49 min

Should security give up on developers?

With Chris Romeo, Izar Tarandach, and Matt Coles

Teams, careers & security culture

The gang discusses whether security should or could give up on developers. We explore what the development world would look like if security did all the security, and the developer's responsibility ended when they committed a PR.

Episode title image: Should security give up on developers?

Listen to the conversation

About this episode

The gang discusses whether security should or could give up on developers. We explore what the development world would look like if security did all the security, and the developer's responsibility ended when they committed a PR. Conclusions are eventually reached.

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 · 49 min conversation

0:00Chris Romeo Welcome to another episode of the Security Table. I'm Chris Romeo, excited to be once again joined by my friends Tanya Ward, Izar Terrindosh, Matt Coles, and we're here to have some fun with application and software security and a whole bunch of other things along the way. We got a fun topic though for today, which I think is going to result in maybe multiple hours of debate and conversation, but simple question that I'm going to throw out to the panel here that I think is going to balloon into something much bigger along the way, and that is Should security give up on developers in 2023? And so I got this idea from a JupyterOne article where they were talking about should, you know, security will give up on users as a line of defense. But it made me think that we think about developers a lot more. And so should security just give up on developers in 2023? Tanya, kick us off here. What are your thoughts?

0:53Tanya Ward Oh my goodness. Maybe I'll revert that. Should developers give up on security? No, we shouldn't give up. It's another way of looking at it. I mean, developers have a lot on their plate, right? Between, you know, they're only being asked to do more with less and, you know, again, following up with the tooling discussion. So no, I don't think we should. I think we should continually figure out the right way to equip them so that they can do the best job that they can in putting the right defenses in place within our products and applications. Zohra, you're kind of looking at me, so, uh, what, what's your thoughts?

1:30Izar Tarandach I'd say it depends. I mean, we have this huge population of developers, right? We have millions and millions and millions of them out there. We, we are— we as security practitioners, we are sorely outnumbered, and For a long time, we have been pushing this crazy notions of security champions and every developer becoming their own security champion, or all kinds of things like that. And then we—

2:06Chris Romeo sorry—

2:06Izar Tarandach and then we complemented that with secure libraries and security defaults and threat modeling and security first time of the day, and security is done. And all that good stuff, and the CVEs continued mounting up. So something, something was wrong there. And lately I've been starting to think that perhaps some developers should be more secure than others. So back to the, it depends. I think that perhaps we don't have to push as much security to every developer out there, but start thinking in terms of what is it that they're working on. And perhaps some developers should have— should be more security-minded than others.

2:55Chris Romeo How would you break that down? Like, what— what's— I'm curious as to your thought on what's— who would be more security-minded? Like, what would be your qualifications for somebody, or the— how would you determine that person needs to be more security-minded and that person doesn't?

3:12Izar Tarandach So I'm free-flowing here. I didn't give this any prior thought, but we keep saying, for example, hey, you shouldn't have any kind of security in the frontend. Perhaps we don't have to give so much security time to frontend developers. Perhaps we should finally, I don't know, concentrate on the backend ones and make sure that they are doing the right thing, because after all, that's where the funny stuff happens. And browsers are coming up with all kinds of better and better and better defense mechanisms anyway. So it's not a give up, it's a why don't we modulate how much security time we give to each class of developer? Is that something that we can even do?

3:58Tanya Ward Yes, sir. Do we do that for quality?

4:02Chris Romeo Did we?

4:03Tanya Ward I mean, I— yeah, like when I used to write code, I mean, it was always like producing the best, trying to minimize the amount of bugs in the system. So why would security be any different? Did I stump you?

4:20Matt Coles Yes, you stumped this one.

4:22Izar Tarandach I'm trying to articulate it here because myself, I have many, many, many, many times jumped up and said, hey, security is just another form of quality code, right? But on the other hand, going back to what I was saying before, It hasn't worked the best way up to now. So perhaps finally something has to change.

4:47Matt Coles And I think what you're—

4:48Izar Tarandach I can't change.

