Starting with a fundamental misconception, complex skill checks and skill challenges are not even close to being the same thing. A "Complex Diplomacy Check" involves rolling the diplomacy skill half a dozen times to find whether you're diplomatic enough, and a complex balance check involves rolling Handle Animal half a dozen times to find out if you've succeeded at training your animal. It's intentionally rolling the same check multiple times to find out if the orthodox skill is good enough Literally the only thing that has in common with a Skill Challenge is an n successes before 3 failures mechanic.

A skill challenge is a scene framing and resolution structure. It still has the three failures mechanic (as long as you've accepted the errata) - but a skill challenge is intended to take multiple people doing what they are good at in order to further the plan. So the wizard does the research on the castle (history), the bard walks up to distract the guards (diplomacy), the fighter gets himself and the rogue over the wall (athletics), and the rogue sneaks in, defeats the lock (thievery), and gets filches the parchment without being spotted (stealth). And if something goes wrong the bard can try bluffing the guards they haven't heard anything (bluff), the fighter can spook the horses to keep the guards busy (nature), the wizard can provide a distraction or back the bard's distraction (arcana), or the rogue or fighter can try to convince people to keep quiet (bluff or intimidate).

This is an entire plan handled with the one mechanic, including the sort of challenge level that will be fun. A world away from the canonnical animal trianing complex skill check in which you use (handle animal), (handle animal), (handle animal), (handle animal), (handle animal), (handle animal), and for variety ... (handle animal) (and if something goes wrong you use, guess what? (handle animal)). And what do you do with that skill check? You handle an animal.

And how do you know which skills to call for in an improvised skill challenge? You let the players tell you. They tell you what they are doing, you tell them which skill to roll. And if they aren't doing anything too useful they don't roll a skill.

As for your example of making the dragon scene up, it leaves me cold compared to what I actually did. Which was again let the PCs lead. As I remember it involved them using streetwise to find a suitable cart, bluff (which covers disguise in 4e) to disguise the dragon properly, dungeoneering (which I use to cover engineering - a stupid oversight in 4e and almost as much of one in 3e although craft and profession cover some of it) because the thing was rickety (failed streetwise), history to find a quiet and overhung route, intimidate to drive back the curious kids (who were the physical manifestation of one of the rolls failing - the PCs choice to use intimidate as other social skills would have worked). A much more comprehensive scene using the talents of almost all of the PCs and in which they were able to pull things back from the brink of failure.

And what did the skill challenge rules provide me with? Framing and pacing. They let me easily work out how far each step should take the scene. You've told me how to disguise the dragon - but not take it across town. Mr. New DM following your advice has only the simplest answer to the first step of getting the dragon across town. He doesn't have a ready made scene which involves everyone, will last a fun length of time, will be the right level of challenging despite the number of moving parts, and that can inherently handle some level of PC failure.

One other point before I continue on the fire and the spoiler blocks. The 4e DMG 1 is a deeply flawed book and was put out with literally a year too little playtesting. Most of it has been errata'd - and 4e (2008) is a very different creature from 4e (2012). The half level +15 guidelines are clearly bad because of things like stat increases and feats.

I also think half the problem we're having with non-communication is that we're fundamentally looking at the game in opposite directions. You seem to want to look to the rules to tell you what the fiction should be. Me, I want the fiction, for the most part, to tell me what the rules are. Prone has a tight mapping to the fiction, therefore I don't need guidelines to tell me when an effect can knock someone prone. Hit points on the other hand are largely a product of the metagame so I can't look straight at the fiction to tell me what the numbers should be.

And this is where the making up your half-aboleth vampire comes in. I don't start with the rules at all. I start off with the concept of a half-aboleth vampire and ask what makes him special, what makes him distinct, and what he ought to be able to do. I then give him the mechanics to do that, whatever that is. (I'm now picturing a fish thing with sparkly slime just to make the idea even more horrific, and that instead of biting turns its victims skin into a transparent mucous membrane then just reaches its tentacles in).

(I'll spoiler if I go back to segment by segment).