**: Oh, as a user.

**Chris Romeo (31:16)**: Yeah, as a user perspective.

**Izar Tarandach (31:19)**: Wait, wait, wait. Let's think about this one. Can I opt out as a user?

**Matt Coles (31:22)**: I don't want, I don't want data. If somebody decides they want more.

**Izar Tarandach (31:27)**: What?

**Chris Romeo (31:27)**: We're getting every buzzword here.

**Izar Tarandach (31:32)**: I want a lot of security badges in Shadow IT.

**Chris Romeo (31:35)**: I can't believe we haven't had GenAI come up yet just to try to complete the entire circle. And what about?

**Izar Tarandach (31:42)**: Where do you think that all these ideas are coming from? Wait, wait.

**Chris Romeo (31:47)**: It's probably GenAI backed by quantum crypto algorithms would be my guess. Oh, just 'cause I want to get another bingo. For those playing the bingo, the security table bingo.

**Izar Tarandach (31:58)**: Oh no, no, no. Let's block that chain right here. Let's, he's gonna go there.

**Chris Romeo (32:05)**: Oh man. We're just in the, we're in the midst of machine learning. Mm-hmm. Ooh.

**Matt Coles (32:12)**: Yeah.

**Izar Tarandach (32:13)**: The only machine learning that I do is reading, uh, user manuals.

**Chris Romeo (32:21)**: All right.

**Matt Coles (32:21)**: So we're at the wall.

**Chris Romeo (32:24)**: What have we learned here? What's been the point of this 32 minutes of conversation?

**Matt Coles (32:30)**: We use, we use, we use CSM online for idea generation, but we ignore the article.

**Chris Romeo (32:39)**: That's a good takeaway. Not a bad takeaway. It's, you know, thank you CSOLime for inspiring us to talk about something different.

**Izar Tarandach (32:46)**: We learned that privacy is hard, mkay?

**Chris Romeo (32:52)**: That's a good joke too, privacy is hard, mkay?

**Izar Tarandach (32:57)**: Wait, that goes into the draft? Where is the—

**Matt Coles (33:00)**: Oh, I'm just scanning the article now to see, there's really not No, we're just completely— we had a better topic.

**Chris Romeo (33:13)**: Yeah.

**Izar Tarandach (33:14)**: Wait, wait, let's look at the last paragraph and see. Finally, finally, I see new category of companies I call SIEM, or more accurately, security operations supplementals. This includes detection engineering tools and services, threat intelligence aggregation firms, advanced analytics systems, and the like.

**Matt Coles (33:39)**: Wait, you, you went there. Why did you go there?

**Chris Romeo (33:43)**: This is, I mean, we, we've gotten so far off the, we've gotten so far off the article that, no, wait, but, but, but, but why do we need a new category?

**Izar Tarandach (33:52)**: Oh, it's CSO online. Right, right. If it doesn't have a quadrant to put it in, if it doesn't have a, a, a category, it doesn't exist. Yeah. No. Okay.

**Chris Romeo (34:00)**: Yeah, yeah, yeah, yeah.

**Izar Tarandach (34:01)**: I got it. Yeah, yeah, totally agree. Yeah, we need a new category and at least 4 or 5 new quadrants and whatnot. Yeah, sure.

**Chris Romeo (34:09)**: Categories make the world go round.

**Izar Tarandach (34:12)**: We should. Totally, totally.

**Chris Romeo (34:13)**: We should have it in, in, in the Inner Harbor in Baltimore.

**Izar Tarandach (34:19)**: Oh, well.

**Chris Romeo (34:19)**: Appreciate that, you. Yeah, I think, uh, I think we've done enough damage to the world.

**Matt Coles (34:25)**: I think we're out of gas.

**Izar Tarandach (34:26)**: And it's only Monday morning.

**Chris Romeo (34:30)**: Hopefully this, I think we had some thought-provoking things. I think, I mean, my brain was moving here. I mean, you guys challenged me in some of my thinking about intersection of privacy and security and complexity and all these things. So this is all good stuff to explore.

**Matt Coles (34:46)**: We need Dr. Wutz to come on and tell us how completely wrong we are.

**Chris Romeo (34:50)**: Yeah.

**Izar Tarandach (34:50)**: You know what? We've talked, we quoted her a lot. I think it's about time we invite her in.

**Chris Romeo (34:55)**: Yeah, she can just do a lecture to, uh, to teach us all to correct all the things that we've said that were incorrect. I know she's a listener.

**Izar Tarandach (35:04)**: So can I opt out?

**Chris Romeo (35:07)**: Uh, no, you may opt in. It's your only option.

**Matt Coles (35:10)**: Well, we can mute you. Is that an opt out?

**Chris Romeo (35:13)**: That's an opt in. Uh, it depends on whether it's whose perspective.

**Izar Tarandach (35:18)**: I'll be writing on that. I'll be writing a book.

**Matt Coles (35:23)**: Help.

**Chris Romeo (35:23)**: I'm trapped. All right. Well, folks, thanks for listening to this episode of the Security Table. Get out there and decomplexify your security and privacy.

**Izar Tarandach (35:34)**: But don't decompose.
