engineering, path-breaking, and (not) being okay with ignorance

Someone whose judgment I respect told me that some of the most closed-minded and ignorant people she ever met were engineers. This doesn't fit my general picture of engineering, but I hadn't made any such catalogue of engineers.

Then another friend said that his impression of engineering students, as a physics student himself, was that they wanted a rule-book so they could mindlessly crank out solutions. This also didn't seem fair to me. It seems like a more direct attack on engineering, too, because the context was specifically academic. Since both these friends rarely say anything not worth pondering, I started thinking about the claim that engineering produces ignorance.

Let's tackle this in the form of a disputed inquiry:

to resolve: whether engineering produces ignorance

YES

It seems that engineering does produce ignorance.

  1. Many engineers are ignorant and closed-minded.
  2. Further, engineering is not a basic science, and therefore it by nature acts without knowledge.
  3. People I respect say so.*
    * I mean this non-facetiously. Given a question whose answer you don't know, and conceding that your mind is finite, and that other minds are different from yours, the opinion of people you respect should affect your thought.

NO

It seems that engineering does not produce ignorance.

  1. Engineering is just craft magnified via mathematics; and
    1. craft always takes feedback into account, thereby removing error, and
    2. mathematics doesn't, in principle, cause ignorance (though neither does it, in principle, guarantee knowledge, unless you're a non-fallibilist Platonist blah blah).
  2. Most engineers I know are far, far more logical thinkers than most non-engineers I know.

resolution

I'm tempted to dismiss the empirical arguments on 'small n' grounds. But that isn't really honest. And I see the picture they are painting.

reward: learning like an ignoramus

As a software engineer, I often kick myself for letting my thought get stuck in a rut. Path-dependency really is a risk of syllogistic thought. The effect is that I often tentatively conclude things that, in hindsight, I realize were not only demonstrably false, and knowably so given the data originally available, but also embarrassingly simplistic or self-indulgent. Level of idiocy and revelation of mental model are both rightly mockable.

Commonly-applied, generally-lauded ways out of these syllogistic blinders are completely magical:

  1. Put the problem aside and 'let the subconscious work on it'. Some kind of CUDA-ish, wild-hypothesizing, thread-tangling magic smushes everything together and for some reason new ideas come out.
  2. Talk to someone else about it, whether or not they know anything about it ('rubber ducking'). Why would adding an ignorant person to a thought-process produce knowledge? Why would the social effect of doing so ('your mind gets looser' or some magic like that) result often enough in correct reasoning, when random perturbations of a path-exploration are far more likely to move away from than toward the solution?

Note that these techniques are not just for hacking. Yes, systematic problem-space searches should be attempted before these kinds of flails. But these 'softening' approaches help working mathematicians too; and while a problem often involves setting so many context variables so precisely that much solvent precision is lost when turning to a different sub-process, mathematical discovery is (in my experience, and from my reading of others' experiences) often precisely the result of successfully weaving the positive, constructive syllogistic and the ignorance-founded, mad-parallel reverie modes together.*
* Anyway this is how it feels when I co-juggle 'mere problem-solving' and 'legit discovery' modes when working through mathematical exercises.

This mind-opening magic is one of the things that lets skill (tekhne) drag you farther than your mind can go on its own.

risk: being okay with thinking like an ignoramus

The general aim of 'legit' knowledge (say, Platonic episteme), as distinguished from 'mere hacking', is to be unhappy with this kind of unproven thought-leverage, and not to accept a theorem until demonstrated in chunks that don't require non-obvious intuitive jumps.

The obvious practical problem with this requirement is that a 'non-obvious intuitive jump' to one person (say, non-Ramanujan) may be supremely obvious to another person (say, Ramanujan). This leaves knowledge as something dependent on the subjective intelligibility of its proof...which seems obviously idiotic.

Leaving that problem aside...

The mental-habit problem with the 'magic' approach is that it trains you to become slow to reject 'maybe...' claims. You need to get slow at rejecting the maybe-claims in order to allow the fuzz-magic to work. But at the moment you haven't yet rejected the maybe-claim, the maybe-claim may happen to be true or happen to be way, nuttily far off.

The patience toward finiteness that the engineer develops, in other words, is indifferent to the discoverability of the thought-regions outside the engineer's present mind-limits.

That is, there isn't any distinction with respect to maybeness between

As far as I know, there is a huge difference between the truth, and even the verifiability, of those two claims. Humility! I'm an ignoramus! Nothing is truly knowable!

But if I'm not okay with staying in the 'maybe...' state, then I will work to distinguish.

Of course...sometimes finite-you just don't have time to stress out over finding the true, conceptual root cause of an error/bug/false claim. Every time you 'move on' (say, after locating the line of code but not formally identifying the thought-error), you tacitly concede that you're slightly okay with doing so. This is often the best choice. But we should be aware of the mental-habitual damage, and I think we should do some palliative mental exercises against it.

...What exercises, you say? ...a Project Euler problem or two...? Yes, absolutely, spot on. Do that after you move on from a stupid bug you had to just fix. But those are unusually circumscribed (because they are mathematical), and we need some broader mental medicine for the (valuable, leveraged) being-okay with 'maybe'...