40 minSeason 2, episode 24

The Illusion of Secure Software

With Chris Romeo, Izar Tarandach, and Matt Coles

Secure software development

In this episode of The Security Table Podcast, hosts ChriS, Izar and Matt dive into the recent statement by CISA's Jen Easterly on the cybersecurity industry's software quality problem. They discuss the implications of her statement, explore the recurring themes in security guidelines, and debate whether the core issue is with people or technology.

Episode title image: The Illusion of Secure Software

Listen to the conversation

About this episode

In this episode of The Security Table Podcast, hosts ChriS, Izar and Matt dive into the recent statement by CISA's Jen Easterly on the cybersecurity industry's software quality problem. They discuss the implications of her statement, explore the recurring themes in security guidelines, and debate whether the core issue is with people or technology. Join the conversation as they analyze the roles of developers, QA engineers, and emerging AI tools in shaping a secure future, questioning if the industry is on the right path to real change.



Follow AI Security Table:

➜ Home: https://securitytable.ai/

➜ X: https://x.com/AISecurityTable

➜ LinkedIn: https://www.linkedin.com/company/ai-security-table/

➜ YouTube: https://www.youtube.com/@AISecurityTable

Transcript · 40 min conversation

0:10Matt Coles Ready? That one works at least.

0:14Chris Romeo Hey folks, welcome to another episode of the Security Table Podcast. My name's Chris Romeo. I am joined by Matt Coles and his pack of dogs. Also, Ezar Terandash on assignment. In the forest somewhere where he'll be reporting on the weather conditions in the forest and the music scene and a number of others.

0:37Izar Tarandach I can hear charging.

0:42Chris Romeo Oh, that's gotta be a flashback to a movie like Platoon or something like that. So, ah, it's been an interesting. Week, and I guess it's an old curse. Maybe you live in interesting times. I think Black Hat DEF CON just wrapped up as we're recording this podcast, and we have a statement that was made, a place that we're going to start. And so CISA has gotten a lot of attention on the security table. We've analyzed a lot of the things that they've put out. And tried to stay as positive as possible. Some of us more than others on the positivity side. But Jen Easterly, who is, let's see, Director of the Homeland Security Department Cybersecurity and Infrastructure Security Agency, who I refer to as CISA, came out and made a statement at Black Hat. And it was a, it was a, it was a very groundbreaking statement. She says, and I quote, We don't have a cybersecurity problem. We have a software quality problem. And I'll just, I'll read the next line because there was another quote. We have a multi-billion dollar cybersecurity industry because for decades technology vendors have been allowed to create defective, insecure, flawed software. So let's start with the quality thing. I don't feel like this is the first time I've heard this statement. What about you guys?

2:12Izar Tarandach I feel like she's probably a listener of this podcast.

2:16Matt Coles Yes.

2:16Chris Romeo Secretly, secretly a listener of the Security Table because she's like, what are they saying about us now? Yeah. See if we can make some changes.

2:26Matt Coles Yeah. This is, I mean, why, why should cybersecurity be different? What, why does it need to be different? And, and I guess. Companies have made it different for a long time. We see this all over the place, but why? We often talk about, I mean, if you, if you think about quality, if you think about quality, we're talking about, does the product, does a thing function the way it's expected to function? And does it not do things it shouldn't be doing? And gee, where have we heard that term before?

3:00Chris Romeo Right?

3:02Matt Coles So when you have security issues and you have a thing, have something doing something it shouldn't be doing, having a side effect or having extra functions that it isn't supposed to have, like, uh, say the ability to execute code remotely, um, then, then that sounds like a quality problem. That sounds exactly like what we would want quality to find. Why should, why do we make security differently?

3:27Chris Romeo Isn't it funny though, that quality doesn't exist? Like QA stands for quality assurance.

3:35Matt Coles Yes. And there's a whole organization and function called quality engineering.

3:40Chris Romeo But most of the, but most, but most places don't have it anymore.

3:44Izar Tarandach Yeah. We spoke about this a couple of weeks ago.

3:47Chris Romeo Yeah.

3:47Matt Coles The name is still there.

3:48Chris Romeo It's gone away.

3:48Izar Tarandach The function doesn't exist anymore.

3:50Matt Coles Yeah.