4:50Matt Coles I think, Aizor, what you're expressing there is sort of understanding, doing the cost-benefit analysis that we all talk about that we should be, and I'll gladly put that in the swear jar charity for that one. But, you know, somebody needs to look at where where is the value, like where should we be spending our effort, right? So you're, so first off you're taking this approach that the question that was asked was should we give up on developers, and one way to look at that, the way you went with it, was, you know, the work that developers are doing to secure the systems that they're building, where's the value, right? Are they effective in their role? Are they able to build systems that have security resiliency and be resistant to attacks? Attack. And that's absolutely a valid way to look at it. I thought about the problem a little bit differently, like developers and shadow IT as an example, right? Are developers a risk to the organization? That's another aspect we could look at, but if we're looking at specifically, and where when it comes to how systems are deployed, you know, if you're building a product versus you're building an app, a web application or, you know, an enterprise service or something that you're going to expose to either your internal, external customers, your developers play a different role in those environments. So, but specifically to what you were talking about, I think, correct me if I'm wrong, you were looking at should we spend as much time doing front-end secure development and as we do for backend development, for server-side development, just in the, I guess, in that server, client-server model, are we getting the same value for the time spent? And I think the answer you've sort of touched upon is no, we're not, right? So we spend a lot of time doing frontend development. We know where we've lost the frontend. For the most part, right? The server is where we need to do all the defense.

7:00Tanya Ward Okay, can I ask, so what is then not— what's not working?

7:05Matt Coles Our current level of defenses, I think.

7:08Tanya Ward Our level—

7:08Matt Coles our capability defenses, I think, is what's not working, but also because of the business pressures— sorry to interrupt this— I think the business pressures to release means we're not putting the level of due diligence in. So this is going to open a whole can of worms of Did Agile screw us over? And should we be doing systems engineering properly? And I'm not suggesting we take us back to the '80s and '90s of application development, but we need something better, and it's not just the technology discussion again, right? Same with our last episode. It wasn't a technology— it wasn't essentially a technology discussion.

7:45Izar Tarandach No.

7:46Chris Romeo Bring back Waterfall. Bring back Waterfall. Let's do it. One release for 2 years. That's what we'll do.

7:51Matt Coles We can do it better. We can do it better.

7:52Tanya Ward We never left it.

7:54Matt Coles We can do it better by doing proper systems engineering. I'd like to— I don't know if— we're not supposed to be taking names, giving names here, but I want to plug some work that has come out of a US agency that has put a really good, very comprehensive guide together on how to do secure systems engineering, and we should be looking at following that.

8:17Izar Tarandach Yeah.

8:19Chris Romeo Why do you think—

8:20Izar Tarandach Proper systems engineering takes time. That's the problem. People don't want to spend time in stuff. They want to spend time in other stuff, shiny, interesting stuff. Our stuff is not shiny and interesting.

8:34Chris Romeo I mean, what's the root of that problem though? Is it just a resource management challenge? Like, why don't we do, as an industry, good security design, secure the things that you're talking about, Matt, that are in this document? Why aren't they done? Is it it just takes too long and we can't push to production fast enough? Like, what's the challenge?

8:55Matt Coles I think, yeah, so certainly, certainly time to production, the complexity of the systems we're working with, and the state of the tooling to be able to support the automation to make this— so automation replaces developers, right, as to some extent improves productivity and gives us better capability. In theory. And so the nature of the tools that we have— now, as we shift to new technology stacks where the tooling can go more hand in hand with it, maybe we'll get better about this. But there's so much legacy code out there, I find it hard to believe that we're ever going to catch up. But, you know, just the number of developers writing code, that code not getting the right level of review at a comprehensive level. Sure, somebody can analyze a particular file, right, that they make a modification for. But that file doesn't live on its own in all cases, right? It goes into a bigger project, which is part of a bigger project, which is part of a bigger system. And how all of those things interoperate and interconnect, it's really complex to do that in a timely fashion.

10:08Tanya Ward And I maybe kind of jump in and add, so where are we any better than we were 5 years ago academically? When, I mean, when I did my degree, which was many years ago, Security wasn't even a part of it. So I think a lot of it is how do we also flip the switch and kind of institute some of these things just in anything that, even from early on in school, like for example, when my daughter is in her computer science class at school, I don't think they talk about security. I think they talk about how to build nice things and how we think about requirements, and as we go into university, that's kind of a similar approach unless you go down a specific security track. So I think there's opportunities from an academic perspective to improve this problem as well.

10:58Matt Coles Yeah, developer awareness I think is actually— is also— so if you don't have the tools or the capability to do analysis, you need the developers to be at the forefront of that, and they don't have the education by and large, at least out of school.

11:12Izar Tarandach That's the thing, and I don't think that it's even a security education problem. And I think that Chris can add a lot here, but you know, Matt, you mentioned tools. We can give them better tools. We can give them more secure tools. Take the classical example of, okay, you got SQL injection. So we have our checklist, our blah, blah, blah, our spiel. We said, yeah, you have to do input validation. And one of the things that you have to do is work with an ORM, an ORM, right? And people say, yeah, that's going to save us. That's going to be— there is always that guy that finds out how to write a query in that ORM framework and run that query his way because they think that they are smarter than everybody else and their query is optimized. And that's where you're going to find your next SQL injection with an ORM, right?

