lily
it may be nice to mention why practical implementations use ≤ for cutoffs
rather than <

typo
i actually don’t know the reason for this
(either the historical reason – why particular people chose to do it and how
  that influenced the implementations of others –
or all the objective effects on typical search trees – although I can conjecture)

lily
The earliest easily accessible reference I know of is Moore and Knuth's
  1975 paper which summarised the state of the art at the time; unfortunately
  it does not clarify why ≥ was chosen and merely show that historically it was;
  likely from their constraint that α < β, but that is an assumption