Secure by Default in the Developer Toolset and DevEx
With Chris Romeo, Izar Tarandach, and Matt Coles
Matt, Chris, and Izar talk about ensuring security within the developer toolset and the developer experience (DevEx). Prompted by a recent LinkedIn post by Matt Johansen, they explore the concept of "secure by default" tools.
About this episode
Matt, Chris, and Izar talk about ensuring security within the developer toolset and the developer experience (DevEx). Prompted by a recent LinkedIn post by Matt Johansen, they explore the concept of "secure by default" tools. The conversation highlights the importance of not solely relying on tools but also considering the developer experience, suggesting that even with secure tools, the ultimate responsibility for security lies with the developers and the organization.
The trio also discusses the role of DevEx champions in advocating for security within development processes, emphasizing the need for a balance between security and usability to prevent developers from seeking workarounds. They touch upon integrating security into the developer workflow, known as "shifting left," and the potential downsides of overburdening developers with security responsibilities.
There's a recurring theme of the complexity and challenges in achieving a "secure by default" stance, acknowledging the difficulty in defining and implementing this concept. The conversation concludes with an acknowledgment that while progress is being made in understanding and implementing security within DevEx, there's still a long way to go, and the need for further clarification and discussion on these topics is evident.
Matt Johansen's Original Post:
https://www.linkedin.com/posts/matthewjohansen_i-really-feel-like-a-lot-of-security-problems-activity-7170811256856141825-lKyx
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 · 44 min conversation
0:09Chris Romeo All right. That's yes, that's a terrible visual. I'll just read it. So, hey folks, welcome to another episode of Behind the Scenes at the Security Table, which is now transitioning into the Security Table. Uh, my name is Chris Romeo, joined by Izar Tarandach, Matt Coles, uh, the Sultans of Security. I don't know. I'm just, I'm always, I'm always branding people. You guys know this.
0:30Matt Coles Where's the, where's the, where's the Mark? Effler rips riff, uh,
0:34Chris Romeo Sorry, we didn't have the budget for that, but uh, maybe if we get a sponsor someday we can have like
0:39Izar Tarandach have a triangle here somewhere.
0:41Chris Romeo Yeah, or maybe Brooke, maybe Brooke Schoenfeld will create us a, uh, on his guitar, he'll create
0:46Izar Tarandach Oh, that would be awesome.
0:48Matt Coles or, or our number one fan. Use some AI to, to do it for us.
0:51Chris Romeo That's true. Okay. So the tasking has been sent out to, uh, everyone to find us a little riff that we can use here. But let me read the, the LinkedIn post. It's going to be the, it's going to kind of set the stage for us and lead our discussion today.
1:06Izar Tarandach We call that
1:06Chris Romeo Johansson posted this on LinkedIn. And what this Post says is the following, and I quote, I really feel like a lot of security problems are actually developer experience problems. The more tools I put in my engineers hands to do their job that are just secure by default, the less they figure out workarounds to introduce risk. I'm more successful the more I get out of their way and figure out ways to enable them to do what they need to do, but with guardrails. If you aren't already, find a champion with your DevEx org. It'll pay dividends. So we're talking about developer experience. We're talking about secure by default. We're talking about guardrails. Uh, so I want to give you guys the chance
1:49Chris Romeo to set the stage or give us the first opinion here to, to get the conversation rolling. Matt, we always go to you. So we're going to do it again.
1:57Matt Coles Well, so actually I want, I, I want you, if you could to, there's a few points. So there's a few points. I mean, luckily, uh, Matt, uh, Johansson here, uh, constructed this, this. This, um, this post as individual sentences. We can pick it apart one by one, pick it apart. So I didn't mean to say that, uh, I meant to say, uh, analyze each, each point individually. But where does it come from? And I want to touch upon the secure by default statement first, because that's probably the more contentious out of all of them. So when I read that, the tool, more tools I put in my engineer's hands to do their job that are just secure by default. the less likely they figure out workarounds to reduce risk.
2:46Matt Coles So, when I first read that, I was thinking, is the tool secure by default? Or is the tool producing guidance or results that are secure by default? So which do you think it is?
3:01Chris Romeo That's a good question. I mean, you can, you can, You can see, you could almost say it's both if we, if we, and we're definitely not a textual analysis podcast here, but we, we analyze the, well, there was the word the, at the beginning of the sentence. Did you not see that? That changes the entire meaning, Matt.
3:20Matt Coles so I guess maybe the other piece here is, and it's another matter of interpretation, Izar.
3:27Izar Tarandach But no, point of contention, we can't. Because the first sentence actually says that the problems that he's trying to solve are actually developer experience problems. The result of the tool is not a developer experience problem per se.