12:06Chris Romeo I think— no, come on, that happens in the real world? I thought that was just an academic example.

12:11Izar Tarandach It's true sometimes that I am a good representative of the real world, but in my world it happened.

12:17Chris Romeo I mean, I've done it myself. I know what you're talking about. Sometimes you need optimization, you need performance, and you can't get enough. The ORM is somehow blocking your ability to execute SQL and get the performance you need to return something. So you got to write some custom SQL to do it.

12:31Izar Tarandach Allegedly. The thing is that until we teach people at the first level of awareness, Look, yes, there will be bumps in the road. And if you really want to go beyond those bumps in the road, talk to someone who knows more security than you before. Ask, ask questions. There's no problem asking questions, right? There's a whole security stack overflow of the quality unchecked. I'm sure that there is some very good advice there.

13:05Chris Romeo I'm gonna—

13:05Matt Coles can I just jump in here, Izza? There's a problem with There's a problem with your statement.

13:11Tanya Ward What?

13:11Matt Coles It's that developers— and I want to be careful, obviously we work with a lot of talented developers between all of us here, so I'm not calling, you know, not saying this is necessarily a specific problem or general problem. It's probably more general problem. Developers don't know that they need security and their management may not push them towards that. And so yes, there's great resources like Stack Overflow and lots of other options now, but how would a developer know that they need to take advantage of those things?

13:48Chris Romeo So, yeah, I mean, that's a thing that you need, that folks need to think about at the organizational level though, right? Like, this is, you know, I've been talking about security culture now it seems like for 10 years. Now it's one of those in terms, and I'm like, I was talking about that a long time ago when no one said it, and now everybody says it. I'm not saying I'm the one that came up with it. You know, it wasn't, I wasn't the beginning of it, But, you know, when you think about a couple of points that have been made, these developers, they're going to university, nobody's teaching about security. Here's a fun thing that I do all the time. Whenever you meet a kid that's in— I say kid because I'm almost 50 years old. It's a college— people are a kid to me. But somebody who's a student in university, like, and they're in CS, they say, oh, I'm in CS, you know, I love— I'm studying computers. Like, ask them how many security courses they've had. Maybe they say they had one network security course. Ask them about their Java course that they have. Did they talk about SQL injection? They're like, "No, I don't know what that is. I've never heard those terms." So we have a problem in the educational system that we don't teach people about security and then we put them into a big company and unless that company has an approach to say, "We're going to invest in teaching these things knowing that some of these people are going to learn these things and then leave and go somewhere else." You know, Richard Branson has a very good quote for this like, you know, someone says, you know, "What if you teach them and they leave?" Well, what if we don't teach them and they stay? It's the alternate approach to that, but there's some truth in that. You should invest in teaching developers about security even though they may go somewhere else and that may benefit someone else's company, but at the end of the day, that's how you improve the world is by teaching more people about security.

15:25Tanya Ward I think that's one problem. The other problem is I think the speed of technology and how fast it's moving. I mean, think about all the different coding languages that have come out in the last several years. And then when you look at those coding languages and you double-click and you figure out, okay, how do you, like, how do you ensure that there, there's like secure coding practices with these new languages? And then you think about this, the other technology stacks like Kubernetes and containers and microservices. So I think it's a complex problem actually because The learning curve is slower than the technology curve. And I think that's where we start to see gaps and problems.

16:06Izar Tarandach So we go back to the shiny object problem.

16:09Tanya Ward Last week it was pane of glass, now it's shiny objects.

16:15Izar Tarandach Imagine what happens if you put a shiny object on top of a single pane of glass. But the, the problem here that you mentioned, like, the last several years— I challenge you to, to, to, to check this the last month.

16:33Tanya Ward Yeah, even last month.

16:34Izar Tarandach I mean, we have frameworks and new languages and new stacks and new whatevers coming up at the speed of mushrooms after rain, right? And people love to jump to those things, jump to the new shiny thing. Why? I don't know, it's new, it must be good. So I think that right now all of us are tending to the side of no, we should not give up on developers, but we are questioning ourselves if we have to— how is it that we have to reach out to them, if we have to keep giving them the high-level stuff that we have been doing— sorry, the high-quality stuff that we have been doing, or do we admit that they need much more basic awareness level, uh, uh, understanding level even of security before we go there?

17:27Chris Romeo Let me flip the question on you a little bit here. So we— I asked that the initial question I asked is should. Let's change that to could. Could security give up on developers?

