How simplified should popular science be?
This is something that came up again in another thread, and should be very important. I know that some people are against the way advanced science topics are explained, and I agree most of the time. But how much (often pedantic) metaphor should be used, versus the arcane (sometimes pedantic) terminology the experts use?
For example, if I want to explain how artificial neural networks are constructed I might conjure up imagery of tubes, buckets, and adjustable valves instead of talking about a network of variables with values equivalent to the sigmoid of a sum of inputs multiplied by each input's weight.