That's when those introductory laboratory classes become nothing more than "training" (which is what Dude was referring to, in my opinion-- since that kind of environment is little more than a mindless "follow the directions. Bleat after step 2," kind of 'learning' environment). Note that I drew a distinction between learning and training.
And even when it's not a procedural lab, but a troubleshooting lab, you know ahead of time that whatever problem you're going to encounter is one the teacher has already provided you the tools to resolve. For example, if it's a software issue, you know a significant clue to the nature of the problem is going to show up in a message somewhere, and it's somewhere you've been taught to look. In the real world, you may find yourself supporting software that was poorly engineered, where many error messages, if they're cut at all, are not meaningful.
The problems are also limited in scope. If I'm learning server administration, I know that whatever problem is occurring in the lab, it's related to server configuration. In the real world, I'm more likely to spend time troubleshooting issues that are caused by bad programming, user error, network changes, security issues, hardware failures, or integration problems with another system. In other words, most of the problems I spend time on are not server configuration problems.
Naturally, I've seen many, many problems occur in troubleshooting labs apart from the ones the teacher had set up. The normal reaction is, as you said, to steer the students away from the problem, because it interferes with class time and the purpose of the lab. Anyway, the student usually doesn't have the skills, access, and/or big-picture process and design knowledge to resolve the unexpected problem.
I do concede that amylou had a good point about research labs. They're turning students loose on real world problems, which is totally awesome. But this conversation was about a child who wanted to create games, and there aren't research labs at universities doing that. I think a lot of university presidents would balk at the idea of student teams writing games using university resources. There's no federal grant program tied to it, and securing grants seems to be the point of most research universities, at least from an administration perspective.
