(Just using Donald's email as a hook)
How about we separate problems. If people think there are better ways
to generate the entropy for the random selection algorithm, write a
draft and tell us clearly. Either the community agrees or it doesn't.
Once we have such an algorithm agreed we can then discuss whether wee
would prefer to use the initial run to fill extra slots, rerun all the
time when we need extra slots, or some combination. Personally, if the
random selection process is simple enough, that would probably help
mitigate my negative reaction to adding the step. For now, we stick
with using the initial run, please.
Yours,
Joel
On 8/23/2022 5:18 PM, Donald Eastlake wrote:
Coming up with randomness good enough for this process is an already
solved problem.
Thanks,
Donald
