By which, he meant the whole TED format, and the format of his own talk. He goes on to draw parallels between that format, the current education system, and the "mainstream" media's failing model. On which topic, see Belmont Club's post about schools trying to ban laptops in classrooms.
I agreed with this from Jarvis:
"Why shouldn't every university - every school - copy Google's 20% rule, encouraging and enabling creation and experimentation, every student expected to make a book or an opera or an algorithm or a company. Rather than showing our diplomas, shouldn't we show our portfolios of work as a far better expression of our thinking and capability? The school becomes not a factory but an incubator."
He also asks this, and here's where we diverge:
"We must stop our culture of standardized testing and standardized teaching. Fuck the SATs.* In the Google age, what is the point of teaching memorization?"
The question shouldn't be rhetorical, because there really is an answer.
The answer is that in order to fit new information in, it requires a framework. Franeworks do require problem-solving skills, but they are NOT all process. They are ALSO made up of things you know. Indeed, they depend on that, or else the framework collapses. Which means the new isn't integrated, just thrown on the wall. And there, alas, goes the value-add of perspective, and much of one's problem solving capability.
The black belt in martial arts is the beginning of real learning. Getting there takes a certain amount of rote. And of collision with unpredictable real sparring, too, so that the rote is integrated and understood.
Engineers do a lot of rote learning before they go out to apply their problem-solving skills and build bridges. The skills are 2 different sets - but I wouldn't drive over a bridge built by someone who hadn't done both sets.
Which is to say that the current model for schools, like that of newspapers, is not 100% wrong. It may be 40% wrong, or even 70% wrong, but not 100%. There's definitely something to Jeff's 20% Rule suggestion. But ditching common sets of things that educated people should know, in favor of pure process or washed-out curricula, has been tried. It has not gone well, and is not the answer.