I also get asked an awful lot (by the younger years) how I type so fast and how they can "learn" to type that fast. Type. For years. Bang, you've learned. This is no shortcut, there is little technique, no amount of learning the home keys will help you type fast. You just have to type, lots, all the time.
While this very true; it helps to also give them a place that requires speed to push them to type fast. I've always told people that asked "how do you type so fast?" that I learned to type really quickly by growing up in IRC chat rooms with lots of people and multiple conversations going on at once. You had to learn to type fast to keep up with what was going on.
Same for coding. You can learn some theory. But to learn to code, you have to code. And with kids it's really easy - pick a game, program it.
The only problem I've ever had with using "games" as a way to learn to code is that the final product may not match expectations. To put it another way. I love programming because it gives me a means to solve problems. Sometimes the problems are concrete as "I need a piece of software on my desktop to tell me when I'm getting a phone call on my phone." That problem is focused, the solution is focused too. If your phone rings and you get a notification on your computer, you know you solved the problem.
Games rarely offer up focused problems and solutions, especially for beginning programmers. A lot of game ideas are nothing more than "I want to make an RPG where I fight zombies." The solution would deceptively be to have a few characters in a bland world and some monsters labeled zombies, but game dev is never that simple and the problem space "grows." It goes from "rpg zombie game" to "rpg zombie hunting game where I must build a cure, save cities, and all while I'm working within this cool battle system." Games could be a great route to code but the path between problem definition and solution is huge compared to more simple stuff.