17:38Izar Tarandach It would—

17:39Chris Romeo what would that even look like? Is it even possible? And like, I'll just help— I'll kick some things out into the conversation here while giving y'all time to think about this and come up with much more eloquent answers than I do. But like, what does that look like? Do we build a pipeline, we put security tools in it, and we send the results just to a specialized security team and the specialized security team fixes the bugs, the security issues that were found in the code? I'm really spitballing. I'm spitballing all over the place.

18:08Matt Coles Please, I just take a crack at that.

18:10Tanya Ward That doesn't seem like a scalable solution by any means. Again, it's like— You know, do you build, you know, do you create Navy SEALs or do you want an army that's well trained?

18:20Izar Tarandach Forget scalable. Do you want developers now telling you that between them and production there stands somebody who's going to touch their amazing, impenetrable, incredible code?

18:33Matt Coles I mean, so it potentially gets even worse. Security becomes developers.

18:39Chris Romeo Ooh.

18:39Matt Coles So now we're back. It's a circular problem, right? Security has to be developers, and now they have to understand how to be developers too. And that, I guess that's a double-edged sword, right? They could be, we should already be looking at ourselves as developers to some extent, right? We have to understand how developers work. But now we're going to do it full-time if that were the case, right? And so we're back to being developers again. So I think that's part of the problem. The other part of the problem I think really key here is, and to Tony's point about not everything can be automated and scaled, we've been talking primarily about software developers here, but system designers, system developers, architects, hardware engineers, others who are involved in system development, and even some of the operations things about how you get a system to be deployed into production and functional. Requires human effort. It requires developer effort. It requires people with specialized experience and specialized knowledge, and security can't take on all those roles. If you take on all those roles, you're rebuilding your organization into your security team.

19:49Tanya Ward Yeah, Matt, you made me think about something when you— because as you were talking, I was thinking of two points. The first one is we keep talking about the word security, and it's like, it's kind of like this scary, scary word. You know, and so I think we need to start humanizing that word because if we start humanizing it, maybe developers won't give up on us. I think the second point is you were talking about roles and I think it's a really good point because there are so many different roles involved in building software. As you've talked about, you have architects, you have systems engineers, you have developers, you have even testers fitting into that equation who may write automation. And how do we help these individuals' role understand the parts that security should play in their responsibility? Because, you know, sometimes when we think about champion programs, it's a very— this is what just developers should be doing. We're not looking at all of these different roles and how they play together to build something that's secure.

20:58Izar Tarandach The thing is that I, I joked about the gate between the amazing code and production, but the truth is security is never going to know these systems as well as the people who are—

21:08Chris Romeo who write them, right?

21:10Izar Tarandach And in all the examples that Matthew, uh, listed, I think that's what jumped to me was that there is this amazing— forget volume, but variety of specialized knowledge out there that security is never going to get to, right? I mean, if somebody's designing hardware and they have to think about side channels now, dude, you can ask me until tomorrow, I'll be Googling as fast as you because I have absolutely no clue, right? And I consider myself a somewhat experienced security person. So for security to try and, and get a central place in the stage and say, I am the big gate and everything goes through me right now. I think that not only is counterproductive, but it's a bit, uh, almost obnoxious. It's a bit, uh, uh, uh, you'd never ship anything, would you?

22:06Matt Coles Well, you'd be stuck in analysis paralysis for the most part too.

22:11Izar Tarandach Tanya, to go back to what you said, it's putting ourselves in place of other things that are as important as You said quality. I would say reliability. I would say performance.

22:19Tanya Ward Yeah.

22:21Izar Tarandach That's as important and sometimes much more important than security. So put yourselves in there and say, no, we're more important than performance.

22:30Tanya Ward Come on. Yeah.

22:32Chris Romeo So, I mean, hold on. Let's poke that one a little bit. So, yeah, let's poke the bear.

22:38Tanya Ward Why not?

22:39Chris Romeo So you just, you basically put performance, resiliency, quality, and security all across the board at the same level. And so, so security is at the same level as each of those, or does security sit at a higher priority than those other three? Like, can you have a secure system that doesn't have a high level of quality? Can you have a secure system that doesn't have a high level of resiliency? Because You know, can you— if security breaks down, we probably lost all the data, all of our customers' data, and there's a good chance we might be going out of business. If quality breaks down, do we have that same level? So can security sit at a higher level here?

23:23Izar Tarandach Well, we love to put ourselves into this higher pedestal and say that security is the important thing and you guys better pay attention to what we're saying because otherwise Tomorrow you're going to be in the news and you don't want that kind of news. You want the other—

23:37Tanya Ward fear, uncertainty, and doubt.

