As was pointed out, 2 blades has an issue with vibration during yaw.
The blades have large intertia fighting yaw when horizontal, virtually none when vertical, so the resistance to the yaw force varies with the blade position. Since there will be wind when there is yaw the blades will be spinning, and the yawing action will be jerkey and put stresses on much of the mill and its supports.
Any higher number of symmetrically-positioned blades doesn't have this problem - or doesn't have it to any significant degree.
An even number of blades also has another disadvantage over an odd number: The wind is decellerated by friction with the ground, so it's faster as you go higher. So drag force is greatest when the blade is straight up, least when straight down. With an even number of blades one is straight up and its opposite straight down for part of the cycle, and a gap between is straight up and another straight down for another part. So this produces a tilt-back torque that varies with the blade position - creating a vibration at N times the rotation rate for an N-even blade prop. With an odd number of blades each blade is opposed by a gap, rather than another blade, and most of this cancels out.
This tilt-back vibration from drag variation is largest for a two-blade and drops drastically as the number of blades increases (because the additional blades coming out of horizontal are gaining force differentials as the ones coming out of vertical are losing them). By the time you're up to six blades you're more balanced than a three-blader despite being even-numberd: A six is two threes mounted so some of any remaining vibration components cancel. Even a four isn't all that bad. (A four-blader would be perfectly balanced for this due to the sine/cosine cancelation, as it is for yaw vibration, if the wind drag varied linearly with height. Unfortunately it doesn't.)
So to avoid vibration issues you want either an odd number of blades or a large number.
From a power collection standpoint the number of blades doesn't matter much: A lift-type blade which is spinning draws power from air for a significant distance upwind and downwind (mainly from downwind) and the wider the farther. By matching the blade width to the tip speed ratio you end up with the next blade coming by just as most of the air decellerated by the previous blade has left the region from which this blade draws power. By delibeartely erring on the side of using some slower air (making the blades "too wide" so they work deeper in the windstream) rather than letting some fast air get away, your blade won't let much escape even if it is running somewhat under its ideal speed. (The energy lost to the current blade from the slowed part of the air it's working in is energy that was collected by the previous blade.)
But when you're fabricating blades, it's more work to make and mount more blades - even if they can be narrower. Thus mill builders tend to gravitate to three-blade mills on large projects: the minimum number for low vibration.
With small units the component strength tends to be great in proportion to load, making yaw vibration acceptable, and the short blades mean little vertical drag variation if it's mounted high enough for drag to be a major issue. Also it's easy to fabricate two blades from a single piece of wood (avoiding the joint, simplifying the hub, and gaining additional strength where it's needed) and a straight blade simplifies storage and transport. So you'll see some small two bladers - especially for portable applications like Wolf's truck camper.
I've occasionally thought of doing a middle-sized four blader from two continuous pieces of wood, to try to get that hub strength advangage. But you'd have to cut away half the wood for the crossover - right where you need strength the most - or end up with the two blade-pairs in different planes - perhaps requiring a tweak to their profiles for efficiency. (If you do four blades separately you lose the hub strength and simplicity advantage, so why not just do three?)
