As I understand it there are big differences between employment practices and customs between the UK and the US. In the UK it would be highly unlikely for someone to loose a job over being late for work or even missing a whole day of work if such a situation was to occur. There'll also more than likely be an over-ride for the pump operator (such as for legally filling petrol cans or car's once an issue had been resolved with a few phone calls). The system will also more than likely be able to be deactivated should there be connectivity problems.

Personally I'm all for it. Vehicle Tax, MOT and insurance are legally mandated in the UK, the small minority that flout the laws cause plenty of grief for those that end up involved in accidents with those that don't have insurance (or safe vehicles) and drive up the cost for the majority who play be the rules.

If the members of your team are globally distributed, I'd drop conference calls for many decision-making meetings. Whilst you may have staff that are happy to partake in a conference call at 7pm local time, they will get frustrated with not being as fresh as others where the meeting is being held at 10am their time. For those that are tired, having already worked a full working day, it will prove very difficult for them to efficiently express their side of an argument. Look at how the open source community solve this - discussions are carried out over a number of days via email, with plenty of time available for each contributor to reply and formulate their answers.

The same law that is being proposed to disconnect individuals is coming under fire from the British library (among many others) as it will leave them quite exposed. What if the libraries decide that internet access isn't worth the risk?

Um, how about cost?

If your site doesn't work, they'll just move on to one that does, not complain to Microsoft that xyz.com doesn't render properly.

Actually, in my experience they will complain to those around them that the Internet thinggy is broken and this other one (firefox) still works.

There are several "provably secure" computer systems. As in you can demonstrate they fulfil certain mathematical constraints and those constraints are absolute. Then you have to write the code and prove the code, then you have to hope the prover is correct and the hardwareis correct. Nothing is 100%.

As to the randomisation stuff - yes I've got examples, and we've hit the same thing in Linux with randomisation. You get cases where memory scribbles cause a problem only if the layout happens to be a specific variant (especially with stack randomisation). From "either it dies or it works" you get "1 in 10,000 times xyz app blows up". That does make debugging much much harder. Of course a good reply to that is "so improve the debugging tools".