It would nice to see "expected speed with variance in standard deviations" advertised.
Superficially a standard deviation would be nice (because sheeple can't grasp the whole traffic concept - even when they're stuck in it on the freeway...), but... (in reality) It would unfortunately only end up opening the door for Zero to be one of the standards it deviated to...More often than not.
So when the "freeway" grinds to a screeching halt (Speed=0) it will be construable as an acceptable part of the (fine printed) "Norm"...Because it is within the "standard".
I was thinking of speeds expressed something like this:
Expected Speed: 20 Mbps with a standard deviation of 2 Mbps
So, 50% would get speeds above 20, and 50% below.
You can't really game a system like that. If you start counting zeros, then you're widening the standard deviation, which is a bad thing. Ideally, consumer would ideally desire standard deviations that are very small. (The first "ideally" is to express that in an ideal world people understand basic concepts like this.
So, if you have a 20 Mbps speed, with 20 Mbps standard deviation, you're highly likely to get zero speed. The left side of the graph will have an abrupt cut off, while the right side tapers and then has a small cut off.
As an example:
100 Mbps with 50 Mbps SD
80 Mbps with 15 Mbps SD
Provider B is a far better bet there.