3:51Chris Romeo And so I just, I have to chuckle about the statement because I first heard this from, from leadership at Cisco back in, let's say 2007, 2008, there was a big, Cisco did a big, uh, quality push, you know, and, and, and people started saying, and maybe it was a little bit later than that, maybe it was in the 2010s or so around there, people started saying security is a quality problem was the statement that was made. And so I don't feel like this is groundbreaking by any stretch. Like this is, this is old. This is something that we've been talking about for decades. Decades. But before we started recording, Izar asked a great question, which we didn't answer, so we'll do it now. Why do we still, why are we still making this statement 15 years later? And it doesn't feel like anything's really changed because it is still a quality.

4:47Matt Coles If I may, Izar, it is still a quality problem. The statement is still true. The problem is we've, we've sort of A lot of organizations have either divested of quality or changed their, their, their definition and their capability for quality. As you mentioned, right, a lot of teams don't have a quality engineer on staff, right? They rely on developers to inject quality into the code that they are writing themselves, by the way. Quality engineering used to be a check and balance against developers, right? You had somebody who wrote the code and somebody who tested the code. And that way, you know, you don't get a developer writing a code, testing the code, and checking the code in and pushing it to prod before, without knowledge of some assurance, right?

5:32Chris Romeo And it is a shame that we've done away with that. The more I think about this issue, the more I think we missed something. When DevOps came on the scene and it all became about, it's, it's the historical problem was it's dev versus ops. Devs throwing things over the wall and somehow, I don't even know how this happened, but somehow quality assurance got dropped in that typhoon of DevOps hitting the scene. Is our—

6:05Izar Tarandach so, yeah, I'm, I'm, I'm okay. So I'm, I'm going to take a bit of liberty here and paraphrase her.

6:10Matt Coles We don't have a quality problem. We have a people problem.

6:14Izar Tarandach All the, the. The issues that we are seeing, in my humble view, are self-inflicted. And they come from the point where we said that the biggest value in this, in any industry that does software or even hardware, is to get to market first. The one that gets to market first wins the race. And Over that race, we, we as, as professionals, as people, not only we became ready to drop those things that we saw as hindrances to, to being the first, we also started making a number of assumptions that turned out to be invalid. And we came up with a number of hopes that turned out to be invalid. So the. The assumptions that we seem to have made are that developers are not only able to write perfect code, they are also able to test that code and to bless that code by themselves without any, any other, uh, help. And we have put into ourselves the hope that security is something that can be easily fixed with a silver bullet, some amazing tool that you're going to run on the system once it's ready to go to market. And 15 minutes before you press deploy, it's going to fix all, all your security issues. So I think that quality is, is a huge demonstrator of the problem, but it's not the cause.

7:55Chris Romeo And so what if we took this a step further and said, now let me, let me extend your people problem and see if you agree or disagree with this. We have a people problem and a technology problem. Yes, the people don't have the ability to create secure code. Hence, if we go back to the, the beginning assumption or presumption that quality code is secure code, if we, if we have a people problem, I would argue we also have a technology problem. Because I think part of the challenge here is that we thought technology was going to provide that check and balance between before what QA did. We thought our SaaS tools and our SCA tools and things were going to help us get around not having QA anymore. And I don't think anybody did that deliberately, but I'm just kind of, you know, thought experiment, kind of working this out, uh, live. But pe— but what do you guys think? I mean, do we have a people and technology problem or just a people problem?

9:00Matt Coles I think we, I think we have a, a more, it's more than, it's bigger than that. But as soon as you start going bigger than that, it starts to snowball. We have a people, we have a technology, we have a process, we have an organizational, we have a market problem. We have all of those things, right? We have technology that's hard to get right. Right? We have unclear requirements and understanding of things and, uh, and, and a plethora of patterns and styles and mechanisms. And, and then you have, uh, a rush to market, right? We have, we've taken the iron triangle of resources, time, you know, of people, time, and functionality. And, and, and made it like a, like a, like a dodecahedron or a sphere of lots of different points. And it's really hard to balance all of these things. We have, we have created this, this monster. Um, I would, I, I guess we tried to, the, the whole tool pro— the whole using tools to fix the problem came from a flawed, comes from flawed assumptions about the nature of the problem, right? And it is only now that we have that start— things are starting to line up, right? Computing resources were, were what they were a decade or 20 years ago. And now you have, you know, uh, memory-safe languages and, and things that are able to be self-healing or self-protecting. There's a lot of overhead in all of that, right? Remember when C++ happened? Remember from C to C++ and trying to forget that you, you had exception handling and stack unwinding and, and a whole bunch of other things that were just adding performance bottlenecks, right? But all of the defensive features that we now have are performance, are performance bottlenecks among other things, right? And it's a, it's a whole web of things. I mean, uh, think of Heartbleed, right? OpenSSL and, and, uh, the TLS You know, the stack that runs TLS and, and, and the state machine required in order to manage all the connections and everything. It's just super complex. Um, and so we tried a lot of bolt-on solutions over time. Only now, I think only recently, we've had an opportunity to, to make it easier for developers to write secure code, which means they can now concentrate on quality. Although we've now taken quality out of the picture. And so, I mean, now, now we started a new, a new cascade.

