Question: At what size (both huge and tiny) does a number become more easily understandable in scientific notation? I'm pondering this question for myself. I bet it's different for different people.
The answer to this question is absolutely different for different people. I became intimately familiar with scientific notation a very long time ago. One of the important, fundamental advantages is that the true accuracy of the number is conveyed by the number of significant digits. Of course this requires that it be used properly. If you see a number like 5.000x10^3, you know that the number has four significant digits, because those zeros wouldn't be there otherwise. If you see 5,000 instead, there's no way to know. Most people would implicitly assume that all three of the zeros in 5,000 are significant, but in general this is not a valid assumption.
Engineering notation seems very similar to scientific notation, but the twist added in engineering notation is not without a subtle complication. In scientific notation, the radix point is always located to the immediate right of the first non-zero digit, and the power-of-ten exponent can be any integer. In engineering notation, the exponent is restricted to integer multiples of +/-3; the radix point is shifted as needed to accommodate this requirement. The scientific number 5.7x10^-4 is written .57x10^-3. Or is it written 570x10^-6? Hmmm. How many significant digits are there in this number?
In scientific notation, the rule is that the least significant digit does not need to be fully accurate but must contain some accuracy. Occasionally, there will be a number with just one significant digit. In scientific notation, 6.x10^-5 is a proper number, notwithstanding that there are no digits following the decimal point. In engineering notation, is .06x10^-3 proper, or is 60x10^-6 proper? If it happens that 60x10^-6 is deemed proper or acceptable by whatever governing body is authoritative for the use of engineering notation, then to me it seems that engineering notation has lost a fundamentally important attribute of scientific notation. If 60x10^-6 is deemed properly equivalent to 6.x10^-5, then it would seem to me that every number written in engineering notation should be accompanied by a tolerance value.