3:45Matt Coles it's the use of the tool?
3:46Chris Romeo It's the usage of the tool. Yeah.
3:48Matt Coles Well, so, and then I was thinking that there's maybe another interpretation of tool here. So, and again, so we need to love, we need to bound our conversation because otherwise we're going to go off in tangents and it's going to get ugly. It may get ugly anyway. So, so do you think, because we don't, we don't know what Matt specifically was at, was going for here, but, um, is he talking about like SAST or is he talking about, uh, production frameworks? So, for instance, are we talking about tools used during the development process? Or are we talking about the tools that developers deploy to, which I think we would commonly call infrastructure or frameworks, right? Is that what he's referring
4:33Matt Coles to? Because that,
4:34Chris Romeo say yes.
4:35Matt Coles a very big difference there. D all of the above. Okay.
4:39Chris Romeo So what is developer experience in the modern world of technology? Developer experience tends to be a team that's responsible for the tooling that a group of developers slash engineers use to build something,
4:54Matt Coles We used to just call them tools teams.
4:56Chris Romeo Tools to yeah, I mean we used to call them tools teams back in the day, but they've gotten a fancier name now That's fine. Everybody has fancier names for things. But yes classically In the early 2000s even 2010s. We referred to them as the tools team,
5:09Matt Coles Or release engineering, or something like
5:11Chris Romeo released. Yeah, I mean for me release engineering was more of a team that was focused on getting builds out the door They manage the pipelines, but developer experience was the tools team. They up, they uploaded, or they, they upgraded pieces of software that developers use to build, uh, the things that they build. And so that's what I think we're talking about here though, is tools that the developers use to build, something. And so when I think of this, I'm thinking with the IDE is in this category. I think all of the AppSec tools that people are, that we're going to, we've talked about ad nauseum on this podcast, would fit into that category. Uh, CICD build related tools fit into that pipeline,
5:52Chris Romeo right? All of these things are part of the developer experience. Anything the developer uses to build you stuff is part of the developer experience.
6:01Izar Tarandach and that includes frameworks and SDKs and all that code
6:05Chris Romeo Yeah. Libraries, open source software, things like that.
6:09Matt Coles So I rate this a parti I rate this a partially true.
6:13Chris Romeo Okay,
6:13Matt Coles So, I would contend, I will contend, that I could give every SAST, STARAST, tool, whatever, uh, SCA tool that is itself secure by default and have a piece of garbage outcome of a product.
6:37Izar Tarandach Yes, but those are tools serving the dev, those are not dev tools. Right? Okay, let's
6:46Chris Romeo they should be dev tools, but I would say classically and even in the modern age right now, I don't consider a SAST tool to be a developer tool. It is a security tool that we make developers run, sometimes against their will.
7:00Izar Tarandach a bit further into what he's saying, okay? It says, the more tools I put in my engineers hands to do their job that are just secure by default, the less they figure out workarounds to introduce risk. Let's look at SAST.
7:13Chris Romeo Well, SAST wouldn't really fit in that category though.
7:15Izar Tarandach right,
7:16Matt Coles oh, so now I'm, I see now I would think that a SAST tool is something you give to developers, meaning they run SAST, they go into
7:23Chris Romeo To do their job, developers don't need SAST to do their job.
7:27Izar Tarandach no. And they are the first ones to say that.
7:31Matt Coles hold on, if you do SAST, if developers are going to write secure code, right, you're going to run it
7:38Izar Tarandach No, that's, that's, that's them doing our job.
7:41Chris Romeo Of course that's good process, but that's not, so what are the, so if I, if, okay, I take a developer first day on the job, I sit them down. I say, here's your laptop. You only have a SAS tool. They can do nothing. They can do, they cannot be productive. They cannot, they cannot do anything in their job. They will quit two days later because they're like, all you gave me was this SAS tool and you locked on my laptop. So I couldn't even load an IDE or any of the things I needed to build software. So that's why
8:07Matt Coles a developer tool?
8:09Izar Tarandach No, but, but that's not where I see the problem. It's not the fact that it's the only tool that I have and that proves that it's, it's not a developer tool. To me, it's not a developer tool because a developer can achieve their objectives with or without the SAST. With the SAST, they are going to achieve better objectives from our security point of view, right?
8:30Chris Romeo a supporting
8:31Izar Tarandach to me makes it a security supporting tool.
