Education is broken; people don’t learn much, it takes them a very long time to learn it, they forget most of it, and then they don’t generalize. It’s tempting to attribute this to people being stupid, and clearly that’s a condition of the problem space, but it’s no excuse for not solving the problem.
The medieval lecture model we use is partly to blame. It originated as a kind of mass production for books: the reader reads from one copy of the book while the audience notes down verbatim everything they say. As manual methods of making many copies of a text go, this is quite efficient, but once the number of copies goes past 500 or so, the printing press is far more efficient.
Sadly, this model survived the introduction of the printing press and even the xerox, continuing to this day in many classes, in which the pupils waste precious attention they could spend assimilating the material on copying it instead, as if they didn’t have smartphones in their pockets capable of copying documents over Bluetooth at hundreds of pages per second.
Nowadays the goal of lectures is not really making copies of written material, but transferring knowledge from the mind of the lecturer into the minds of the students, at its best leavened with some entertainment or practical demonstrations to clarify the abstract concepts. But it is a terribly inefficient way of achieving this kind of knowledge transfer; it’s roughly as effective as reading a textbook [citation needed] but consumes the time of the lecturer as well as imposing expensive coordination costs on the lecturer and students, who must not only arrive at the lecture hall together but also pause to think together if they are to have a chance of understanding the material.
(In British universities, the rank below Professor, equivalent to US “professor”, is in fact “Reader”, followed by “Lecturer”, which is a mangled French word that also means “reader”.)
As a means of inspiring enthusiasm, however, lectures are almost unmatched; only music, movies, sports, and blood rituals seem to be able to compete, and of these all except music are much more costly. So lecturers often opt for lectures in order to manipulate their audiences emotionally, usually to further the interests of the lecturer (for example, persuade colleagues that their research interests are worthwhile, or to whip up political sentiment) but occasionally in the interests of the audience.
Sadly, nowadays our lectures are quite often afflicted with slide decks full of text. While these are of some communicational benefit when a language barrier separates the lecturer from their audience, far more often, they further reduce the bandwidth of information transfer while vitiating the lecture’s most signal virtue, that of inflaming the passions.
Learning depends, of course, on many other factors beyond knowledge transfer. For example, you need problem sets to guide students to construct their knowledge, you need some kind of feedback loop to make sure that the students aren’t learning errors as fact, you need spaced practice to prevent the loss of knowledge to disuse, and you need motivation to interest the students in the knowledge. But in what follows I’m focusing on purely the knowledge-transfer component of the education process.
Bligh's book What’s the Use of Lectures? and Armstrong’s Natural Learning in Higher Education go into more detail on what lectures are good for and how to promote better lectures. But I come not to praise lectures but to damn them, to damn them to Hell forever.
This is because reading a textbook and attending a lecture are far from the fastest or most thorough means of learning. In numerous controlled studies students progress about two standard deviations faster with a dedicated one-on-one tutor than with traditional classes. (I think this means about 30% faster.)
You could argue, though, that one-on-one tutoring is too expensive, and can only be justified when the goal of knowledge distribution is light (i.e. it is desired for a small number of students to learn the material) or the benefits of transferring the knowledge are so great that it justifies occupying the time of a large number of skilled tutors who could be practicing the knowledge rather than transferring it. Bloom does in fact argue this: “...more practical and realistic conditions than the one-to-one tutoring, which is too costly for most societies to bear on a large scale.”
However, this is not in fact the case. Most lecture classes only lecture to about 24 listeners, the majority of whom (let’s say ⅔) do not learn the material well enough to teach it. Of those who do, almost none are allowed to teach the material for some three or four years.
If we track the number of people to whom the knowledge has been transferred over time, it can grow exponentially in this model, but it works out to a doubling time of a year or a bit more.
By contrast, in the “one-room schoolhouse” model, the teacher mostly teaches the most advanced students, who teach the less advanced students, who teach those who are less advanced still. The process of teaching what you have recently learned serves both as a check to unmask the pernicious fluency illusion and as additional practice time. Typically in the schoolhouse, according to rumor, this also achieved a doubling time of around a year.
But we could do much better; the crucial question is how short a time you have between mastering a lesson and having to teach it to someone else. If, for example, you study a lesson one week, balanced half-and-half between working with a tutor and doing exercises on your own, and act as a tutor in that lesson to two other people in the next week, then the doubling time could be under a week. (During the first week you were presumably teaching the previous lesson to two other people during the hours that you weren’t studying the new lesson.)
Even if one-on-one tutoring were only as effective as attending lectures, this model would still defeat it; only four weeks after tutoring the two initial students, you’d have 2+4+8+16 = 30 students already up to speed on it.
Bloom points out (in his paper on the “2 Sigma Problem” I mentioned above) that tutoring also helps with focus (students spent 90% of time on task instead of 65%) and with positive attitude.