11:42Izar Tarandach I, I, I like the, the way that you put it, that in, in the snowball framework, but I think that it's even a step further. I think that it's recursive. I think that's because it's a people-created problem. When we generate new technology, it carries the problem of the people who created it. And then that goes to the next step and, and, and forward.

12:08Chris Romeo So it's technical debt to the nth power.

12:11Matt Coles Yeah. Well, and on top of technical— oh, sorry, is there— No, no, no. Just the technical debt side. Oh, by the way, all of these organizations that have been pushing stuff to market and have backlogs, I mean, like, like, like, there's this stack of backlog. Where'd all that go? Who's addressing the technical debt? And by the way, that technical debt isn't necessarily just security problems. It's quality defects. It's, uh, it's refactoring waiting to happen. Uh, and it, and just, it's just digging a hole, digging a hole, digging a hole. And, and at some point, because nobody, nobody starts over, or very few, very few.

12:53Chris Romeo Isn't that the nature of software though?

12:55Izar Tarandach No, let me put it like this. People will put much more time and money into A/B testing for the right color of the background of the column on the side of the application than they are going to go into solving technical debt.

13:09Chris Romeo Well, because that's a part of—

13:10Matt Coles Or which browser you're going to support.

13:13Chris Romeo Yeah, but if you, if you, but it all comes down to what helps you sell more products. No. At the end of the day.

13:21Matt Coles Okay.

13:22Izar Tarandach That's a very interesting way of putting things because if you gave the option to a customer saying, I have X time, I can either put it into getting just the right shade of blue or into solving this thing that can put your data in jeopardy in these edge cases. But nobody asks the customer because we have a bunch of product managers out there who will look at technical people and say, we know the customer. We are the customer. We've been the customer. And no, you haven't. No, you aren't because you keep creating the same problems again and again and again. And not to paraphrase Einstein, you people are mad. You just keep doing the same thing again and again and again, expecting something different.

14:15Chris Romeo It's the definition of insanity.

14:16Matt Coles So that's a, so that's a great segue into the other part of this conversation is when do customers start asking for security over that shade of different shade of blue?

14:27Chris Romeo We should just to set that up. So CISA came out with a Secure by Demand guide, which I just a few minutes ago figured out they mean demand in the buying something perspective. Because when I first saw that, I was like, secure by demand? Is that like, I don't know, man. Like, is it something that just happens? Like you think about it and then you raise your fist and then the thing locks down or something?

14:52Matt Coles Have you never heard the term demand plan? You never heard demand planning?

14:57Chris Romeo No, I don't do any demand planning. I just, I make a lot of demands in life. Don't get me wrong. I understand that side of the equation, but I don't, I don't actually plan it. It happens in the moment where I'm like, wanting whatever I want. So, okay. So, yeah. Secure by demand.

15:14Matt Coles So, so the, so the CISA, CISA is secure by demand, right? So when customers are acquiring software or systems, CISA's, so this, so this came out of the executive order of the, you know, the Biden presidential directive on cybersecurity and, and the executive order that went with it. And, uh, You know, there was a call for, so there was a call for organizations that were building products to take ownership on behalf of their customers of security. And so now we have the, the flip side of that is getting customers to demand more from their suppliers and providing a guide for suppliers to be able to acquire, to acquire and ask for security in the systems that they, that they're purchasing. Oh, is there's only my batteries are about to die.

