We study certain combinatorial aspects of list-decoding, motivated by the exponential gap between the known upper bound (of $O(1/γ)$) and lower bound (of $Ω_p(\log (1/γ))$) for the list-size needed to decode up to radius $p$ with rate $γ$ away from capacity, i.e., $1-\h(p)-γ$ (here $p\in (0,1/2)$ and $γ> 0$). Our main result is the following: We prove that in any binary code $C \subseteq \{0,1\}^n$ of rate $1-\h(p)-γ$, there must exist a set $\mathcal{L} \subset C$ of $Ω_p(1/\sqrtγ)$ codewords such that the average distance of the points in $\mathcal{L}$ from their centroid is at most $pn$. In other words, there must exist $Ω_p(1/\sqrtγ)$ codewords with low "average radius." The standard notion of list-decoding corresponds to working with the maximum distance of a collection of codewords from a center instead of average distance. The average-radius form is in itself quite natural and is implied by the classical Johnson bound. The remaining results concern the standard notion of list-decoding, and help clarify the combinatorial landscape of list-decoding: 1. We give a short simple proof, over all fixed alphabets, of the above-mentioned $Ω_p(\log (γ))$ lower bound. Earlier, this bound followed from a complicated, more general result of Blinovsky. 2. We show that one {\em cannot} improve the $Ω_p(\log (1/γ))$ lower bound via techniques based on identifying the zero-rate regime for list decoding of constant-weight codes. 3. We show a "reverse connection" showing that constant-weight codes for list decoding imply general codes for list decoding with higher rate. 4. We give simple second moment based proofs of tight (up to constant factors) lower bounds on the list-size needed for list decoding random codes and random linear codes from errors as well as erasures.