Good article. Thank you. You make some excellent points.

I agree that source access is not sufficient to get a secure software and that the many-eyes argument is often wrong. However, I am convinced that transparency is a requirement for secure software. As a consequence, I disagree with some points and especially that one:

It is certainly possible to notice a vulnerability in source code. Excluding low-hanging fruit, it’s just not the main way they’re found nowadays.

In my experience as a developer, the vast majority of vulnerabilities are caught by linters, source code static analysis, source-wise fuzzers and peer reviews. What is caught by blackbox (dynamic, static, and negative) testing, and scanners is the remaining bugs/vulnerabilities that were not caught during the development process. When using a closed source software, you have no idea if the developers did use these tools (software and internal validation) and so yeah: you may get excellent results with the blackbox testing. But that may just be the sign that they did not accomplish their due diligence during the development phase.

As an ex-pentester, I can assure you that having a blackbox security tools returning no findings is not a sign that the software is secure at all. Those may fail to spot a flawed logic leading to a disaster, for instance.

And yeah, I agree that static analysis has its limits, and that running the damn code is necessarry because UT, integrations tests and load tests can only get you so far. That’s why big companies also do blue/green deployments etc.

But I believe this is not an argument for saying that a closed-source software may be secure if tested that way. Dynamic analysis is just one tool in the defense-in-depth strategy. It is a required one, but certainly not a sufficient one.

Again, great article, but I believe that you may not be paranoid enough 😁 Which might be a good thing for you 😆 Working in security is bad for one’s mental health 😂