16:13Chris Romeo Roving consultant. So I guess, Matt, if we set this document up, I'm looking at it right now. I grabbed the PDF version. There was like a web version, I think was a tiny bit more of a summary. It looks like there's a little more detail in the PDF itself. And so It is broken down. The, the, the meat of the PDF is the software manufacturer questions section. And then under that, it has various headings like general questions, authentication, eliminating classes of vulns, evidence of intrusion. So it tracks with the Secure by Design guidance that we've seen so far.

16:50Matt Coles Yeah, exactly. So, so the Secure by Design was telling, was telling suppliers, vendors, how to secure their systems. And now the demand guide is telling customers how to ask for those things to ensure that they get done.

17:03Izar Tarandach As this is, but it is still not, not addressing the problem because let's tear this thing up.

17:09Chris Romeo Let's tear it.

17:09Matt Coles It's gonna address the problem. It's gonna try and address the problem through, through, through money, right?

17:14Izar Tarandach I mean, but here's the wrong thing. It's assuming that customers will come and actually demand security. You know when customers are going to demand security? When we start selling software by $39.99. But if you want the really secure option, the plus, the premium, you're going to have to pay $45.

17:33Chris Romeo Yeah, but that's already— the market's already shown that that doesn't work.

17:37Matt Coles And we're already past that.

17:39Chris Romeo The piece— my piece of evidence to that is, I don't know if you've seen the backlash against people who try to charge for single sign-on. As a premium feature.

17:47Izar Tarandach Oh, I didn't. I totally missed this.

17:49Chris Romeo It's been a big, it's been a big brouhaha in SaaS, whether you, because what some people would do is they put the single sign-on option like in their enterprise tier and then their, their first tier doesn't. And so the argument is base security should not be something you have to pay an ultimate premium to gain. And in the market today, single sign-on is table stakes. It's not.

18:14Matt Coles Yep.

18:14Chris Romeo Something that's got a lot of innovation.

18:16Matt Coles And by the way, that was in, that was in the executive order. If you recall, the executive order said organizations should not, should not demand more for more for security.

18:26Izar Tarandach And that's great, but it goes against the psychology of the buyer. If people are not thinking that they're paying more for a premium thing that just they are getting and not everybody else, they're not going to go for that. And if we don't give an economic incentive to quality and security, companies are not going to go back and, and rehire the QA people and develop that, uh, capability again.

18:49Matt Coles Yeah, but it's, I mean, well, so here's a bigger problem that I think we're missing something, right? We're talking about enterprise software. The security problem isn't just with enterprise software. How many free apps do you have on your phone?

19:06Izar Tarandach Oh, I'm totally talking consumer. I'm totally talking consumer.

19:10Chris Romeo Unless it's from a reputable source, I don't have any—

19:14Matt Coles either way though, you can have those free apps and the security problem isn't with the app, or isn't with, isn't with that app, right? It may be a data collector. It may be an aggregator. It may be a host of other, other things in the whole supply chain or infrastructure. Right.

19:32Chris Romeo I mean, you're, you're, you're an Android user. You're an Android user, aren't you?

19:36Matt Coles I am. Yes.

19:36Chris Romeo You know, I can tell because in the Apple ecosystem that doesn't, I don't think those problems exist because of the way they've isolated. They've isolated a lot of things so that you can't do things as easily as you can in the, that's a whole, we should have a different episode on that.

19:54Matt Coles 1.9 billion people were, were, were affected by the, uh, the national data breach, uh, uh, Recently, I'm pretty sure your records are in there somewhere.

20:04Chris Romeo I gave up worrying about data breaches about 15 years ago because all our information is out there. It's just, it is just live with it, deal with it.

20:13Matt Coles But the fact that you're on Apple, the fact that you're on Apple doesn't matter.

20:15Izar Tarandach It's different, right?

20:16Matt Coles And if you need those apps, those individual apps may be, may be more secure because they're isolated, but the, but your data in the backend, all bets are off still.

20:24Izar Tarandach And, and they managed to isolate the fact that you lose 70% of the functionality that we have on Android for free, and Apple has to pay for it every 3 years, right? So that's an isolation that's completely different. But hey, let's not fanboy here.

20:39Chris Romeo I mean, you know, the only advantage of Apple is that it just works. So I've had Android before and, and they were trash devices that you had to be a system administrator.

20:52Izar Tarandach Definitions of work.

20:54Chris Romeo Exactly.

20:57Matt Coles It's freeism, freedom.

