've just tried it and it makes no difference whether there is hand there or not, the lower 3 bits are not predictable, which is what I am interested in.
Excellent. You have what you need.
So the rest of the post must be driven by your curiosity. Nothing wrong with that.
So I have to disagree, in my experience using analogRead on a floating pin for the 3 lowest bits does not make it predictable.
What was the relative humidity at the time? Any lamps near by? What about a non-incandescent one? How close was your computer? How stable was the power supply to you Arduino? Was it night or day when you performed your test? How far away was the nearest thunderstorm? Was your WiFi on or off? Was the AC or heater running?
When you brought your hand near the floating pin, what was your other hand doing? What was under your butt? Metal chair? Fabric chair? What clothes were you wearing?
Did you check for a repeating sequence or did you look for a single repeating value?
Did you repeat the experiment a few times after the board is first powered?
Did you reposition the cord powering the board? How about the cord powering the near-by computer?
Think about it. You are reading a signal which at 10 bits and 5 volts means that:
LSB0 = 4.8mV or 0.0048 V or 0.098% of full range
LSB1 = 9.7mV or 0.0096 V or 0.195% of full range
LSB2 = 19.5mV or 0.0195 V or 0.391% of full range
Yup. That little bugger is sensitive. Which potentially makes it a good choice for collecting entropy.
The problem with a floating pin is that all the questions above (and a few not there ;)) effect it. The same things that make the voltage on a floating pin wildly erratic can also make the voltage very stable. That's why the technique is unreliable.
Can you really move your hand near (but not touching) the pin to get exactly the number (0x05 for example) you want?
No. But that is not what you originally asked.
We'll have to agree to disagree I suppose, but I'd be interested in hearing your experimental setup so I can try it here.
It? By "it" are you referring to the stuck-value-by-not-touching-hand experiment?