8:34Chris Romeo it's a supporting tool or technology,
8:36Matt Coles Okay, so,
8:37Chris Romeo to the
8:38Izar Tarandach Not only that, not only that, but in that class of tools, there isn't anything that a developer can do to end with a less, no, there is, they can change config and, and stuff like that, but that's not going to lead them, apart from perhaps creating less tickets or whatever, or alerts or whatever, it's not lead them to, to, to, to their outcome, which is at the end of the day, better code. Perhaps if you have a tool that's very aggressive in the CICD and it kills the build and they can shut that tool down, then okay, that lets them achieve their objective of putting code out in a less secure manner because the tool had bad defaults. But what I read in this, in this line here, is exactly what we
9:24Izar Tarandach spoke about, I think, last week and the week before and the week before that. Give them a library, give them an ORM. If they think that they need to, they will find ways around that goodness of the ORM to write their raw queries and introduce injection. So that's what I think is going on. If we start giving them the things that we're talking about with the guardrails and libraries that work by default in a secure manner, it's If those things are too in their faces, if those things are too restrictive, they are going to be an incentive for them to find smart ways around the controls that we put in place. So the whole thing defeats the objective. When you, as Matt says, find a champion with the
10:16Izar Tarandach DevEx org, you are bringing in somebody who can be the voice of balance in the world. You want this to be secure, but the developers want it to be flexible. Here's the middle term where everybody can live. Happily.
10:30Chris Romeo think you got to take a, you have to declare a moment in time where you're making this shift from developer tools that are giving developers all the freedom to developer tools that are have, that are secure by default.
10:43Izar Tarandach Yes. Agreed.
10:44Chris Romeo I don't think we are there yet. I don't think, I think Matt is describing a future state. that we should aim for that he says he's trying. He, you know, the more tools I put in my engineer's hands, he's, he's iterating on his, on his process of how he's working with developers. I don't think we are there yet, but I think there is, there are things that we can do, for example, with the IDE to make the decisions that the developers are making influenced or constrained by the guardrails, but also directed into a secure by default stance.
11:17Izar Tarandach So let's take the id one popular example today is a visual code. Almost everybody's using it one way or another. Extremely powerful thing, like almost as powerful as emacs, right? But the thing is that you, you, you bring it up, the first thing that you get is there in the corner restricted mode enabled or restricted mode flashing or whatever. And what that does, that means is that it's not going to interpret a lot of code, both local and external, that you could be importing as modules, stuff like that. The second thing that happens is the developer turns that off. It says, trust, trust, trust the code in here because it's me writing, I'm great, trust it, or whatever. right? Now,
12:02Chris Romeo so well throughout our history.
12:04Izar Tarandach true, but everybody feels good because there was a security by default thing there in place
12:11Chris Romeo Well, you
12:12Izar Tarandach and the developer, wait, wait, wait, wait, I'm not in the whole scenario, so they turned it off, cool, everybody continues doing their thing, write their code, whatever. Now they go and they find a plugin, a plugin that they need to bring in. that plugin could come from a central repository with Microsoft, it could be some repository on GitHub, whatever. It comes in, they start using it. Nobody knows what that thing actually does, right? So my point in here is, you have this tool, developers need it, it's the experience that they choose, they build the experience by importing modules and functionality, and security has absolutely no control or nice way of control or visibility into what these
13:01Izar Tarandach guys are bringing. So, the approach up to now has been that security takes a step back and says we are going to create a local repository with things that we've vetted beforehand and we know where they come and we expect everybody to only pull modules and functionality from that repository that's vetted. Would we call this secure by default?
13:28Matt Coles Could be, and, and to follow up on that, so thank you for realigning my expectation here, uh, but now I want to ask a clarifying question along that, along that lines. When we talk about supervised fault, And we talk about risk. Risk to who or what? Are we protecting the end system and the customers? Or are we protecting the business that's creating it? Because what you're describing, using vetted, validated components within an IDE, you know, as a developer plugin, is a business risk problem. Not a customer solution
14:05Chris Romeo Yeah, but isn't, I mean, at the end of the day, this is a great question, Matt, because I've been thinking about this over the last couple of weeks. At the end of the day, for me, every issue is a customer issue.
14:18Izar Tarandach If you're a SAST,
14:19Chris Romeo my build infrastructure gets broken, ultimately the biggest thing that scares me is that my customer's data gets disclosed. I don't care how it happened, so everything for me funnels towards customer data disclosure.
14:31Izar Tarandach Or if it interrupts your service.
14:34Chris Romeo or denial of service or whatever, right? So, that's why I don't separate business,
14:39Matt Coles well, it makes sense if you're in there so we've we talked about this a couple weeks ago with with product versus, you know, service provider, right? And whether that's business service provider or cloud application service provider right where your infrastructure is is where your customer data lives versus and your service resides and your business model is based on as opposed to like a like product development where if the CICD pipeline goes down and you can't ship a product you know on day one well if it comes up on day two oh okay so you ship you delay a day it may be may be bad but it may not be catastrophic to your end customers right um so the risk level is different but we're still talking
15:25Matt Coles about business risk, and I agree there's a transitive effect there to customers when you're talking about your business is up as a service provider.
