This question was answered several years back, but, just for fun, here is a simple proof of the upper bound. We give a bound on the expectation, then a tail bound.
Define r.v. di to be the depth of node i∈{0,1,…,n−1}. Define ϕi=∑ij=0edj.
lemma 1. The expected maximum depth, E[maxidi] is at most eHn−1.
Proof. The maximum depth is at most lnϕn−1.
To finish we show E[lnϕn−1]≤eHn−1.
i≥1ϕi−1ϕi
E[ϕi|ϕi−1]=ϕi−1+E[edi]=ϕi−1+eiϕi−1=(1+ei)ϕi−1.
By induction it follows that
E[ϕn−1]=∏n−1i=1(1+ei)<∏n−1i=1exp(ei)=exp(eHn−1).
So, by the concavity of the logarithm,
E[lnϕn−1]≤lnE[ϕn−1]<lnexp(eHn−1)=eHn−1. □
Here is the tail bound:
lemma 2. Fix any c≥0. Then Pr[maxidi]≥eHn−1+c is at most exp(−c).
Proof. By inspection of ϕ, and the Markov bound, the probability in question is at most
Pr[ϕn−1≥exp(eHn−1+c)]≤E[ϕn−1]exp(eHn−1+c).
From the proof of Lemma 1,
E[ϕn−1]≤exp(eHn−1). Substituting this into the right-hand side above completes the proof.
□
As for a lower bound, I think a lower bound of (e−1)Hn−O(1) follows pretty easily by considering maxidi≥lnϕt−lnn. But... [EDIT: spoke too soon]
It doesn't seem so easy to show the tight lower bound, of (1−o(1))eHn...