20:59Chris Romeo Enjoy, enjoy being a sysadmin for your phone. That's what I would tell you as an Android user.

21:05Izar Tarandach I was thinking the other day, I haven't rooted my phone in years. Like, I think double-digit years. Why? Because Android just works.

21:16Matt Coles Just works.

21:17Chris Romeo Just works. That's because you're, that's because you're getting older. That's what.

21:22Izar Tarandach That's true.

21:23Chris Romeo It's not as exciting anymore. No, I mean, it's the same thing. Like the more gray hair we get, the less exciting things like rooting a phone become.

21:31Izar Tarandach And all of a sudden I'm depressed.

21:34Chris Romeo That was not the best.

21:37Matt Coles Go hack your 3D printer.

21:38Chris Romeo Let's tear this, let's tear this document up. Okay, here we go. Let me just tell you the first thing that jumps off the page at me. I love it. There's a little box in the top right. It says artifacts you can collect from a software manufacturer, a software bill of materials, SBOM, that lists third-party software components used by the product. Yay. Why can't we just say something about how you should actually use it? One more sentence, please.

22:01Izar Tarandach Because you, you know what else you collect? You collect stamps, you collect coins, you just keep them somewhere.

22:07Chris Romeo That's the bloody problem with this is that it's like collecting stamps. And I just did an AppSec podcast episode with Steve Springett where he enlightened me on all things CycloneDX and I got back on board with it. But then I see something like this and it's like, just tell people that, and comma, and use this to better understand your risk from those third parties that you're doing business with and decide if you want. Yeah, see, exactly.

22:36Matt Coles But it doesn't really do that. It doesn't do that.

22:38Izar Tarandach The problem is you're going to use it to feel good about the fact that you have it because it's not going to enlighten you on the risk that you're carrying because nobody reads the thing and goes, hmm, that's ML2.

22:50Matt Coles Yeah, but also, also because just because a component is in the SBOM doesn't necessarily mean it's being used. It doesn't necessarily mean it's being used directly or indirectly, that it's been securely deployed, that it's been securely configured, that it isn't full of holes that have yet to be discovered, you know?

23:11Izar Tarandach But there are tools for that, right?

23:14Matt Coles Okay. We're back to, we're back to that whole QA thing, right? Yeah.

23:19Izar Tarandach Okay.

23:19Chris Romeo So then if we keep going, so here we go. All right, let me pick on this again, 'cause we just gotta get the picking on out of the way here. General questions, the first section, the first one, has the manufacturer taken CISA's Secure by Design pledge? What progress reports has the manufacturer published in line with its commitments to the pledge? Are they trying to create their own certification stamp for our industry and just kind of like guilt people into being a part of it?

23:47Matt Coles Is this, is this like the, the, the, there's a, there's another, there's other common pledges, uh, for some cultures. Uh, this is completely voluntary and really you gotta ask people to do the pledge.

23:59Chris Romeo Really?

24:00Izar Tarandach I just keep asking myself how many times is this going to happen that something goes boom and somebody looks around and said, but they promised.

24:13Chris Romeo I never thought of it that way, but they promised as you have a little tear coming down your cheek.

24:20Matt Coles So, yeah, so what do you, so what do you do? So what do you do if somebody is on, you know, signs up to the pledge and then boom happens?

24:28Chris Romeo Yeah, there's nothing. There's no, there's no ramifications. There's no liability. There's nothing here other than I feel sad because they didn't.

24:37Matt Coles There isn't even public shaming. There isn't even public shaming, is there?

24:41Izar Tarandach What happens when they, when they get a central idea from threat modeling and start saying my pledge is not your pledge?

24:52Chris Romeo It's true.

24:53Matt Coles That is true. We should have a threat in threat modeling for pledge-worthy.

24:58Izar Tarandach We should threat model pledges.

25:01Chris Romeo Stay tuned for the next episode where we threat model a pledge. Okay, there's another section here, authentication. Products should support secure authentication. Love it. Uh, should do single sign-on, should do multi-factor. Or other phishing-resistant forms of authentication like passkeys. Leave that one alone based on our last episode's conversation. Uh, have they eliminated default passwords? Like all of these, I mean, there's none of these things we're going to argue with. You got maybe the passkeys we could argue with a little bit or something, but this is all stuff we've been saying for years and it's stuff that's in the Secure by Design original document. Yeah.

