Similarly, do not discourage anyone from pursuing photography by telling them that they have to start with something like film.
This extends to trying to tell people that they should learn programming by starting with (for example) assembly instead of python.
When I started photography as a hobby years ago, the community in the forums never suggested I used film, neither did friends (a couple of them working at pro level) and colleagues. I know there are some snob people out there but even the ones I’ve met never actually said to start with film.
That being said, every one of them gave the same advice: don’t worry too much about the gear; instead, go learn light and composition.
After you manage that you can do the same with any other camera to improve quality.
I think I’d suggest film because it’s just much more fun and physical, not because it gives you better pictures.
It makes the learning curve steep, which can be offputting for a lot of people.
A polaroid is not a bad choice because the compose-shoot-view_results loop is about the right length of time, and the limited number of shots in a pack makes the "compose" step important straight away.
Having a knowledgeable but not dominating friend helps. "Try kneeling. What do you see?" "What about waiting for that cloud?" That kind of help.
ASM seems to give people a really bottom up mindset and it takes a while to adapt to Python-ish practices. ASM people think "Oh I could do it these ways from scratch", Python people think "I know just how to express that in terms of this highly optimized numpy object".
I took two semesters of computer science in college—working in Java. I hated it. I just wanted to build websites that did things and those courses didn't get me any closer. I built a bunch of apps that had to be compiled and nobody would ever use them. I swore programming off until I discovered WordPress a year later and started building WordPress themes.
And what you call bottom-up approaches have value too. You talk about "how to express in numpy?", as if you were dead set on using numpy. But it numpy the right choice? It is highly optimized, sure, but is it highly optimized for your use case? A "bottom-upper" will start to think "how will I solve this problem" and then look for something that will make their task easier. It may be numpy, it may be "from scratch" or it may be something else. Ideally both approaches will lead to the same conclusion. The numpy guy may realize that his favorite library is not the right choice in the same way that the ASM guy may realize that his solution is already implemented in a library.
Anyways, a good programmer needs to have an idea of the whole spectrum, from hardware architecture to assembly to low level languages like C to high level languages like Python. Start where you want.
Hehehe I think only film snobs do that. Film is fun but these days for beginners, grab a digital camera because of instant feedback: you'll automatically see the image you've taken. And if you use more sophisticated camera, you can also see the histogram. Useful for analyzing your exposure. Not impossible with film camera.
Once you are comfortable with your exposure, then jump into film.
digital does not -> interpreted language.
p.s. if your goal is professional s/e it is hardly a bad idea to start with a compiled (i.e. film) language. something similar likely applies for photography.