Besides of course sexual selection, considering in this section only "formal learning" activities.

Consider e.g. the 2020 University of Oxford, where many many people are taking courses without any laboratory work (and also without much use at all) like literature and history, and they are paying about 9k pounds/year for it: how much it costs to study at the University of Oxford?.

Basically all of this could be done online from books.

Laboratories are impossible however, because expendables of every experiment you do cost from hundreds to thousands of dollars, not to mention crazy upfront equipment costs.

For this reason, the brick and mortar aspect universities should focus exclusively on laboratories, and ensuring that the students with the most relevant knowledge (which can be readily obtained online) get access to those laboratories. Students should of course fully master every aspect of theory pertinent to their experiments. principal investigators should hand pick whichever criteria they want to select their students, possibly based partly on exam as a service if they find it a useful metric.

Furthermore, the use of laboratories should put great focus on novel research. A lot of laboratory instruction could be done from video of an experiments. As much as possible, we should use laboratories for novel research. Related: Section "Videos of all key physics experiments".

This is the actual main function of university for many people as of the 2020s. And it fulfills it quite well. A breeding ground.

In a closely related sense, university is simply a symbol of personal status. Not a place where you go to learn. And especially in the Anglophone world of fancy colleges, university also doubles down as a form of long term luxury hotel. Even if it ends up meaning debt.

There's nothing wrong with sexual selection. This type of natural eugenics is an important part of humankind. It is however just sad that any type of learning falls so much behind. A close second would be fine. But as it stands, it is just too far off.

Related:

For a critique/history of this insanity, see also: Section "Colleges of the University of Oxford".

Ciro Santilli's general feeling is that university should not own IP, it should belong to the researchers. Instead, university should help researchers make their startups, so they can become big, and then we can tax them and reinvest in the universities.

Of course, this goes through the nonprofit impact measurement difficulty. Maybe we could instead limit the IP to some reasonably small percentage, like 10%?

But still, as of 2020, if feels like universities are way too greedy.

These are the most evil examinations society has.

They mean that until you are 18, you have to study a bunch of generic crap you hate just to get into university. Rather than studying whatever it is that you truly love to become a God at it as fast as possible and have any chance of advancing the field.

And then, if you decide that you want to change, which is not unlikely since you haven't really try to study what you signed up for before then, it can be very hard and time consuming, leading to a bunch of adults with useless degress they will never use at work.

With the invention of the Internet, all teaching material can be free and open source. Only laboratory space has any cost (besides the opportunity cost of participation in actual projects in a research team).

Per-country list: en.wikipedia.org/wiki/List_of_admission_tests_to_colleges_and_universities

Where nuclear weapons and nuclear power, and a ton of derived research is made.

For a fun and brief random software encounter with that universe, see the VisIt section of stackoverflow.com/questions/5854515/interactive-large-plot-with-20-million-sample-points-and-gigabytes-of-data/55967461#55967461.

The summary from www.geeksforgeeks.org/tree-traversals-inorder-preorder-and-postorder/ is a winner:

In principle one could talk about tree traversal of unordered trees as a number of possible traversals without a fixed order. But we won't consider that under this section, only deterministic ordered tree traversals.

There's a billion simple looking expressions which are not known to be transcendental numbers or not. It's cute simple to state but hard to prove at its best.

Open as of 2020:

Bibliography: