By William Poundstone
Some scientists have long predicted the end of the world. By nuclear war, or environmental disaster, or rouge AI, or...
This book goes into these calculations, their history and the substance behind them.
They are usually based on "Self Sampling" statistics augmented by a Bayesian math. The anthropic principal is a great example-- all laws of physics must support the creation of a universe that allows for observers (like humans) to exist, otherwise the observers wouldn't be around. For example, if the universe was 2 dimensional, or 4 dimensional, then gravity would behave very differently and we wouldn't exist.
Then there is also the Lindy effect-- when you randomly meet an object (or person, or process) on average you will have met that object half way through it's lifetime. If you know how long the object has existed so far, then you can predict the full life time of the object.
But garbage in, garbage out. Sometimes this math can be used to justify bull shit. Maybe your analysis is wrong. Maybe you are missing something.
So much of this is subject to opinion. Like "What is the probability of a rouge AI taking over the earth? ala Terminator."
You never really know until its too late.
Some scientists have long predicted the end of the world. By nuclear war, or environmental disaster, or rouge AI, or...
This book goes into these calculations, their history and the substance behind them.
They are usually based on "Self Sampling" statistics augmented by a Bayesian math. The anthropic principal is a great example-- all laws of physics must support the creation of a universe that allows for observers (like humans) to exist, otherwise the observers wouldn't be around. For example, if the universe was 2 dimensional, or 4 dimensional, then gravity would behave very differently and we wouldn't exist.
Then there is also the Lindy effect-- when you randomly meet an object (or person, or process) on average you will have met that object half way through it's lifetime. If you know how long the object has existed so far, then you can predict the full life time of the object.
But garbage in, garbage out. Sometimes this math can be used to justify bull shit. Maybe your analysis is wrong. Maybe you are missing something.
So much of this is subject to opinion. Like "What is the probability of a rouge AI taking over the earth? ala Terminator."
You never really know until its too late.
No comments:
Post a Comment