15:34Chris Romeo Yeah, because without customers we don't have a business.
15:36Matt Coles that's right,
15:36Chris Romeo we can, we can measure all the business risk we want. If the customers are gone, we're just, we're just tootling our thumbs,
15:42Matt Coles right. So secure by default, I will, I will argue that secure by default is not the only metric here, and unless we, unless we further define what secure by default is, right, so secure by default is securely configured, right, and in theory patched for vulnerabilities. Right, for known vulnerabilities.
16:03Izar Tarandach Mmm, I'm not sure if I'm there with you.
16:06Matt Coles Okay, I did that, I did that on purpose. I did that on purpose.
16:10Chris Romeo have to do with, I mean, you got the intended result. What does patching have to do with, uh, with secure by default?
16:17Matt Coles Okay, so secure by default is at least secure configuration, right? And you do secure configuration because you want to prevent somebody from Uh, having an exposure of something that can be bad, right? So you don't, you don't configure HTTP because you don't want data in transit to, you want data in transit to be secure, so you use HTTPS only. You authenticate with strong credentials because you don't want to have a bypass of that authentication. You have strong RBAC in place. right, as a configuration option because you want strong access controls. So if you don't patch as part of that secure by default process, you're leaving exposures, potentially, especially if those are attack surface type,
17:01Matt Coles uh, vulnerabilities.
17:02Chris Romeo I think we've got a break secure by default into. I want, I want to segment it. So I think secure by default, I think design decisions. We have a number of loaded design decisions up front that if we do not do or do not enact, we'll result in a crappy product whenever we get to production. Now there are supporting operational processes that'll have to, that are part of calling something secure by default. You can't tell me it's secure by default if you don't have a patching process. Oh, well, we can't, we don't, there's no upgrade. Well, then you're not, you're not secure by default. If there's no upgrade path, if there's no, if you're not triaging vulnerabilities to that product and releasing
17:43Chris Romeo new versions, hopefully in an, in an automated way that I don't even have to think about or worry about. Like the coolest thing I've seen in the last number of years is like software patches from products I've purchased that just happen. Just magically. And then sometimes they tell me, sometimes they don't, right? So, but big picture for me, we have secure by default design decisions. We have operational things that are supporting, that have to be there for something to be secure by default, but they're not secure by default. You can't say, well, because I patch, I'm secure by default. That's, that, that doesn't, doesn't add up.
18:18Matt Coles Sure, let me, actually I wanna, I wanna touch upon something since you just brought it up. Uh, if you have an automated patching process in your tool that you've given to developers as a developer tool, and You can push patches that are, that potentially have functionality side effects. Have you broken developer experience?
18:44Chris Romeo What do you mean by patching what? Patching, auto patching like the end product that we send to customers, or auto patching my IDE, or what? I don't,
18:52Matt Coles auto patching, auto patching the IDE.
18:54Chris Romeo Auto patching the IDE.
18:55Matt Coles So we know that secure, we know that patches that come out for system, for things can include functional side, functional side effects. Deprecation of APIs, deprecation of functions, removal of functions, change of behavior of functions.
19:08Chris Romeo But I mean, that's why you have a developer experience team though, that they don't just, they don't just let visual studio code updates flow to everybody whenever Microsoft posts them.
19:18Matt Coles So just having a secure by def,
19:20Chris Romeo has a triage process to ensure that they aren't breaking the whole world.
19:24Matt Coles so then it's just having a secure by default tool is not sufficient. Mm
19:28Izar Tarandach no, Matt, wait,
19:29Chris Romeo I never said it was the only thing. I never said it had to be the only thing. I'm not saying you don't have other things that support it. It's not standalone by any stretch. Sorry, Izar, go ahead.
19:38Izar Tarandach No, wait, wait, developer, developer, developer experience. I, I need to, to understand this better. To me, if you have a Porsche or if you have a bicycle, both are able to take you from point A to point B. You're probably going to enjoy the ride much more in the Porsche. Developer experience, same thing for me. Visual code or VI, both are going to enable you to write code. You're probably going to enjoy the experience much more in visual. Visual Code, because it's going to do a lot more for you. To my probably mistaken eyes, that defines developer experience. We are giving our developers the best environment where they can flourish and be innovative and be everything, and we are whispering in
20:24Izar Tarandach the back, and be secure, and be secure, right? And, and, I, I, I, I, to, to Matt's point,
20:35Matt Coles Which Matt?
