>>>>>>Go write an auto-pilot system for a 400-seat airliner.
>>>>
>>>>Would it be easier to write one for a 200 seater?
>>>
>>>No. But the ramifications of a design flaw would be less. We both picked arbitrary points in the spectrum Towards the one end is the single catastrophic failure resulting in hundreds of deaths; at the other a fixable mild frustation when, say, submitting a web form.
>>>
>>>Because of the nature of the tools generally discussed here I'd hazard that the closest that clients of most UT posters come to a risk of physical injury is being hit by a print head - or possibly an over-enthusiastic cash drawer :-}
>>
>>Probably the coolest place I ever worked was a pure custom software development shop. (For the real old timers here, it was RDI Software, renamed Geneer while I was there). Every Friday after lunch we would have an all-hands meeting which often included demos of work in progress by one of the project teams. One Friday it was a process controller that had been developed by the Factory Automation group. The guy who was doing the demo pointed out a spinning gizmo inside the machine and said, "Touch that and you're dead. I'm not joking." He added wryly that it impressed him to be working on software that can kill you.
>
>I must confess to getting much more satisfaction/enjoyment when working on software where the outcome has concrete physical manifestation (other than something moving around on a monitor or a gradually emerging piece of paper) - preferably with real-time processing requirements.

I did gantry robot positioning with a 1kw UV source on photopolymers. Any mistake immediately turned into a concrete physical representation of itself. And it was "much more satisfaction/enjoyment"