The main drive of Starcraft's success is likely marketing, polish, and history. However, Starcraft does succeeded in a fair few areas and it is worth taking a look at. Unfortunately many of the things that Starcraft does well are inseparable from the things I'd expect StStephenHawking
(and others) to dislike about it.
I've been disappointed that Starcraft focuses on "playing against the UI", but it has advantages. Activities with short feedback loops, or that can be repeated without depending too much on the actions of other players, are relatively easy to practise and are good sources of a sense of achievement. Executing build orders, macro (micromanaging your economy), and micro (micromanaging your army in a battle) are all activities that fit this description. Games of Starcraft have relatively long periods of no direct interaction between players. This gives players the opportunity to hone their solo-skills. I'm not saying improving at Starcraft is easy, it just seems more straightforward than improving at Zero-K.
Zero-K removes many of the unambiguous solo-skills in the name of letting players "fight their opponent, not the UI". Good and bad builds still exist in Zero-K, but the difference between them tends to be in decision making instead of execution. It is relatively easy to detect faults in the execution of a build order in Starcraft. It is much harder to see why a decision is bad, and I think this makes it harder to feel a sense of progress. It seems like Starcraft's focus on execution helps players fell like they know how to improve.
Once you look for them, there are quite a few things that make it harder to know how to improve at Zero-K.
Starcraft games are more discrete. In Zero-K you should maintain a constant rate of expansion and, at higher levels, constantly scout, poke, and raid your opponent. Starcraft is more about expanding in bursts and short decisive battles.
Compared to Zero-K, Starcraft maps are basically all the same. The starting bases are especially well standardised. This allows people to pick the same race and build every game, and to better compare their performance between maps.
An upfront payment economy punishes players that mess up much faster than a flow economy. Once you have 1000 minerals you know that you did something wrong. It takes longer to excess in Zero-K and it is often harder to pinpoint why.
The units in Starcraft are more responsive than the units of Zero-K. Of course, this is by design because smart units with instant turn rates and un-dodgable projectiles would be degenerate and a lack of smartness results in the kind of micro that Zero-K aims to avoid. The fact remains that units do exactly what you tell them to in Starcraft, while in Zero-K they often have to turn and negotiate other limitations of their movement physics.
Here is the essence of the control schemes of the two games.
Starcraft - Give players simple commands, but highly responsive, commands. The intended challenge is the communication of complex maneuvers (eg. spliting, kiting) under the bandwidth restriction of the commands. Units are excellent at executing the simple commands promptly.
Zero-K - Give players a wide array of powerful commands, but restrict the units ability to act on them. The intended challenge is in deciding what to do under the restriction of the unit physics.
Obviously Starcraft has decisions and Zero-K has challenging micro. Starcraft also has unit physics, but it rarely gets in the way of a players desire to make a unit walk somewhere or shoot a target.
I think there is a tradeoff between responsiveness and command complexity. Applying the extremes of both doesn't seem like it would work well. The various infinite-APM Starcraft AIs show that unit behaviour can look pretty degenerate when units are both highly responsive and able to recieve complex commands. At the other end, unresponsive games with simple commands just feel bad to play (in my experience).
This is what extremely responsive units can do with perfect control:
The commands are even slightly more responsive in Starcraft. Zero-K's commands are issued when the mouse button is released, as almost any command may be dragged to issue a line or area version of that command. Starcraft has no line or area commands so is able to issue its commands when the mouse is pressed. I'd be interested to know the effect of this. Perhaps some Starcraft (or other RTS) players were turned off Zero-K because they subconsciously perceived the sort of command lag inherent in this distinction between the two games.
Starcraft has much simpler units than Zero-K. This helps responsiveness and possibly makes the basics of the unit easier to grasp. For example, they lack turning circles, occlusion and the possibility of missing. They are more complex in terms of non-physical attributes such as build cost and armour class, but this complexity is somewhat hidden from new players. The thing that new players are first exposed to, the unit physics, is much simpler.
The physics of Zero-K units contains pitfalls that experienced players learn to avoid with good command habits. These pitfalls tend to be required to make the unit behave correctly, so they are hard to remove. For example, if you point-move some units and tell them to attack something the units at the back will be unable to fire. This causes the unit clump to push itself into the enemy forces, making it weak to AoE and deal barely any damage. This trap comes from the fairly important mechanic of allied projectile occlusion.
Starcraft units are also simpler in the sense that they can only do one thing at a time. Most Zero-K units can do two things at once: move and fire. Consider some of the reasons that Starcraft units tend to have less trouble with target prioritisation.
They hold fire while moving, so it is relatively easy to have them move and then instantly fire at the desired target.
Their burst rates are relatively high. No 20s Lances in Starcraft.
They always hit (unless things such as blink occur).
They fire in strict circles, with no occlusion or relativistic effects.