And testing in the typical style that fails when a student can use a textbook or google makes just as little sense.
Memorization, while not sufficient, is an absolutely necessary part of learning. No amount of googling can help you become skillful at something.
Note that searching for new information or old information that you may have forgotten the specifics of is also perfectly normal, and an important skill in itself. But even there, Google will usually be more of a tool for recall (what was the order of those two parameters? What was the name of that substance that they use for liver issues?) then completely new fact finding.
For example, if I ask you what is the complexity of finding an element in a red-black tree, it would be OK to quickly Google the speicifc answer. But if you have to Google what complexity means, or what a red black tree even is, or if you're not sure whether the complexity is better or worse than searching a linked list, then it's clear you don't have working knowledge of red black trees.
So, in other words, you propose a test. Presumably it's open-book, since you are testing an ability to apply knowledge and not an ability to merely remember it.
This sounds less like a rebuttal and more like a validation.
While testing for "reciting" them is certainly suboptimal, it's cheaper to prepare a test of the sort than a test where you are effectively putting that knowledge to use.
A balanced test would do a bit of both (so it does not take ages to prepare and grade, yet ensures you know the basics and can apply them in some problems).
But I was never under the illusion that this constitutes working knowledge. I was able to come up with solutions for some problems only because I had some basic knowledge memorized, which allowed me to effectively manipulate ideas and concepts which I did know of. Even then, I used to semi-joke how the hardest thing in theoretical math is all the memorization required.
If you've ever explored any scientific field deep enough, you'll know that it's festered with tricks found by accident and roundabout ways to get to a solution that, if you haven't seen it before, you have no hope of applying in practice.
Or you might think that consulting a book or two (or Internet) will allow you to prove Fermat's theorem while you are being tested for 2h.
Very little of what I used on a day-to-day basis came from rote learning, a lot came from the deeper understanding that I got from actually actively engaging with the concepts.
I'm not proposing doing away with schooling completely, but the model should change (much in the way more progressive schools are doing anyway, just a shame that not many parents can afford to send their kids there).
The question is not how much non-rote learning is required to adequately perform any task, but whether rote learning is a necessary component of it.
By engaging with research papers (often requiring multiple readings), you "accidentally" also memorize a lot of the material (even if not for ever and ever), which makes it possible for you to "try to improve it" etc.
So all I am saying is that rote learning is a necessary component of applying any suitably advanced material. Whether that happens organically by engaging with the material deeply enough or on purpose by simply rote-learning is orthogonal to the need for that knowledge to be there.
At issue here in this thread is whether testing for such knowledge is signalling anything of value. I still believe it is because you'll have that knowledge with either approach, even if only one of those approaches will gain you more functional knowledge: basically, such a test is a proxy for you having engaged in the material.
Which is why I suggest a mixed approach in a sibling comment: test for knowledge of "factoids" to simplify test development and application of such knowledge where you are required to manipulate "learned" concepts (so only part of the test is hard to create and keep consistently hard/easy).
We have access to a lot of knowledge at our fingers, it's true. But if the data is all stored there instead of our brains, then we won't be able to recognise that fact A (which we just read) is similar to fact B (which we read a long time but didn't bother to remember because we could just look it up anyway). The implications of that depend on the context, of course, but in a general sense, I think it'd slow our development as a species down.
Exams are generally supposed to test your understanding, not your recollection abilities - with some exceptions, of course, but in general that's supposed to be what they're for.
That said, for many years I shared the view. Some time ago a convincing argument for "memorization" was presented, and I am slowly changing to the view that it is a reasonable way to learn and to become more creative.
When we locked down in the UK, millions of people were furloughed for 6 months+. How many of those people retrained, studied etc.? Not as many as we might hope because education is firstly about motivation and secondly about information.
But those tests are harder to design, harder to grade and you cannot reuse them - since someone will just post the solution on the internet. And you need to prevent students from just sharing the solution during the exam.
Drills and memorizaton aren't the only thing in education, but they are key.