It comes down to the quality of the solution, versus the added headaches of managing dependencies. It's not a religious issue, but one of experience and engineering. It's also a good idea to make sure that if you're inventing something that someone else might eventually provide, you at least loosely couple it, so it can be swapped out later.
With those thoughts in mind, don't get stuck in analysis paralysis. Use judgment and move on.
To sum up: He's not saying that developers are science illiterates, but that they're not necessarily any better outside their fields than other non-experts are.
The one thing I'd respond to with regards to this assertion is that a lot of software engineers get exposed to complex systems. Any of your better troubleshooter types has had enough experience with a system of sufficient complexity to humble them against thinking they understand it all, and by extension they have a pretty good detector for when someone's making such a claim without backing it up. I think that's where you see a lot of the software people being critical of AGW extremism - there's a place for the attitude, "I recognize man's impact in this area, and I concede CO2 is likely a factor that's impacting the environment, but I'm not convinced you understand the system well enough to make meaningfully accurate predictions that should be allowed to drive policy".
It's perfectly reasonable to disagree with that kind of statement, but extremist predictions that haven't panned out have undercut the legitimacy of more respectable voices, and feed rational skepticism of predictions.
Therefore, somebody wanting me to invest money has to make a considerably better case during deflation than inflation.
Not really. You just don't need to chase risky crap to get 7% returns. You can be quite comfortable taking low-risk investments running 3%, and netting greater real gains.
Always draw your curves, then plot your reading.