20:35Izar Tarandach If you, the last one that you, that I think you made, if you patch that, that tool, that developer experience tool, and it takes a bit out of the developer experience, the developers will still be able to write code. just not with all the functionality that they are used or demanding. So they will come with the pitchforks to the door of security, but it's not going to stop the company from functioning to,
21:03Chris Romeo we've gotten, we've gotten hyper focused on the IDE versus focusing in on what the developer is actually building. We want the secure by, we want the tool to be secure by default in such a way that it drives the developer to the right decision and takes away any crappy decisions. Somebody
21:22Izar Tarandach goes to the library for me.
21:23Chris Romeo somebody adding a feature that enables the HTTP protocol. in this day and age, should not be allowed to happen. That should be prevented somewhere. Whether it's the IDE putting a little buzzer and going, wah, sorry, wrong answer, you cannot do that. Whether it's the check in process at GitHub, GitLab, whatever you're using, having some rules of engagement that say you can't do that. But, but, Somewhere we need the developer experience to catch that problem. And right now the challenge, this is all crystallizing for me now. The challenge we've had as an industry is that we've pushed all of that stuff off to these, uh, other tools. We've pushed the responsibility to SAST and the other one that
22:08Chris Romeo starts with D and many other tooling tools that exist out there versus bringing it back into the developer's workflow.
22:17Izar Tarandach So that, that's where the, the guardrails come in place, right? I mean, if I start bringing that into the developer workflow, basically what I'm doing is, I can't believe I'm saying this. I'm shifting left. I'm taking
22:32Chris Romeo boo, boo.
22:35Izar Tarandach I'm taking those tools, right, and bringing it in into the developer's desktop, which for some reason, some people thought that it was a really good idea. But, uh, the, the, the thing is that you're just transferring things, you're moving things from one place to another. You're not really, Changing the way they're done, changing the expectations, or changing the responsibilities. You're just giving more shitty work to the developer, and you're making that thing that they use to, to, why people need Macs with 64 gigs memory? Is it for, uh, uh, uh, running AI models? No, it's for running Visual Studio. It's, it's like, it's, it's becoming ridiculous, right? And we want to pile even more on that.
23:24Izar Tarandach So. I don't know, I think that these are very separate things that connect in that word experience. So the last line of Matt's post, uh, what was it?
23:44Matt Coles How about champion?
23:45Izar Tarandach guys. Yeah, uh, if you aren't already, find a champion with your DevEx org, it will pay dividends. So now we want to go to DevEx. and explain security to them, and get a champion that, and everybody knows what I think about champions, and, and, and, again, transfer the responsibility. Do your work the way that suits me. Do your DevEx thing the way that I think it should be, because it's
24:12Chris Romeo but we've argued this point on the, at this, on the, around the security table a dozen times, at least about who's ultimately responsible for security and where do you get the results? You can't, yeah, but you can't get the results from you because there's not enough of you.
24:27Izar Tarandach you, you can't, you, you're totally right. But even if I transfer the responsibility to a developer, the final outcome is still my responsibility, because I decide what responsibility I'm transferring to them. Right? So, at the end of the day, it becomes two extremes. Actually, two extremes and a middle. You have security in here, defining the requirements, you have the developer in the middle actually doing the work, and then you have security at the end verifying that the work was done. I don't think that any of these three pieces in here is going to be better well, better served by introducing dev experience in the mix. I mean, I love the idea of security by design. I love the idea of, uh,
25:21Izar Tarandach Tools that provide security by design. I, I, I'm still not very close on what's a tool that's secured by design. I think that we, we, we need to ask Matt actually what he means here. And, and just a plug in, Matt has a great newsletter called Vulnerable You. No, Vulnerable You, and it's really cool. So,
25:41Matt Coles Sorry, secure by default was the statement that Matt, uh, Johansson said, not secure by design. That's important. It's important. Secure
25:49Izar Tarandach Yeah, yeah, yeah, yeah,
25:50Matt Coles by default are two different
25:51Izar Tarandach It's a default, completely
25:52Chris Romeo what I've figured out as I've continued to study these things is everybody thinks they know what that means and nobody freaking knows what it means or how to do it. Which is hilarious
26:02Izar Tarandach know I don't.
26:03Chris Romeo it, but most people are like, people are throwing it around now because of CISA's document and everything. And I'm like, I'm kind of, I'm starting to unpack that for some research I'm doing. And I'm like, nobody knows what this means. And there is no defined definition of how to do it.
26:17Izar Tarandach You, you know when secure by default?
26:20Matt Coles We tackled this
26:21Chris Romeo by default is easier than secure by design to wrap your brain around.
