“In essence” is doing enormous work here, and it will be basically impossible to have any kind of discussion if that work is considered acceptable.

This kind of word-twisting can be used to make pretty much anyone into a murderer, at which point “discussion” will come down to who the mob chooses to listen to.

The analogy to hiding the URL would be something like a seatbelt that only unlatched at the car's stop only in neighborhoods it considered "safe".

And, how could the fleet command be ignorant of the culture problems on this ship?

Seems like those NCO issues should have flagged long-term problems to both the ship's command and to the fleet. I don't understand why they would ignore something like that.

I get why readers don't do that. There is very little written so well that it deserves that trust. Generally we have to step back from something and ask "what does that mean? how do I reframe that into something I can understand?" Because the material's models are too poorly thought out to be useful or too poorly communicated to be discovered. Trying to adopt such models is futile.

But for me, SICP is one of those rare books where I am better off trying to absorb its frameworks, rather than constantly trying to reframe the material into something I understand.

When I step back and _think_ the way the book suggests, the problems solve very quickly. When I try to get a solution from some combination of my intuitions and known tricks, I thrash about endlessly.

It is a very carefully written presentation of profound and important ideas, and deserves to be approached differently than the vast majority of books.

So far I have found that the Wizard Book always provides the required algorithm, or enough discussion to work it out.

The "filtered cross-join" model allows students who have learned SELECT and WHERE to think of a JOIN as an extension of those primitives, a combination of two tables on which they can filter.

Venn diagrams might be useful to visualize an outcome, but they will not support stepping through primitives to a solution.

In "filtered cross-join", JOIN can be an extension of SELECT that combines two tables. The combination is a set of rows, each each of which combines all the fields of one row of one table with all the fields of one row of the second. This is easily visualized with two two-column tables of three rows.

They can then use WHERE to find those rows with matches on the key field.

With this model, students can build up JOIN as an abstraction of simpler primitives. When they are struggling with a problem, you can ask them to first step through those primitives to accumulate the solution. Say, query the "raw" join and examine a few rows to see which they want returned. What is true of those and not true of the ones you do not want? ("I want those where these two fields match, and none that do not.") Ok, how do you express that condition in SQL? ("Where . . . this equals that?" "Hmm, try that" "HEY THAT WORKED!")

With that basis, they have a model that can extend to more complexity -- joins across several tables, joins on the same table, joins with conditions other than field equality.

Building up to and using this model, you can have the vast majority of students writing joins with confidence in two days.

My suggestion to the site would be, use an example that has two columns on each table, to provide the key field on which the join will be performed.

The scholarship is just bad. The whole argument is full of mis-statements of critiqued positions.

Anyone interested in the article is better served to start reading Plato, Locke, Hume, Smith . . . .

> Professor Salaries will have an increasing Gini coefficient

I think your own model says this is too noticeable. Better strategy would be harder-to-notice changes, like smaller teaching loads, more sabbaticals, earlier retirement, travel abroad to conferences, etc.

Yes, if the tuition were reduced by increased endowment subsidy, the political cover for the endowment would not be much affected.

But that strategy would limit the value of that cover (for the university staff) by 1) ultimately limiting the size of the "politically justified" endowment and 2) limiting the overall resources they control.

Re: 1), if the university contains cost inflation below endowment retained returns, the endowment eventually grows to fund 100% of tuition. At that point, the question "why do they need more?" becomes obvious and unanswerable.

Re: 2), note that the article implies an "agency" effect with university personnel, in that they benefit by their control of the institution's resources. In companies, such effects are seen in executives taking higher than market salaries and excess perks. All that diverts income from equity owners to managers. In a university these might be above-market salaries, research budgets, job security, prestige, trips, subsidized housing, etc etc etc -- diversions of endowment returns from "teaching" to the staff.

The agency model argues that university staff benefit from _growing_ costs, which give them more resources to control. The story that education costs rise faster than inflation and middle class incomes removes the limit on the size of the cost base / resources controlled. It also "politically justifies" an ever-growing endowment.

Thus lowering tuition simply by paying ever increasing portions of tuition does not serve the university staff. The point of the article, I believe, is that those staff have an ongoing interest in _increasing_ the costs to be subsidized by the tax-free growth of their endowments.

(It is also worth noting that such arguments justify more than protection of endowment returns: state funding for public schools, federal "overhead" payments for research grants, etc. We might even wonder if the endowed universities play a "cost-setting" role for higher education generally. They operate at ever-higher cost, sustained internally by the endowment returns; those costs then help justify the budgets of unendowed schools. "Hey, if you want the kids of Michigan to have a first-rate education, this is what it costs, just look at Harvard." If so, and MIT, Harvard et al began controlling costs, those controls would then ripple over time throughout the whole system through those benchmarking effects. Thus even unendowed schools would have long-term political interest in protecting those endowment returns.)

Note that we don't have to fully believe the whole story to learn from it. The "market" for higher education market is clearly pretty weird, and explaining that will require so pretty involved stories.