23:38Izar Tarandach But, but the fact is that when Tanya mentioned quality, we all agreed that security is a facet of quality. On the other hand, if you have code that's non-performing, it could be as secure as anything, but if it's not actually doing what it needs at the time that it needs to do then is it actually doing something? And if it's very secure and it's not doing something, then why? It's like having a Porsche in your garage and not going out, right?

24:11Matt Coles Now, having said that, there are dependencies here, right? And let's add privacy to this mix because we haven't yet, right? You need security in order to support privacy goals, right? You can't have— I will go out on a limb and say you can't have privacy if you don't have security, but the true reverse is not— the reverse is not true, right? You can potentially have security without privacy, but you can't have privacy without security. And so where else are those relationships? Performance. Let's talk about performance for a second, right? Performance code— security may influence performance.

24:53Tanya Ward It does.

24:53Izar Tarandach It usually does.

24:54Matt Coles And usually does, right? So those, those are potentially mutually exclusive. Quality, I think that's the underpinning. If you don't have quality, you don't have anything else, right? So I think we've established that. And we have other illities here, right? Maintainability, supportability, compliant— well, compliance.

25:15Chris Romeo Yeah, that's another, that's another word on the list.

25:17Tanya Ward Compliance.

25:20Matt Coles What's the illity there? Compliability? Compliance!

25:26Izar Tarandach Security!

25:29Matt Coles But actually that brings up another interesting discussion point perhaps. We've been sort of focusing on purely defense like— and prevention, and that's been the core, right? If you're worried about remote code injection, if you worry about malware or your writebook from a code injection, you're going to try and prevent all buffer overflows. And we know we can't stop all of those from existing. So then is defense and preventative defense the primary focus, or should it be the primary focus, or should we start or be more looking at more like more commonly now either a blend or focusing more on detection, I'll say detection, monitoring, management, and less so on, you know, and it really pains me specifically to say, you know, to make resiliency and proactive defense to be not the primary focus, but we're losing the vulnerability war here, and so maybe it's time to shift gears.

26:44Chris Romeo I thought you were going to say shift left for a second.

26:47Tanya Ward I was like, oh no. As you were kind of talking through that, I think, you know, we again going back to the word security, I think there's so many different pieces to that word. And so there's like, how do we build secure products through the practices and secure coding with our developers? I also think there's how do we build products that can kind of can have a level of resiliency to do detection if they notice some abnormalities in it, like if performance started going out of, you know, off, off spike, they can actually start flagging things within it. And as you said, detection. And I, I think they're too, you know, when we think about developers, I don't really see developers fitting in that side of the house. I really feel like they sit more in, when we think about building secure products, they fit more in that kind of secure coding, and these are just other things to help customers understand, you know, how to kind of put the eyeballs out there if something is potentially hitting their system.

27:52Matt Coles Yeah, actually, and can I just— I just want to jump in there for just a sec. I have not seen it, and I would love to be corrected here if any of you have a good resource or want to talk about this, but the idea of self-healing, self-defending and self-healing systems. Have we really gotten there yet? Right? Do we have, we don't have, we don't have those features. It's, it's, it's like AI, right? It's not fully here yet and we've been waiting for it for a long time.

28:21Chris Romeo I feel like that was a campaign though, a number of years ago. For real. It was a, it was a marketing campaign of self-defending, self-healing, and then it just like lots of marketing campaigns, it just kind of dissipated and people stopped talking about it. I don't know that we ever got anywhere close to self-defending. Like that was even before AI was a big thing. So I don't think they were even talking about AI doing some defense action. I think it was just that, that was just really a lot of fluff is what I, I only saw.

28:51Izar Tarandach Let William Gibson write your marketing material. But, uh, the, the, the thing is you guys, it's so, so many in there.

29:01Tanya Ward It's so priceless.

29:03Izar Tarandach It's just so many interesting points in there that I, I wish I, I still had the head. It's, it's 5-something to go back and go point by point. But one thing I can tell you, Matt's touched into a facet of things that is fast becoming my, my new, uh, best subject, and that is the, the, the fact that, uh Monitoring and detection and, and all that good stuff is becoming more central to our efforts. And that to me joins the developer side of things when we think in terms of there's this huge arsenal of tools out there that has been used for observability that we haven't quite started using for security. There are some efforts there, but there's a huge amount of data in there that the developer intrinsically knows and understands because they're producing it, right? The signals, the traces, the logs, all that kind of good stuff. And we have not yet gotten to the depth that we can at the developer level to provide them with that security. It's funny to me because like years and years and years and years ago, sitting in front of a Sun 3 workstation, and there's this new binary that you didn't know what it was, you would run truss. Later on, on Linux, you would run strace, right? And those are tools that par excellence, they are debugging tools that developers would use in those times and even today to do observability on those binaries, understand what system calls are happening here, what's the arguments, where is stuff going, right? And then to Matt's point of teaching more less defense and a bit more of the taste of attack. A couple of years ago, I was teaching someone about buffer overflows. And you go into a Linux virtual machine and you start showing them stuff, but then you start telling them, okay, now you have to turn off this thing and you have to turn off ASLR and you have to compile it without this flag and that flag. And the person was just looking at me and saying, wait, if I have to turn all this stuff just so that the buffer overflow actually works, then is that a problem actually? I mean, have we gotten to a point where some of our defenses are good enough so that we can finally say we have removed a whole class of bugs from the table? And that jumps into the discussion that we still want to have about Rust.