26:26Izar Tarandach secure by default to me, and I know that nobody asked, but to me, it means that once upon a time I worked in a place that was very, uh, forceful, about publishing hardening guides. Every product goes out with a hardening guide. To me that means the product is going out by default, open, and it's on the customer's side to apply these hardening guides to close it. Secure by default to me means it leaves the door closed. It goes out
26:59Matt Coles We, we, we had this, we had this argument, this discussion about loosening guides. That's the new term that CISA wants to use. yeah,
27:07Izar Tarandach boo, boo, boo.
27:08Matt Coles because releasing, releasing with a guide should. It should, it implies that it was left, it was open when it left the factory, but it should be hardened, and that's why we give hardening guides. But, but loosening is not the right, I think we've decided, we've, we've had this conversation that loosening is not necessarily the right thing. It's a secure deployment guide that gives you options for deployment. But it should be, by default, one of those secure options. Oh,
27:38Izar Tarandach so we go right to the, the heart of the thing in this dev experience discussion. If we do something that's completely closed, like the IDE or the libraries or whatever, and we say we are not giving you guys a loosening guide because that's a bad thing. I think that that's what Matt means when he says they are going to find ways around. So if you don't give the losing opportunities, they will find them and you won't have control over them because you believe that, that everything is by default,
28:05Chris Romeo But you have to make it easier for them to use the new tooling and the new guidance that you're enforcing in the tooling than if they had to go do it themselves. So you can't make it harder so they have to try to sneak around it. You have to enable them with this and think of it as a positive instead of a negative.
28:25Matt Coles so Matt basically, Matt said though, it's my job as a security person, or his job as a security person, to give developers tools that are secure by default. so that they don't have to find workarounds. And what Izar was saying, and I agree with him, is if you give a locked down system, the reason why I don't recommend hardening an operating system before you give it to a developer is because they're gonna have to unharden it to do their job. And so if you give a secure by default component, and let's talk about frameworks and components more so than business tools, although the IDE is probably a good example of a business tool, developer tool that if you give it in a secure by default state,
29:09Matt Coles and it doesn't work the way the developer needs it to work, they're going to have to find workarounds. And that counter that contradicts Matt's argument that if I give a secure by default tool, developers will be happy and can do their job.
29:22Chris Romeo we've gotten hyper focused on the IDE. but we're not talking, this is not talking about hardening a IDE environment. And it's, it's, it's the decisions that are coming and being implemented in the IDE are the things that measure this. It's, it's secure by defaultness. I'm making that
29:40Matt Coles I
29:41Izar Tarandach I think that it's talking about giving to the developers something that's already closed from the shop. It comes to them closed. This is what you're going to use to work,
29:51Matt Coles Right. And, and, and, and you can't, because, because, well, so let's, let's, let's go with, with the line of, of the thread that you were just, was just suggesting there. So if you're talking about what gets built in the IDE, and you give, so first off, you're not going to give pre hardened code, right? You're not going to give secure by default code because the developer, that's the developer's
30:11Izar Tarandach No, yes, you will, you will, you will, libraries,
30:14Chris Romeo it's the
30:15Izar Tarandach components, libraries, third party.
30:17Chris Romeo I don't think that secure by default IDE means that it only allows secure code. But let's talk about, let's use an open source example, okay? So, if I want to add a component to my project, should I as a developer inside of a big company be able to go scour the internet and find any package I want and load it into my, via my IDE. I say no. I don't, I, in a large engineering, a startup is a different thing. Startup is, you get, you, you got to have that, that ability to move and shake and do things differently. But in a established engineering organization, I do not want packages being added from some place where we have no idea how trustworthy they are. They could be, have been backdoored, they
31:00Chris Romeo could be malware. And so I want the IDE to prevent somebody from adding that package that hasn't been vetted into a
31:10Izar Tarandach Yeah, we, we spoke about that with the visual code example, right? Are you going to get it from a repository or you're going to go everywhere? But
31:18Matt Coles as long as there's guardrails, as long as there's guardrails for innovation, that's fine, right? So developer,
31:23Chris Romeo there's gotta be a process where you can say, Matt, you can say, Hey, Hey, but I found this cool library. It's not on the list. Somebody's got to review it and we got to get this into our product because it's great. That's fine because then, but then we, because we don't, we don't just have random software being added into our software. And we wonder why our build systems got hacked or our, or some other type of challenge that came as a result of an open source component that was, that was breached.
31:48Izar Tarandach so from experience there will be those draconian places, and by the way, this happens with Chrome extensions as well. No, here's the list of the ones that you can use and whatever else, shut up, we don't want to hear about it. will be those that will say here's a set of procedures and tools that you can use to make sure that your risk is as little as possible, and after you run your proposed extension or whatever through this, then we are sort of okay with you using it. And there are those places that say do whatever you want, right? But, Let's bring that to, to, let's get out of the IDE for a second. Another thing that we all love and love to hate, Java. There is that file that java. whatever.
