# Why don't textbooks explain proofs' discovery?

This question concerns only proven statements. I don't know if research papers do, but most math textbooks don't. Counterarguments:

1. Space?
1.1. The increased length from explaining the discovery is justified; if proofs' discoveries aren't explained and revealed, how can they be learned and discovered, especially to solve unsolved questions?
1.2. The discovery may be more vital than the proof: easier is deducing the latter from the former, rather than searching the latter for the former. So if space is a problem, why not focus on the discovery and leave the proof to the reader, instead of vice versa that is the status quo?

2. Ignorance of the discovery? Maybe these flashes of genius cannot be described? This seems unlikely for statements proven long ago.

3. Desire for students to toil, and painstakingly discover the discovery themselves? But this is inefficient; if certain theories expended centuries for a famous mathematician, how can a math student discover within a day?

4. Suppression of knowledge? This is least likely, and paranoid; maybe authors don't wish students to learn the discoveries easily, because then future students would outperform them?

E.g. Cauchy-Schwarz Inequality is often proven without its discovery explained. Textbooks don't explain how a student would divine $\bf{0} \le \|\bf{u} - \cfrac{\langle u, v \rangle}{\|v\|^2}\bf{v}\|^2$.

• I originally asked the above here, but maybe it suits this SE more? – Greek - Area 51 Proposal Aug 10 '17 at 1:47
• I've never seen a book NOT give Cauchy-Schwarz as an exercise. – PVAL Aug 10 '17 at 11:55
• Hard to believe that nobody has yet cited the work of Imre Lakatos, who described mathematics as a practice that erases its own tracks. – mweiss Aug 11 '17 at 3:35
• After reading your point 4, I suspect that you want to see hints, cruxes, or ideas for proofs, not real histories of their discoveries. These are written sometimes. – beroal Aug 27 '17 at 11:55
• @beroal That also, yes. – Greek - Area 51 Proposal Sep 1 '17 at 5:51

In many cases, standard theorems in a theory were first proved as special cases before the theory itself was even invented. For example, Lagrange's theorem in group theory predated the invention of group theory. He proved it as a result involving certain sets of permutations (of variables in polynomial equations) which were only much later recognized to form a group in the modern sense. A book in abstract algebra will typically begin with the axioms of group theory -- axioms which were not clearly spelled out in Lagrange's life time. To explain what Lagrange himself actually did you would need a long detour into the theory of polynomial equations, one which would have only an indirect link to the modern result known as Lagrange's theorem (which was only proved decades after Lagrange's original result). This is only a single, somewhat preliminary result in group theory. If you tried to do it for all results in group theory textbooks would blossom into thousand-page tomes that would lose almost all students.

There are some text books that have a somewhat narrow focus and which do have a more historical approach. For example, Harold Edward's Galois Theory does get into what Lagrange actually did. While such books are valuable, and all math students could benefit from at least occasionally working through such things, they aren't a very good way to get a good understanding of the modern theory. Mathematics is hard enough without overburdening its exposition with excessive history.

I'm not exactly sure if I understand what you are asking, but I will try to give at least a partial answer. I think there are a couple reasons why an explanation, of how a proof was first arrived at, is not given:

1. It isn't known

For a lot of proofs not even the author can exactly explain how he arrived at the given proof. "Well I thought about how to prove the problem, thought about what my intuition said, tried to make that rigorous, found the places where I couldn't quite get enough rigor, played around with some other ideas, and finally came up with the proof when I realized that I could just reduce this to a coloring problem which was much easier to solve." And that's one of the more specific examples. Often people spends weeks or months on a problem trying many dead ends before they come up with the correct approach or as happened to me figure out that they are trying to prove the wrong thing and prove the exact opposite.

1. The original proof was quite ugly

This is very often the case for many harder problems. When you first work on a hard problem you might be starting a new direction. The example that comes to my mind is the proof of independence of CH. The original Cohen paper is quite hard to read since when he did it, the technique he invented wasn't formalized and smoothed. He and others went on to refine the technique into something that almost a baby can use (for the easy problems anyway), but this took some time. Explaining how he came up with the original proof would probably be hard and wouldn't really help much.

1. It is time consuming

Explaining how a proof was first arrived at will either be very time consuming (explaining all the dead ends, ideas that didn't work, refinements etc.) or will leave out 3/4 of the important stuff (all the dead ends, ideas that didn't work, refinements etc.). Usually in Mathematics classes (especially at the lower levels) students have more than enough trouble just grasping the refined and cleaned up proofs and ideas, much less being able to follow the complex process that led to their creation.

1. We don't know/ lack of rigor of the original proofs

A lot of especially the original proofs are old. Cauchy lived in the late 18th early 19th century and I doubt he wrote down how he came up with the proof. Further more though quite a bit of older mathematics wasn't proved to current standards. Some of Newton's and Leibniz's results were such and while the results hold in essence the actual founding theory which makes them work nicely wasn't created until much later.

• I like your point 2.) Also, even when the original proof isn't ugly per se it might be less elegant than later proofs, and elegant proofs are typically easier to digest. Also, original proofs sometimes have desideratum that are different from later proofs. Abraham Robinson had to justify nonstandard analysis, so he necessarily used a lot of model-theoretic machinery. Now that it is accepted, it isn't so necessary to belabor the point. Thus modern expositions tend to adopt a more axiomatic approach in the safe knowledge that it can all be justified. – John Coleman Aug 10 '17 at 14:19
• As pertains to 2 ["Explaining how (Cohen) came up with the original proof would probably be hard..." @JohnColeman as well] you might check out my answer to MO 159935 and its links to a question I asked at MO 124011 and paper found [without paywall] here. – Benjamin Dickman Aug 13 '17 at 16:29

In addition to other good points already made, there is also the mythology that there are "ultimate" proofs, and that the messy human processes that occur along the way to development of "the correct" proof are best ignored... insofar as those human activities are not the "platonic" ideal of mathematics. Timeless, etc.

Similarly, many peoples' affection for "formality" of presentation would not allow them to give examples prior to a definition (which was, in reality, only conceived long after the examples were well understood, and motivated the definition). In my opinion, this is a stylistic conceit that does not help people understand mathematics.

Said in another way, there is an extreme "Whiggishness" in accounts of mathematics, as though (cf. "Whig account of history") people in the past were pathetic and benighted, and everything has been improving toward an ideal, which is represented most closely by the current situation. If we buy that, then there's not much interest in examining the sad, ignorant flailing-about of old-timers like Newton, Leibniz, the Bernoullis, Euler, Lagrange, and so on. :)

More seriously, the sophistication required to understand the complexity of mathematical developments probably is not available to most 18-year-olds, or 22-year-olds (in grad school). So maybe a bad investment of time and effort?