Re #1: I agree. But with a pedantic semantic distinction:
There are a lot of objectively wrong ways to do something. But among all the right ways, there are trade offs.
There are a lot of objectively wrong ways to do something. But among all the right ways, there are trade offs.
There is no right or wrong, only working and not-working.
If you only need to move a 60lbs bag of sod across the yard, carrying it yourself, using a wheelbarrow, and using an F-150 Lightning all have exactly the same result and none is inferior. They can all be done in the same time by the same person with no real difference. It's only when the parameters change, like moving 400 bags of sod, that the first two require tradeoffs.
You're assuming that the "result" data structure is only a single boolean, is_task_done. That's not how most things work, GP's bubble sort will pretty damn well show up in locked-up UIs and injured-dog-slow bottlenecks, that's as observable as the Sun at 9 AM. That would happen pretty fast too - don't delude yourself with "Move Fast And Fix Later" with quadratic algorithms - 1000 squared is a million, and at 10000 it is a hundred millions.
>carrying it yourself, using a wheelbarrow, and using an F-150 Lightning all have exactly the same result and none is inferior
Off course not, the method used would show up in your health and exertion data, your fuel costs, the amount of noise and heat you emit while doing the work, etc... Every single way of observing a boolean is_bag_moved result can also be used to observe a metric measuring those things, and thus determine the method you used.
You're trying to defend an extremly strong claim : If algorithms A and B gives the same output in memory, NOTHING else matters. Here's an concrete example, consider the following sequence of algorithms:
Algorithm 0 : Print("Hello")
Algorithm 1 : do {Nothing} then Print("Hello")
Algorithm 2 : 2.times do {Nothing} then Print("Hello")
...
Algorithm N : N.times do {Nothing} then Print("Hello")
This is an infinite sequence of algorithms that all have exactly the same behaviour in a time-space agnostic manner. Would you say every algorithm in this sequence solves the proplem of printing "Hello" exactly as well as every other algorithm ? If yes... interesting. If no, then your original claim is false as-is, it needs qualifying with what observable behaviour means, as this example shows, it's a difficult notion to exactly define, a naive definition of "Everything that can be observed in computer memory or on screen" would leave you open to ridicilous conclusions.
The Perl community themselves eventually extended the motto to "There's more than one way to do it, but sometimes consistency is not a bad thing either"