32:35Izar Tarandach sc. whatever, that I forget the name, where you can define which ciphers are available in the decrypted extensions of Java. If there was a way to enforce that every single developer would use only that version of that file, I don't believe there is, but if there was, would we say that this is an example of a developer tool that's secure by default, and that There might be some cases where a developer could come and say, I can't work with these ciphers, I need something else.
33:16Matt Coles I don't think it's a developer tool.
33:19Izar Tarandach Compiler's not a developer tool?
33:21Matt Coles The configuration of your it's not your compiler. It's the configuration of your runtime, I believe. It doesn't affect the
33:28Izar Tarandach Right, right, right, right, right, right, you're right, you're right, you're right, you're right, you're right,
33:31Chris Romeo again, it's a design decision. Right? It's a, we're back to that. That's a design decision. What crypto algorithms do you allow to be included in a project? Now,
33:40Izar Tarandach But you are guardrailing it, you're guardrailing it by only giving the known good options for that design decision.
33:47Chris Romeo yeah,
33:47Matt Coles but it's not
33:48Izar Tarandach default.
33:48Chris Romeo it around the other way. Do you want a developer to be able to add DES into a, that configuration file and start using DES encryption, which we all know isn't even encryption. We can, it's barely, we can barely call it encryption at this point because it can
34:03Matt Coles DES with the Triple DES with the sync key Triple DES with the sync keys is at least Something.
34:10Chris Romeo Yeah, but that's not that. Yeah. Triple that. Yes. Now you said triple DES, which is a different algorithm. Yes. It's using the same, the
34:15Izar Tarandach But, but that's not a flip. That's not a flip, that's the continuation of the scenario.
34:21Matt Coles Well, so, so then, then the extent by extension, are we, are we bringing, are we bringing, um, specifications, whether that's coding conventions or, or in this case, a crypto specification. into the fold of secure by default and or by developer tools. So as an example, if we set a standard for what crypto algorithms to be used or can be used and or cannot be used, we will end up with things that are secure by default. Well, secure by design, potentially secure by default. Uh, and, but, but that's, that's a standard. That's a specification. That is not a tool. And it's not a developer experience,
35:06Izar Tarandach but it's not a tool, yeah.
35:08Matt Coles but it's not really a developer experience, is
35:10Chris Romeo I mean, the, the, the, that, that's certainly everything you just described is right on, that is part of secure by design, potentially secure by default. But let's bring it back to this conversation and say, do you want the technology that your developers use to enforce that for you? Because that would make it, that would make the technology that the IDE they're using secure by default. If they go to, or maybe it's when they go to check in the code, like we can also push it out of the IDE into the source code control system.
35:40Matt Coles Or a CICD, or
35:42Chris Romeo yeah, I like that a little bit less because now we're letting them waste time on something that's never going to get approved. And they may spend three days implementing something and they go to check it in and the thing says no way you can't use DES encryption. What are you doing? Whereas if it's in the IDE and we're able to enforce that and say, hey, this is something we don't do. We don't allow DES encryption anymore because it's not 1964.
36:05Matt Coles it's part of the SAST, or it's identified by the SAST
36:08Izar Tarandach so here's, here's another take. What if we found a way to do all this securing by default that he's proposing in those tools? Would it be fair to say that going on an 80 20 arbitrary divide, 80 percent of the developers don't care enough? or don't reach the level that they have to disagree with the defaults of the tools that they are using. They just want to get cold out of the door. And the 20 percent that do disagree are hopefully going to engage in a dialogue with security around those defaults because they understand that they are breaking the defaults. So could we say that if we go the way that he's proposing, By closing things as part of developer experience, what you get at the end of
37:02Izar Tarandach the, of the, the, the, the experience is that either you're going to have things that are secure by default, 80 percent of the cases, and 20 percent of the cases, people are going to come and talk to, to security and actually have a dialogue around those defaults and perhaps figure out that, hey, we need different defaults, or, hey, we need a different way of doing this kind of thing. right? So could it be that what he's proposing, rather than leading to a more secure and beautiful future, actually leads to a higher level of dialogue between security and developers through developer experience? Does that make any kind of sense?
37:46Chris Romeo Yeah. Yeah.
