As if I needed another reason to sleep badly... We just had an alarm system installed in our home. The school decided it would help us... sleep peacefully? I always thought that alarm systems were for people in suburbs with stuff to protect or people in dodgy neighborhoods who feared for their lives. Our '95 Jetta had an alarm system that scared me once or twice. The one in our house woke us up three times during the night (here in India we say 'thrice') with all of its whooping and blaring.
The nature of such a system is full of interesting considerations. Interesting from an academic standpoint, concerning if you have to live with it. To design such a system, we have to keep altering the mechanism based on our ability to sense different things. We can ponder this and similarly inferior logic that explains the way smoke alarms beep when supper is ready, the way a forgetful person becomes locked out of their email account, the way a medical test produces false positives (sometimes much more frequently than real positives), the way an auto-flash sometimes wrecks your photograph, the way some algebra generates extraneous solutions, and the way homework grades and test grades correlate poorly. So here is the gist of the dilemma and the reason I slept badly enough last night to spend time writing something like this.
We want a home-security alarm that sounds when:
A bad person bent on evil enters my house... but this is impossible without an omniscient system that guards every conceivable point of entry, including the kitchen drain. So we settle for an alarm that sounds when:
A bad person bent on evil enters my house through the door... but this is impossible without a system that understands thoughts and intentions, and besides, the system might be confused if anyone ever wanted to hurt us and take our things with a pure motive. So we settle for an alarm that sounds when:
Any person enters my house through the door... but this is impossible for all but the most sophisticated machines. It will also require that we take the responsibility to disarm the system while innocent people are entering through the door. So we settle for an alarm that sounds when:
A door opens... this is getting more feasible, but we are suddenly unable to say with confidence that there is a person involved, or which direction that person is going through the door. Boy would it be annoying to have an alarm system that sounded when you exited your own house. The door opening seems like a simple mechanical motion, but desiring to detect it with one simple boolean variable and one simple electronic sensor will force us into a simpler condition for setting off the alarm. So we settle for an alarm that sounds when:
A small magnetic sensor on the door detects that the door has been opened... but the sensor cannot tell that the door has been opened, it can only detect the presence or absence or change of some electrical impulse. So in the end we settle for an alarm that sounds when:
A small magnetic sensor on the door detects a change in the strength of a magnetic signal. This is possible, so we install the thing and turn it on and learn that the condition that sets off our alarm is true much more often than there is a bad person with evil intentions entering my house. Ironically, we didn’t even turn the system on last night because we didn’t want to mess with it, and it still woke us up thrice.