My take is that those that are truly successful in CompSci have both a love of the utility that computers have AND the escapism that they enable through games and play generally.
Back when I was 8 or so, I had my first exposure to video games at the hotel we happened to be staying at in Anaheim outside of Disneyland. Asteroids, mostly. I was hooked. Within the next 2 years, I'd found a way to buy my first computer (a used Tandy Model 1 with tape drive--yep, like I said, I'm old). I whiled away my afternoons loading games off of tape and editing their sources to figure out how to cheat at them.
Later I traded up to a Commodore 64, then a Mac SE, and HyperCard got me through high school (along with a very early Casio graphing calculator). I wrote little games, demos, and all kinds of other mostly-worthless junk in BASIC and HyperTalk. I wasn't a particularly great student (in particular, I was spectacularly lazy), but I got a fair start learning the first three of what I call the five basic CS topics:
By the end of HS:
1. Substitution - Using variables in place of concrete values
2. Iteration - things like for...next loops
3. Problem Decomposition - breaking things down into component parts like functions/subs/whatever
Not until later:
4. Object-Orientation - binding data with its associated configuration (aka code to everyone else)
5. Recursion - writing routines that call themselves and enable decent into hierarchies
(Feel free to argue whether things like algorithm analysis, data structures, state machines, and whatnot are separate or fall into these categories--the reader obviously knows how I feel about it)
So by the time I was done with high school, and almost entirely without any kind of formal training, I was decently grounded in 1-3 mostly on my love of video games as a motivator. Soon after, however, my ridiculous lazy streak kicked in, and you really can't get to advanced topics while being profoundly lazy. I got to university and had my proverbial ass handed to me--brick walled on differential equations, too lazy to write anything of any substance, and what killed me utterly was that it was clear I had no clue how to sell my ideas to others and make them a reality (thanks to the Intel internal bureaucracy for that).
"I like computers, but if I don't play games, then computers become entirely work and then I won't like computers anymore."
It wasn't just about productivity and efficiency anymore, and it wasn't about being a timesuck and an escape, either; it was about maintaining moderate motivation--to love computers for both their own utility AND for the entertainment value of loving a good hack and getting sucked into a different world. Both, not either by themselves, and they're not mutually-exclusive.
I eventually restarted my college career and graduated with my CompSci bachelor's from San Jose State in 2008--16 years after starting at Arizona State in 1992. But I was motivated by BOTH a love of computers for their utility AND that escape into another world. At the end, I was named SJSU CompSci's 2008 Graduating Senior of the Year, was a Software Engineer for a while, and now run a team of incredible software engineers. And I still hack code almost daily (silly management responsibilities get in the way sometimes, though).
What's the story, then? Motivation in moderation. If you see computers too much as a tool, you can't maintain your motivation on will alone (though it will take you decently far). If you see computers entirely as entertainment, you can't maintain your motivation on that either. The trick is to balance the two and maintain your love of tech from both angles.