25:35Matt Coles And by the way, this is what probably large organizations that have full, fully vetted and fully capable supply chains already do. Right. So who's this, who's this document targeted at? Is this target targeted at the 17-year-old who's downloading an app onto their phone?

25:50Chris Romeo I don't think so.

25:51Matt Coles Or is this targeted at a mid-sized business that doesn't have a security team on staff?

25:56Izar Tarandach I think it's targeted at the next big startup that's going to change the world, but leave security as a bolt-on for the end of the thing. You know, when it's deployed and people are buying and actually demanding security.

26:09Chris Romeo So next section is eliminating classes of vulnerabilities. Let's rehash this again. I feel like we spent almost the whole episode tearing this concept apart, but it's back again. So they, they, the question they say is what classes of vulnerability has the software manufacturer systematically addressed in their products? For those that they haven't addressed, do they have a roadmap showing how they plan to eliminate those classes of vulnerabilities? So, for example, like consistently enforcing the use of parameterized queries to prevent SQL injection attacks.

26:40Matt Coles So this ties back to their Secure by Design guides, you know, the SQL injection, buffer overflow, cross-site scripting guides.

26:49Chris Romeo What does this mean systematically addressed?

26:53Izar Tarandach Like, wait, before we go there, I just had a small realization here. As you were reading it, I had a bit of a déjà vu. We heard this before, and then Matt mentioned it goes back to the same. This is the third time that we are reading the same document, the same content. It just started as here's some guidelines, here's a pledge, and now here's how you ask for it, how you demand it. But it's exactly the same content. It's just being packaged in different ways that people can come and, and address the lack of it.

27:28Matt Coles You know what? It's brilliant. If that, if that was intentional, it's brilliant because it gets, because it gets the concepts in front of the suppliers, the buyers, and the people in between, right? The pledge is just fluff, but, but I mean, if you have consistent content, consistent messaging, and it's not boiling the ocean, Right? You're not, you're not MITRE with a CWE framework of, you know, hundreds of weaknesses or the MITRE ATT&CK framework with hundreds of ways of getting in and techniques and mitigations. And you start small. It's actually pretty genius, right? Get, get, get a, a start on something, get some quick wins, maybe quick wins. It doesn't solve the problem, but it does. Reduce the problem to some extent, right?

28:20Chris Romeo A little bit. All right.

28:21Matt Coles It's baby steps. To your point, Chris, about, uh, so what does this mean? So if you have, if you have a SaaS application, right? If your application, if your product is a SaaS application, you have addressed SQL injection throughout. You have addressed cross-site scripting throughout. As a product organization, if you have one product and you do it in one product, great. If you have hundreds of products, Systematic is, uh, very challenging, right? Because every product is probably developed differently and you have to address it in a way that includes, you know, making sure all your developers know how to do SQL, uh, to mitigate SQL injection and have a coding standard and have tools looking for that coding standard. Yeah.

29:01Chris Romeo You need a, you need a guardrail.

29:04Matt Coles Exactly. You need a paved road and a guardrail.

29:06Chris Romeo Well, now paved. So remember when we say paved, a paved, when I think of a paved road is something I give you as a developer. That makes your life easier.

29:14Matt Coles So here you may get a database, you may have a database access layer that's already been pre-written that already addresses SQL.

29:21Chris Romeo Yeah, that's true. I guess you could, you could approach it that way if you were like a single product startup. I think that's a lot easier than if you're a 100 product enterprise to say, just because we know how enterprise products are assembled, right? I bought this one from this company over here. I bought this one from this company over here. It's If people think that when a company acquires a product via mergers and acquisitions, that they rewrite that thing from scratch in their system, that ain't how it happens. They take it in, they say, well, guess what? Now we have two source code control systems. One of them is GitHub and one of them's Bitbucket. And then 10 years later, we'll hopefully bring them together and there'll be this magical moment where every, all this, all this source code now is in one place. But while we did that, we acquired 3 other companies. And so now we have 5 source code control systems, we have 3 software development lifecycles, and a partridge in a pear tree.

30:14Matt Coles Right. And oh, by the way, that technical debt may or may not have come over with the acquisition.

30:18Chris Romeo Oh no, it came. And it, and it, and it 10xed once it landed.

30:23Matt Coles No, no, no.

30:23Chris Romeo Met all the other technical debt. They made more.

