[repelsteeltje]

I think it will take a while for people to realize this effort looked great, but wasn't the right approach. Or no silver bullet, at least.

The presentation with a simple diagram that combines this data with an sbom to yield "information" gives me navel gazing vibes of UML being the future of coding.

Just as architecture didn't equate to well designed and maintainable software, I fear this initiative won't fix horribly outdated and vulnerable deployments. Software life cycle, deprecation, abandonment, supply chains are mostly a process problem, standards and technology won't fix that.

[Arnavion]

It doesn't force someone who already wasn't checking their dependencies for CVEs / maintained-ness to start doing that. It does make someone who *was* doing that be able to show they're doing that in some standard way.

In other words it doesn't force you to add an SBOM + EOX checker step to your CI pipeline. But if your compliance auditor wants you to check your dependencies, adding such a standardized step makes it easier to satisfy the auditor.

[repelsteeltje]

I'm basing this mostly off first hand and anecdotal evidence - but through the years I've found that the major contribution of audits lies in having to think about the checkboxes every now and then. And what they mean in the context of my organization or project.

Rarely have I found that compliance to the goals was an issue in themselves. Or that making changes to tick a checkbox correlated to material improvements.

That is to say that if this leads to more efficiency and makes it easier for compliance audits and such, I fear is stream lining the least impactful part of its goals.

[hiatus]

> Rarely have I found that compliance to the goals was an issue in themselves. Or that making changes to tick a checkbox correlated to material improvements.

I am confused when I hear people say stuff like this. I guess if you turn on a tool and never look at it again, it won't result in material improvements. But complying with regulations or a particular compliance regime should _absolutely_ result in at least _some_ material improvement to your security posture. Like you can implement segregation of duties just as a checkbox, or use the requirement to revisit the way you gate changes to production, as just one example.

[repelsteeltje]

It depends on where you're coming from. Your code base, that is.

If it's already outstanding, you spend a lot of time revalidating what you already know and it's often a noisy process with many false positives.

If it's in a horrible state, however, the regulation often leaves a lot of wiggle room where you do some work to achieve, say, PCI compliance and then spend a lot of time arguing why this and that don't apply in your specific case.

So admitted, the is probably some improvement in the latter case but it's hardly proportional.

So IMHO, it doesn't help those of good will & expertise and does too little for the negligent. It adds noise and in the end quality still depends on factors other than compliance and certification.