31:37Chris Romeo Yeah, I don't think, I mean, I don't think we're there where we can say buffer overflows have been completely mitigated at this stage, but I wanna come back to something else you were talking about from observability because I wanna bring RASP into the conversation here. Runtime application self-protection. Like when I think about what's changed the game in AppSec in the last couple of years, I think RASP is on, is on that list of, you know, we could probably talk for hours about how threat modeling's on that list as well. But I think when I think about what RASP provides for us of being able to have that observability and make that security decision inside the runtime environment for an application, Listen, I don't work for a RASP company. I'm not trying to tell people they need to buy it. I think that's been a big movement forward though of once again enabling that observability, detecting things like SQL injection where you have almost 100— you could say it's 100% hit rate if you're doing it inside the runtime app of the application. There is no false positive there because the SQL is there. You've got that observability there. And so I think RASP is something that has been really industry shaping for AppSec and the places that all of us like to play. And I think RASP answers a lot of those things, Isara, you were talking about there as far as observability and getting developers the information they need about the thing they're an expert in as far as the execution of the application.

33:06Izar Tarandach And yet it's not everywhere, why?

33:10Chris Romeo Education.

33:11Matt Coles Well, I think it's a system design question too, right? You have to design for it. And oh, by the way, here's security again, 'cause now you're generating data. So it's a circular, it becomes a circular problem again, right?

33:24Izar Tarandach There's another thing there that you're not, you are adding a layer to the system that's not part of the system or that's not been designed on the system. Or that it's code that somebody else wrote, that it's not one of the libraries that you wisely chose to the system. So you are impacting that other thing's performance. And as we know—

33:46Tanya Ward I was just going to add, I think that we're being— that's going back to— I think that's idealistic. I think let's kind of put the lay of the land out there of what a developer's life is. I hate to bring up the word, so I'm going to just say I'm not going to bring it up, but you're on sprints. You have velocity curves. You're all like user points, stories. You're trying to get your basic coding stuff done. Then you're trying to make features. And then on top of that, you have people knocking on the door saying your code isn't optimized because it's not performing as fast. And then you have somebody knocking. Your testers are throwing 50 million bugs at you. Then you have security coming knocking at your door saying, well, you need to do this. You need to do that. So I think RASP is a good concept, but I think The foundations really need to just— we need to get better at the basics before we go to the advanced. Unless you believe the advanced can outweigh the basics, then you don't really need to do the basics because the advanced will end up taking care of just everything.

34:48Chris Romeo One of the dangers of RASP— you just unlocked one of the dangers though that I see in RASP, and that is people become complacent. Oh, the RASP will take care of it. The RASP— I don't have to worry about SQL injection. Even if it gets in, the RASP agent will knock it out.

35:02Izar Tarandach Same way as they did with firewalls, same way as they did with WAFs. Remember when WAFs were going to put all of us out of a job?

35:13Chris Romeo Somehow it didn't happen.

35:14Matt Coles I don't know.

35:14Chris Romeo I don't know what happened.

35:16Matt Coles Well, you know what's next, but I'm not going to say it.

35:19Chris Romeo Please say it, Matt. Say it.

35:21Matt Coles No, no, I don't want to spend more money on the swear jar. I've racked up quite a bill already.

35:30Chris Romeo The old zero trust.

35:33Matt Coles But actually, no, it actually sort of, it does bring up an interesting aspect of maybe we're looking at the problem wrong. And maybe, so maybe back to one of the earlier statements that Chris brought up of could we take developers out of the picture? Maybe we can in a limited fashion. So today, if, and if we look at, so first off, I think DevOps or DevSecOps is not a well-baked idea even still today, personally.

36:11Chris Romeo Wow, controversy, wow.