37:47Izar Tarandach Yes,
37:49Chris Romeo I think the, the connection once again. We've all, I think, agree on this point that part of, part of being a good security person is understanding the plight of the other functions across the business from developer experience to product to engineering to the best way that we support them is when we walk a mile in their shoes and understand what they're going through. So yeah, I mean, I think you're, I think you're going in the right direction here about the, this is going to create more dialogues. If you, so yeah, that comes back once again to the exception process. The secure by default tool cannot be draconian in that it's locked down and there is no, sorry, we don't allow that. Uh, it has
38:35Chris Romeo to have an exception process where you can have a request sent to Matt as our security engineer and he can review it and go, um, okay, I can't find any reasonable reason why you'd want to use DES encryption. Make your argument, give them the opportunity, but it could be something a lot more, you know, it's okay. Here's a, here's an even better example. The, um, the, the engineer, you have a really, uh, engineer who knows crypto and they say, well, um, I want to use, uh, some of this, this new quantum crypto. in our product, which is potentially, there are quantum safe algorithms or something like that. They want to use, they want to use something that's, or they want to use something that's actually
39:16Chris Romeo potentially a higher level of security than what you're doing, what your minimum standard or your maximum standard kind of prescribes. Then Matt could have a dialogue about, well, you know, we don't think quantum safe algorithms, we're not that afraid of quantum computers yet. But thanks for looking into that. And we'll, we'll, we're going to be chasing that in our minimum standard for the future. And we'll definitely make a note to circle back with you. Cause it looks like you've got some really strong opinions,
39:41Matt Coles So, I
39:42Izar Tarandach I'm using a memory safe language. Hehehehe
39:46Matt Coles actually disagree, I think I might disagree with that just a little bit. Security should be consulted, I think, in that case, because there's value in that conversation. But you don't want to, if a product wants to go beyond and innovate, because maybe they have pressures coming from their customers. Right? Then we should be supportive of that within reason. I, I, I, that's the setting the guardrail. It's not the exception process, because what we're talking about here is making a choice that goes above and beyond the minimums, and you should never set a maximum, right? Security should never say, well, you, you should, you, you should at least use AES 128, but you can't use anything more than
40:32Matt Coles AES 256.
40:33Chris Romeo Yeah. But what if they want to use a different
40:35Izar Tarandach No, no, no, that's not true. That's not true. What if you have a performance problem and if you go too high, that's going to bring things down? There is such a thing as too much security. No, no, no, but security could be influencing that performance problem. What I'm saying is there is such a thing as too much security for any given
40:51Chris Romeo There is a such thing as reasonable security too.
40:55Izar Tarandach There should be.
40:56Chris Romeo It
40:56Matt Coles reasonable security doesn't preclude, reasonable security does not preclude the ability to go above and beyond, right? Because a developer, development team may decide they want more than
41:06Izar Tarandach But that's future proofing. What today is reasonable tomorrow could be the minimum.
41:11Chris Romeo yes. So
41:12Matt Coles like RSA, like RSA, going RSA 4096 on a, on a key, key length, as opposed to RSA 2048, which is the, which is the
41:20Izar Tarandach Right, but if you tell people you always have to use 4096, anybody writing for those things is going to come and say I can't do that.
41:28Matt Coles That's right. Developers will find ways around though, if those are too stringent guidelines.
41:32Izar Tarandach Yep.
41:33Matt Coles But that's not secure by default. That is guidelines or specifications, right? Guardrails, the choice of tools in the post originally, is where I think we're getting hung up on. The definition of what a tool is, developer tool specifically, versus the things that developers use. that they configure, that they build, that they produce. Those are the things that we're talking about more
42:01Izar Tarandach I think that the verdict here is that we have to invite him to explain himself.
42:06Chris Romeo It took us 42 minutes to get to the point where we decided we should have just had the author of this post here so he could just tell us what he meant.
42:13Izar Tarandach yeah, I don't think we quite grasped what he was going for, but
42:17Chris Romeo I mean, I
42:17Izar Tarandach we're in the ballpark, I think.
42:19Matt Coles Yeah. I mean, there's a lot of discussion points here and, and, you know, maybe he didn't intend, maybe he didn't, wasn't thinking in, in any of these areas, but from a secured by design and secure by default standpoint, you kind of have to.
42:32Izar Tarandach Right, right.
42:33Chris Romeo Yeah, I think that's a good place to end our discussion for today about, uh, this developer experience, secure by default, guardrails, and DevEx champions. I feel like we made a little bit of progress. I don't know if I could measure how much, but I think we made a little bit of progress forward, at least in our own understanding. We had to crystallize some points to make this whole thing work. So, uh, folks, thanks for listening again to The Security Table and stay tuned for another episode real soon.
Transcript supplied by Buzzsprout. It may contain transcription errors.
View the original transcript on BuzzsproutOriginally published as The Security Table. Part of the AI Security Table archive.
