You know, even barring the fact that some of the Reg writers may actually have a modicum of knowledge and experience, I think you should bear in mind that many of the Reg readership have some pretty extensive knowledge and experience in matters IT.
At the end of the day, hacking a power plant's infrastructure, or hacking a web site, it's all the same. You need knowledge of the protocols the remote system responds to, a reasonable idea of the infrastructure between you and them, and a knowledge of what applications/hardware sits between their connection point, and an area inside their network you can establish a beachhead on.
So that means either a fairly large amount of probing, or, more likely bribery and intelligence gathering. There is no "magic application" that you push button, receive access. Cracking into a system is long, boring, tedious work, the bulk of which isn't done in front of a computer. To penetrate say, I don't know, my personal web server, you might not need a lot of expertise, I use a set of readily available open-source apps, and apply 80/20 rule to the security on my web server. All eight visitors will be traumatized if it gets cracked, but the reality is it probably won’t, it’s not big enough for the 20% of the crackers out there who could get through to bother with.
That said, if that web server was something I threw a few months at securing, used obscure tools configured in very non-standard ways, and periodically changed configurations, (I could even have a pseudo-random change script to do so automatically, and a corresponding one on my backup and update servers to keep them able to transmit data with the changes,) then getting into that server would be a miserable pain in the ***.
Now, give me a few million and a research team of 20, and I sure as hell could 'secure' a power plant, while maintaining connectivity to critical networks for updates/monitoring/configuration: all the above ideas, with a magic "detect attack, disconnect from network" trigger, a few backup networks, maybe even a non-internet connected communications link to a remote site, (radio? dark fibre? Military networks?) in case of extreme attack. Throw some BODIES at MONITORING, (you know, trained professionals who know what they are looking for,) and bob's your uncle.
Certainly that's an overly simplified way of looking at it, but that's the point. We, the Reg readership, know it is overly simplified, and we can all read between the lines. The whole concept of "the evil hacker boogymen are coming to get us is just a giant bucket of deja moo: the feeling you've heard this bull**** before.
As long as you design a network along the principles of the least privilege, and you don't connect anything to anything else unless it absolutely must be connected, (and you implement systems that will cause a disconnection if under attack, as well as appropriate overrides,) then any computer system can be secured. The problem is rarely, if ever the computers, the problem is, as always, the people involved.
From the twits who brows myfacepornbook with added browser exploits on top, to the design decisions being made by politicians and managers instead of IT professionals, people are the weak point. The greedier that the people in charge are, the more corners they will cut. The more corners they cut, the harder such systems are to defend. If the ‘merkins want to defend their country against cyber attack, the only way they will do it with regulation, not weapons. Regulate that critical industries must meet certain minimum security standards, and have people of a minimum talent and ability monitoring such systems 24/7.
That, however, isn’t as "sexy" a solution, and doesn’t get the pork out of the barrel and into the hands of their friends. Long live the land of the 'free'.