I didn't say anything about a char-op build. My basic question is still unanswered: Didn't you have to learn what builds are better than others? How did you learn that 14 is too low for a primary stat? I'm betting you learned it organically, through trial and error. Why shouldn't he (our now theoretical player given the lasted update by the OP) get the chance to experiment? To try? To fail? And to later succeed? Doesn't success after failure make you feel even better than just success alone?
Let me approach this yet another way: the very process we are talking about is optimizing. Optimizing is taking something that works and making it better. By its very definition it is not a task done by a novice. When you are teaching something to someone you don't start with an advanced topic first. You start with the basics and work your way up to the advanced topics at the student's pace of progress.
You are starting with an advanced topic. He has no frame of reference with which to compare optimized versus non-optimized. You might hand wave the optimizations but once you do that you are not instructing, you are dictating.
I suspect we will hit the agree to disagree wall shortly.
All the more reason to give help him create a decent build first. Figuring out why his character doesn't work well might not be obvious. I would think that he would learn better with someone to explain what works and what doesn't. Playing a suboptimal character to my mind is an advanced topic. The basic lessons would be how to play a decent character well. If the character isn't great, then its much harder to see if the problem is the build, his tactic, the DM screwing him over, or 4e just being a sucky game.
In learning, there is a concept called scaffolding. A common example is the keyhole essay, taught at around the junior high level. Keyhole essays are not a great format, but they are an adequate format. Before you can learn to do more complicated writing styles, you need to learn how to do the simpler styles to see how they work.
Help building a decent character in 4e, to my mind, would be a great way to learn the system. There's a lot to figure out in 4e, especially for any class that marks. But with a suboptimal character, what you learn from play won't necessarily self-teach you how to either to build a great character or how combat works. There are too many variables.
Essentially what I see you arguing is that having a teacher is necessarily worse than learning on your own. That doesn't strike me as a particularly true statement.
As to how I learned stat allocation in 4e, well, to me it was mostly obvious. At least the basic distribution was reasonably obvious (put your highest number in your key stat, second highest in your secondary stat, etc). What is absolutely best, I have no idea.
Now, you seem to be saying that I don't want an inexperienced player to be able to experiment or make mistakes. I don't see anything like that in my statements. I clearly think that just handing him a character sheet and then telling him what to do on every turn would be a bad idea. Yes, he should be able to experiment. What I am saying that a guiding light showing what most other people have found to be a pretty good idea is more valuable than just tossing the player to the dark to let him discover every single trap for himself.