Your 2x rule seems to be based on the Nyquist sampling theorem, which says that a signal of bandwidth B can be accurately reconstructed by sampling at a rate of 2B. However, in practice, a sampling rate higher than that is usually recommended. Normally the specs for an oscope will tell you what the max bandwidth is for the scope, especially if it is a quality manufacturer like Tek. As for clocks, if you are looking at a clock as an analog signal, for example looking at logic delays, you might want to see events on the scale of rise times and fall times, which requires a much higher bandwidth than 2x the clock frequency.