36:14Matt Coles I agree with him. You're giving a lot of power to, you're giving a lot of power and responsibility to developers with not necessarily all the capabilities or tools or or knowledge that they need. And so there's a, there's a positive and there's a negative to that. But if we think about this as sort of an evolution, maybe what we need to look towards is, let's take a Kubernetes environment as an example. If developers weren't responsible for the Kubernetes environment and you instrumented it to run containers with RASP and logs and all the infrastructure and all the monitoring and all the things that we as security need, and it's, I guess, if it's transparent to the user, to the developer, and they just drop in workloads into that environment, they get, we get security, they get productivity, and we're all happy-ish, right? So Maybe we need to look at this differently. We in security need to influence the technology decisions so that the infrastructure can support our needs and developers can work within a shell that they can then feel comfortable with so that they can be productive and do what they need to do effectively.

37:36Chris Romeo Wow, Matt, coming to, coming around to kind of where I started this conversation about, you know, could we do this without them? So Tanya, Izar, you got any responses, rebuttals to that idea of a zero-dev security? Zero-dev security, is that we're calling it now?

37:53Tanya Ward I've got some thoughts, but it's all right, I'll let you go first.

37:57Izar Tarandach Thank you, Tanya. Thank you. I, I, as I said, when, when Matt just put it down, I, I agree with him. There's a lot of stuff that's half-baked, and of course it goes back to the word of the day, shiny stuff. But on the other hand, I think that the whole proposal of the— well, part of the whole proposal behind the, the, the whole thing was that, hey, security, you guys want something to be all around the infrastructure, that's fine. So just write this whatever it's called, and it's going to make sure that everything gets included into everything else, and all the instrumentation is going to be there, and all the libraries are going to be in place, and and everything's gonna be fine. And I think that we have gotten to a place where we have a lot of that. Unfortunately, we have so much things running as code today that we haven't, I haven't found yet a quick way to do code as code. Every time that I try to write something, I spend so much more time building the whatever is, needed around it as code, right? And I think the developers are suffering from that too. I mean, that we were so afraid of giving them a single point of failure that we ended up giving them millions. And all this infrastructure ends up being put together with a lot of duct tape and hope. And sometimes a trust.

39:30Tanya Ward I know, I agree with both of what you've said. I would also add, I mean, not everyone, everyone has different development environments. Everyone have you know, how they build their products or applications are different. So not everyone always use containers, Matt. I always go back to, you know, we're thinking about tools. How can we make these tools smarter to better help our developers? Like, how do we reduce false positives? How do we get them really focused on what they really need to be focused on as part of that pipeline? And I'd be interested to kind of hear all of your thoughts on, do you think that the industry is better serving developers or not serving developers?

40:15Matt Coles So before we answer that, I just want to clarify. So I was using containers as an example, as an area where we could potentially do a good work of building an environment that developers could make use of that have the security properties that we need. It is not universal. It will not work for everything. Right, hardware interfaces, it's, I can't imagine unless we're also building hardware modules that that would be feasible. So it's very focused, very specialized I think, but that may solve, you know, if that solves 80% of all the data breaches of, you know, high-performance web environments, great, that's a plus, that's So I just wanted to throw that out there.

41:00Chris Romeo Seems like that should be possible now too. Like for, in special cases, like to your, the caveat you made, like there's, if you're doing something different than normal web application, then all bets are off. Like we can't, we can't make these statements about it. But I mean, I think there's something there to having, building an environment and it's a pretty tight set of guardrails. That are constraining what the developer can put into this and the controls that are gonna exist are prescribed and are part of the assurance that we gain from the entire system. But yeah, I mean, I think there's a lot of one-off cases that'll break that model as well.

41:41Izar Tarandach But the moment we start laying down those things, you always get somebody coming and saying, oh, you are getting a talented painter and telling them that they can only use two colors.

41:50Matt Coles Right?

41:50Izar Tarandach We are putting those guardrails and people are feeling constrained by them.

41:55Matt Coles That's a whole other piece. I would look at it as— what we're really saying is we're hiring a painter and we're putting the scaffolding up and attaching it to the frame of what they're going to be painting on, and they can then choose their tools from there.

42:12Tanya Ward Right?

42:12Matt Coles So again, you're looking at it— we're looking at it with the wrong constraints.

42:15Tanya Ward It seems to me what you're saying though is that if we're We're trying to reduce some level of risk, but in that, it also seems that we're not really trusting our developers. We're painting a beautiful little play box with a fence around and saying, you have to stay here, you can't go over the fence, you can't go through the fence. You know, but at the same time, in doing that, we're also protecting, you know, we have a level of protection in what we're building, like to prevent attackers from getting in. And which one rules? Like, is it—

42:47Izar Tarandach You know, we started saying, can we trust them in, in different words from where we, we started, right? And I think that the place that I'm driving at is, and this is going to, to, to sound horrible because I do think that programming is an art, right? At some point we have to tell them you're not a painter. You are a Lego builder and, and here are the Lego pieces and go build this thing that I want with these Lego pieces, right?

