The Prototype's Language

The article really needs to explain how a comparison of Python to Java is supposed to explain what languages people choose for new services in 2024. Why aren't other languages on the table?

Even taking it at face value that the choice is between Python and Java, the article leaves so much unaddressed.

> a good plan executed now trumps a perfect plan done next week

In what universe does this apply to a payments system?

Hatred for Java is burned deep in my soul, and I've spent an unreasonable amount of time on this very web site carefully articulating reasons why I think Java is a crufty relic whose design reflects a bunch of discredited beliefs from a quarter century ago. But gosh, if I had to choose between Java and Python for a payments system, I think I might choose Java.

I'm not sure I would want to use Python for anything that remotely touches money. It's far, far down on my list of languages I would use for something that has serious consequences for being incorrect.

> The single most impactful technological choice when building payments is the programming language.

Got it. So why was Python chosen again? Call me crazy but I want things that touch my money to be rock solid.

> The moral of the story is that in formal systems, how you express problems impacts what problems you can solve.

An interesting exercise many programmers don't do but should is to program purely in paradigms other than imperative OOP. The solutions you come up with are very different when you don't have objects, for loops, and if statements in your language's vocabulary.

Try writing a real program in Prolog or SQL. Not just database calls to retrieve data, but an actual full app with a UI and user interaction and network calls and all the rest. You will likely feel like a novice programmer if you haven't trained your brain to think in set operations like union, join, intersect, etc. When everything in your program is a set instead of a stateful objects, you do set operations over them instead of method calls. That's uncomfortable and disorienting at first.

As for the article, I didn't get much from it past the bit I quoted.

FWIW that PG quote is fully out of date; I do not associate Python with programmer competency. The first thing I ask when a programmer says they know Python is "What other languages do you program in?", because I've come across far too many Python programmers who can't tell a stack from a heap. YMMV

I'd happily prototype things in python 2.7 but not in 3.something. The ratio of looking up documentation to building stuff has gone badly downhill. Sadly most dependencies no longer run on 2.7 because 3.0 was better in some sightly unclear fashion so python 2.7 is now, finally, essentially dead.

But whatever, we've still got lua 5.1ish.

The rest of the article is coherent, including the properties you want in the language, I just think python in particular has catastrophically lost its way.

The post just skips over _why_. It states that Python is 'more modern' than java as an axiom. This is not a particularly easily adopted axiom. For example, python is in fact older than java. Which surely isn't what the author is talking about.

But what _is_ the author talking about? It just isn't said. It merely states that the argument 'python is slow' is no longer relevant. It doesn't even state a reverse argument (why python would be better than java), and thus doesn't even begin to allow waylaying _those_.

As a trivial example, perhaps the author means: A nominally, strongly, explicitly typed languages is obsolete. In which case - how does that explain how, for example, typescript has happened, which actively _re-introduces_ these aspects to a language?

Clearly then, 'python is superior' cannot be taken as a truth universally held and requiring no logical argument.

The rest, sure, whatever - "Programming language used has a significant impact on what you can write and how one would write it" is.. kinda obvious, no? Turing machine logic says that it is _literally_ incorrect (anything you can compute in one, you can compute in the other), but that _is_ something all those with a modicum of programming experience probably take for granted. You _can_ write a compiler in brainfuck, but, that'd be insane.

The tricky argument isn't stated, the obvious arguments are.

I surely must be missing the point of this post.

I can imagine a single uncaught exception bringing down the entire payment system...

Meanwhile, here in reality, I've been digging deep into Java precisely because Python just isn't cutting it when I'm trying to make the switch into "real" software engineering from my software project lead role (where I end up coding half the time anyway, I just don't get any recognition for it).

No no no, it’s 2024, we write everything in Rust now