30:25Matt Coles The doc— the documentation of the technical debt of what was there. So the, the Jira backlogs may not have come across.

30:31Chris Romeo True. Nobody ever ports those over. Just let those delete. Oops.

30:35Izar Tarandach I'm sorry, but just, just going back to the people problem. You forget that once you go into mergers and acquisitions, one man's technical debt is another man's job security.

30:48Chris Romeo There's a t-shirt idea.

30:49Matt Coles They're features, not bugs. They're features, not bugs.

30:53Izar Tarandach Yeah.

30:53Chris Romeo All right. So I think you're, you, you had a bit of an epiphany here, Izar. I think you're, you're onto something here. This is the same information repackaged 3 times. Uh, I will agree that it's brilliant only when I see the industry having changed in 3 to 5 years from now. If there's no me measurable change in 3 to 5 years, then this is all just noise that's being—

31:16Izar Tarandach so I'm going to Cassandra it and I'm going to say this is not going to change anything. This is going to make a lot of people feel very, very good. Warm in their tummies, this is not going to change anything.

31:30Matt Coles How much does it have to change in order to be measurable and meaningful?

31:34Chris Romeo Yeah, I like meaningful.

31:36Matt Coles This is a serious question, right? How do we judge meaningful change, right? If somebody, if an organization eliminates all SQL injections, but you still have data breaches because a database lives out on the internet or it's caught up in a data aggregator. Has it been meaningful change? How do you measure this?

31:57Izar Tarandach Okay, so this is going to sound funny, but it comes with a, with a measure of seriousness. I think that's the moment we can lower the rate of WTF per minute when we are looking at systems, then you know the thing changes.

32:14Matt Coles Okay.

32:15Izar Tarandach It still happens to me that I look at things that have been developed 10 years ago, 5 years ago, 2 years ago, 2 weeks ago, and I go, no, seriously, what was somebody thinking when they did this?

32:26Chris Romeo Yeah.

32:26Izar Tarandach With a bit of thought, you figure out that they were thinking of good things, but without the quality and without security and without the commitment to make something better. They just want to make something fast that's going to sell.

32:38Chris Romeo And I'll go a step further and say, I don't think there's going to be any measurable or meaningful change as a result of any of this stuff. Until we reach the point where we have to go foundational here. So if you want to eliminate SQL injections, you have to take out the ability for parameterized, for non-parameterized queries to be done at all. You have to put some type of technology in place that just doesn't let it happen. That's the only way you can systematically address—

33:06Matt Coles Are you talking about a Rust for SQL? Is that what you're asking for?

33:11Chris Romeo I don't know what I'm at. I'm asking for a concept. Isar is going to spell it out for me here.

33:15Izar Tarandach Yeah, actually, I'm going to go one step further, Chris. I think that we are at that inflection point.

33:20Matt Coles Really?

33:20Chris Romeo Okay.

33:21Izar Tarandach And we do have an opportunity right here. It's going to be missed and we're going to kick ourselves in the butt in a couple of years about it. But now that everybody is moving towards having a coding assistant in their pocket or in the window next to their code. Wouldn't it be wonderful if all these AI copilots and shit would actually write secure codes and tell you, no, no, no, no, no, no, don't do that query that way. Use this ORM thing. The problem is that because the thing is recursive, whatever we are creating nowadays to do that stuff does not do it. So we could go foundational right now and create a new foundation where those things that we say, oh, developers don't have the time, don't have the training or have the interest for those things. All these AI copilots could easily solve the problem because they could be the, the, the, what's the name of the, the thing from Pinocchio? The, the, the green, it's not a grasshopper. What is it? Um, the conscience of cricket.

34:20Chris Romeo Jiminy Cricket.

34:22Izar Tarandach So they, yeah, any AI copilots could take that, that role, right? And be the guy saying, hey, you know what? That's, that's a nice line of code, but I don't think that you should write it just like that.

34:32Matt Coles Oh my God. Clippy for code, huh? No, no, no, no, no.

34:36Izar Tarandach This is different from Clippy. This is different from Clippy.

34:39Chris Romeo Okay.

34:41Izar Tarandach Oh my God.

34:41Matt Coles Every time that somebody says Clippy, I go back to that joke about the suicide letter. Anyway, but no, you know, it's a, it's a great, it's a great idea. You're assuming, you're assuming that AI is going to, to take on that Right.