43:21Chris Romeo I mean, think about the modern application though. Like, the modern application is like— yeah, exactly, at least for modern web applications.

43:28Tanya Ward But that's no different. But where is it?

43:32Izar Tarandach Where is the framework when people start painting off the, of the colored boxes, right?

43:37Tanya Ward But where does that end? Is our— that's True for designing, it's true for UX interface, it's true for every aspect, right, of, of building a product.

43:53Chris Romeo So we're advocating for low, no-code everywhere. Just use it everywhere. Like we don't need languages anymore.

44:00Izar Tarandach It, it's, it's interesting that you pull that to designing cuz many times I have seen reactions from teams while threat modeling where people would say, why do you have to do this so confusing? Why do you have to do this so complex? There's a much simpler way of doing it. And usually, and I think that it's safe to say, usually the simpler way, if it's not more secure by itself, it's easier to spot the non-secure places because you are not churning into that complexity. Right. So I think that we are tending to it as, as Chris said, we, we are tending towards everybody agreeing that yes, we have, we, we are building Legos. Of course we are building Legos with a lot of creativity. We are build, building new Lego pieces every day, but we should not be painting the, the ceiling of the, the Sistine Chapel. Every time that you want to create a single-page application. And that's where security meets the road, I think, meets the developer in saying, hey, listen, you have to be aware of these things. You have to understand there is a minimal set of things that are required for this thing to be able to, to function in the real world. If it touches your performance, if it touches the limits of what you need, what you need to do, then you're painting out of those guardrails that we told you. So come and discuss it with us so that we can help you.

45:38Matt Coles And by the way, it's not a, it's not a trust issue. Well, it's not specifically a trust issue, or it should not be framed as such. What security brings to some extent is a different view on requirements. And so when we build that sandbox. If we were to build the sandbox, let's take the panner example, right? We'll build the scaffolding and the bounds, and here's the, here's the, your canvas, and we've secured your canvas because we know that OSHA here in the US, for instance, is going to, you know, sue us out of existence if, if you get hurt. And so we're here to partner with you so that we all meet our goals and we all meet our objectives and we don't get sued out of existence. Now, obviously, very extreme example here, but And if we have that conversation as security professionals that we're here to help make the project a success and not we're here to stop you from doing something stupid because we don't trust you not to, then— and obviously this is easier said than done, right, because it comes down to how do you eventually get— how do you make it to that level of conversation so that they really believe you when you say that, that you're here to share your experience.

46:57Izar Tarandach Stop you from doing something that we think is stupid.

47:01Tanya Ward I think ultimately there has to be a level of collaboration between the two. So we, we need to be better informed how we better help them, and likewise they need to understand the requirements that we're putting forth, right? Because industry, technology, everything keeps changing. So I think, Chris, at least in my opinion, to go back to your initial question, I don't think you, you can give up on developers, right? When it comes to security, it's It's always got to be a dance and a strong partnership.

47:32Izar Tarandach Yeah.

47:32Chris Romeo And I think we're all gonna, yep, we're all gonna, we're all gonna agree with that. I think at the end of the day, none of us are gonna be like, no, we should just give up on, unless they have AI, give up on development. Yeah. But it's a fun thought exercise.

47:43Matt Coles Yeah.

47:44Chris Romeo If we had, you know, if we had ChatGPT, OpenAI doing it for us, but like, it's a good thought exercise though, cuz it does, it does cause a, cause us to consider the boundaries and, you know, what are we asking of people and, you know, a lot of different things. So, it's good to go through that exercise. I think, Tony, I think you summed it up for us pretty well there as kind of a closing statement or a, you know, a takeaway here that, you know, we do all agree that security and development, it's about that partnership, it's about that collaborative. That's the only way we're going to get where we want to go. To Isar's very early point, we're never going to scale enough. You're never going to have a one-to-one security to development. It's just not possible. And so by security and development partnering together, that's really the path forward for the future. So thanks to everybody for listening to this episode of The Security Table, and stay tuned for the next episode where we're going to talk maybe SBOM. I also got a couple other ideas here, DevOps, DevSecOps. We're going to argue that one out on a separate conversation. We're going to save that for a whole other day. So thanks, folks.

Transcript supplied by AssemblyAI. It may contain transcription errors.

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