It's not bad advice to suggest that developers be stupid if they are the type to learn from their mistakes. If they are not the type to learn from their mistakes, why are we worried about them?
Being given incorrect information, and making mistakes, and not knowing you've made mistakes, is not a useful learning technique.
And while these people are making mistakes and learning from them their customers are suffering buggy code and everything that goes with that - data loss and insecurity at the extreme end and unexpected behaviour at the mild end.
W3Schools has pointed out that users take a risk with its content, and it made no claim over the quality of its code. The only element they have authority over is its clarity, and that is what's important. The information may be lacking, but I stick by my guns when I say that crap code can still make some concept that a tutorial revolves around clearer. If other elements are not relevant to the tutorial, I do not believe it is fair to blame the author for excluding them.
That's awful advice for beginners, who have no idea why some things are good and others are dumb.
Some people will validate their pages. Some of those people will know what the errors mean, but decide that it's okay to have those errors. Other people will get a bunch of errors, and not really understand what's going on, and will just publish anyway because "it works in the browsers I tested on".