[caf]

The ABC has an article with a quote that explains it better (http://www.abc.net.au/news/2012-10-03/tunnel-closures-cause-...):

  CityLink spokeswoman Kate Murphy says the tunnels had to close as a precaution.

  "It's not just overhead signs [that are not working], it's also our safety systems
  as well," she said.

  "For example, in the tunnels the deluge systems or the smoke extraction systems
  which we may use in an emergency, we can't actually operate those as well, 
  which is why we've had to take the dramatic measure of actually closing the
  tunnels."

[DrStalker]

I had to hunt the article to find that, since the first impression is just speed limit signs/traffic lights being out which is not major. (and plenty of posts here suggest other people thought the same thing)

The variable speed signs always have extra signs along the lines of "(80) if variable limit sign is disabled" as a fallback.

Not being able to handle leaks, fume buildup, fires and other emergency conditions is a completely different story.

[rurounijones]

The fact that their safety systems (Which anyone would expect to be separate from the traffic control systems) are also affected makes me wonder about the "computer glitch" angle.

[Zenst]

Like 99% of news reported computere glitch's it is some human mess up that caused it, either by design or error or a homer moment were somebody pressed the wrong button. Could even be somebody overloading the electrical phases in the datacentre and tripped the UPS into action for it to barth at the imbalance (seen that one).

This is why they say computer glitch and count on the majority of non technical people to not question it.

[dspillett]

> Like 99% of news reported computere glitch's it is some human mess up that caused it

If the system allowed the user to do something that impacted safetey in such a way, then the computer glitch is that the computer should stop that sort of thing happening. Human error happens a lot and you have to account for it in your designs.

Of course there are times when you have to trust the user because it is not practical to do those tests (in which case the human should have procedure to follow that mitigates the risk), and instances where the user wil override the checks (when the machines say "I really don't think you should do X because Y, click 'I understand the consequences of my propsosed action' to proceeed") - then it is genuinly human error is something goes wrong.

[batgaijin]

Hmm... actually yeah that is a really good point.

As much as I don't want to sound like a conspiracy person, that level of connection between those components is a bit too much to believe. Especially in Australia where they have insanely high standards for construction/electrical engineering.