34:59Izar Tarandach That's, that's, that's a given.

35:00Matt Coles Okay.

35:01Izar Tarandach It's going to happen. People are going to use it more and more until the whole thing implodes on itself because the code is so crappy.

35:10Chris Romeo New models are going to be trained. That's worse than having developers write it by themselves.

35:13Izar Tarandach AI-generated crap, right?

35:16Matt Coles But, or you're consuming, or you're consuming 95% of your CPU and storage for the AI models so that you can write good code. You can't compile and run.

35:26Izar Tarandach But we are at that inflection point where we might have an opportunity to do something better because people are talking so much about changing the way that they do things, right? We in the security field, we have been asking developers for years, for decades to change the way that they do things. But there was never a, we may have been trying to do this with tools, like all kinds of linters and static code analysis and all that good stuff because we are saying, okay, we are going to point out what's wrong and we hope that the next time that you write it, you're going to write it right. Which is not the case, unfortunately, it wouldn't happen. Here we have this opportunity to actually change the way that people do things to a better way. But I go back to what I said in the beginning, it's a people problem.

36:14Chris Romeo Yeah.

36:14Izar Tarandach And it's not going to change.

36:16Chris Romeo But so on the technology side though, you still need, I still need a guardrail that prevents. Non-parameterized queries from even being possible. For this to be systematically addressed, it can't be, it has to be eliminated at the source where it can't be possible. So if you do it as a developer, the system just doesn't allow that query to run. It just firewalls it off.

36:40Izar Tarandach Until the point where Joe Primadonna, the chief architect of whateveritis.com, is going to come and say, that framework is impinging on our creativity and we cannot 10x our objectives. And achieve synergy with the world because we can't do queries the way before.

36:58Chris Romeo People are going the other way. People are going the other way. People are, people are not going to throw away the frameworks at this point because they save you. Like, are you, I mean, are you going to build your own React JavaScript front end engine? Because it'll take you longer than it does to build the product to build your own front end.

37:15Izar Tarandach Until Joe Primadonna decides that building the next React is going to be the thing that's going to make their name out there.

37:22Chris Romeo Yeah, but their main CEO and business person is going to say, no, we're not spending, uh, 10x the money to build something that you want. Use the framework that's been provided.

37:36Izar Tarandach Or, or they will come and say, uh, dear investors, we have Joe Prima Donna and he is developing this whole completely new React paradigm and it's going to change the, the, the industry. And we don't have to do anything. We just have to let their creativity go and do its thing. The, the, the problem is that the arguments in here, they go both ways, right? But at, at the end, at the end of the day, you have people trying to push, sorry, product out there. And the incentives for pushing product are that you either need it or you think you need it. And those are the, the, the, the realities, the economic realities that we as security people or even as quality people have to fight against.

38:23Chris Romeo Yeah. All right, Matt, give you the final word here. Isar and I have been talking nonstop for 5 minutes. So final word.

38:30Matt Coles Well, since we're, since we're coming back, coming down to the end of the episode, I, I want to throw out a concept here just as a bombshell idea of all the thinking back to the quality problem. With all the money that security teams and secure developer training and CISOs and, and whatnot are getting, are getting in terms of funding, which admittedly wasn't a lot and, and now is ramping up to be a lot. What kind of a coup would it have been if the director of quality or VP of quality engineering had said, I, I'm going to take on security as a quality problem. Give me that money and tackle it. We'd have quality engineering still as a function and they'd be tackling security.

39:17Chris Romeo Yeah.

39:19Matt Coles Just a thought.

39:20Chris Romeo It's an alternate universe kind of a thought experiment.

39:23Matt Coles Where would that be? What, where would we be today if quality had tackled security and not, oh, it's different. Don't talk to me.

39:32Chris Romeo Save that for next episode. That's a good one. I want to explore that one in more depth. I don't want to, I don't want to just touch the surface here. I think we should dive into that. Let's play alternate universe where things happened differently 10, 15 years ago and then brainstorm what it would look like today.

39:48Izar Tarandach We would be doing A/B testing on the color blue.

39:53Chris Romeo And with that, folks, we'll close this episode of the Security Table. We recommend hope that you'll sit around and also ponder A/B testing of the color movement.

Transcript supplied by AssemblyAI. It may contain transcription errors.

Originally published as The Security Table. Part of the AI Security Table archive.