Executable pseudocode: Python, and how a hobby project ate the world

It began as a Christmas-break side project in Amsterdam and became the default language of AI, science, and everyday programming. A field guide to what Python is, why it won, the real costs of its design, and what its numbers actually say in 2026.

There is a small demonstration you can do with Python: show a page of it to someone who has never programmed and ask what it does. Often they can tell you. for student in class_list: print(student.name) reads much like the instruction it is. That property, code that reads almost like the pseudocode you would sketch on a whiteboard, is not an accident, and it is the main reason a modest scripting language from 1991 now sits at the centre of artificial intelligence, scientific computing, and introductory programming courses.

This guide is the first of three on languages that shaped how programmers think: Python, then Prolog, then Lisp. Python goes first because it is the one you are most likely to use this week.

01A Christmas project in Amsterdam

Python's origin is unusually well documented, because its creator wrote it down. In December 1989, Guido van Rossum, then a researcher at CWI, the Dutch national centre for mathematics and computer science, wanted a hobby programming project for the Christmas break. He was after a scripting language that would sit between the shell and C: higher-level than C, more programmable than the shell, and pleasant for the Amoeba operating-system project he worked on. He borrowed heavily from ABC, an earlier teaching language developed at CWI, keeping what he liked (readable syntax, indentation) and fixing what he did not (rigidity, no extensibility).¹ The first public release, version 0.9.0, went out in February 1991.¹

The name has nothing to do with snakes. Van Rossum was reading the published scripts of Monty Python's Flying Circus and wanted a name that was short, unique, and slightly mysterious; the official FAQ confirms it, which is why the documentation is full of spam, eggs, and knights who say Ni.²

02The design that won: readability as a feature

Most languages treat readability as a virtue; Python treats it as a constraint. Indentation is not decoration; it is the block structure, so the code's visual shape cannot contradict its logic. The philosophy is set out, half-seriously, in "The Zen of Python": Beautiful is better than ugly. Explicit is better than implicit. Simple is better than complex. Readability counts.³ Add an official style guide, PEP 8, that most of the ecosystem follows,⁴ and Python code from strangers tends to look like code from colleagues.

Two more design choices did quiet, long-running work. First, the "batteries included" standard library: JSON, HTTP, sqlite, zip files, email, so useful programs get written without installing anything. Second, Python is a strong glue language: its C API lets performance-critical code live in compiled extensions while Python orchestrates from above. That is not a minor detail; it is the architecture of the entire scientific stack. NumPy, the array library underneath nearly all Python data science, is C doing the arithmetic with Python doing the thinking, and it became foundational enough that its design was written up in Nature.⁵

The language also learned to bend without breaking. Dynamic typing makes small programs quick to write; for large codebases, optional type hints (PEP 484, 2015) added a gradual type system without changing the language's runtime character,⁶ a pragmatic compromise in keeping with Python's temperament.

03Growing pains

Python's history has two well-known scars, and both are instructive. The first is the 2-to-3 transition: Python 3.0 (2008) fixed deep design mistakes, notably text-versus-bytes, at the cost of breaking existing code, and the community took over a decade to migrate. Python 2 was finally sunset on 1 January 2020, with the core team's blunt explanation: "we became concerned we couldn't security-patch both versions with the volunteers we have."⁷ The lesson the industry took, never split a language ecosystem in two, has shaped every language migration since.

The second scar is the GIL, the Global Interpreter Lock: a design that made the interpreter simple and fast on a single thread, but meant one Python process could only run one thread of Python bytecode at a time, which is painful on many-core machines. After decades of "you can't remove it," PEP 703 was accepted, and Python 3.13 shipped an experimental free-threaded build with the GIL genuinely optional.⁸⁹ A thirty-year-old language re-working its concurrency story rather than freezing in place is a healthy sign.

Python's strength was never raw speed. It was that the time from idea to working program is shorter than in almost anything else, and most programs are limited by the person, not the machine.

04The numbers in 2026, read carefully

Popularity claims deserve scrutiny, because different measures track different things. On the TIOBE index (search-engine visibility), Python reached an all-time record for any language, 26.98% in July 2025, and still leads in July 2026, though it has cooled to around 19% as the index's compilers debate whether the AI-driven spike is levelling off.¹⁰ In Stack Overflow's 2025 developer survey (what working developers say they use), JavaScript remains the most-used language overall at 62.3%, but Python posted the survey's biggest jump, roughly seven percentage points year over year, on the back of AI and data work.¹¹ And on GitHub (what code actually gets pushed), Python overtook JavaScript as the most-used language in 2024, ending a ten-year run,¹² before TypeScript took the top spot in 2025 as typed languages surged in the agent-assisted-coding era.¹³

Read together, the summary is that Python is either the first or second most important language in the world depending on the measure, its growth is driven by AI and data, and the one area it does not own is the typed, large-team application codebase, which is exactly the gap type hints are meant to close.

05Why AI chose Python

It is tempting to say Python got lucky with the AI boom. It is more accurate to say Python spent twenty years becoming the language AI would need. Scientists adopted it early because it read like their notation and connected to their Fortran and C libraries; Nature was telling researchers to "pick up Python" back in 2015, years before the generative-AI wave.¹⁴ By the time deep learning arrived, the ecosystem, NumPy's arrays,⁵ Jupyter's notebooks, and later PyTorch's tensors,¹⁵ was already where the researchers were. The frameworks do their heavy lifting in C++ and CUDA, with Python as the control layer on top. GitHub's own analysis credits this dynamic for Python's 2024 surge: the generative-AI boom pulled a wave of data scientists and researchers, not only traditional software engineers, into programming, and they arrived speaking Python.¹²

06Where OcxlyDev lands

We use Python daily; the verification harnesses, sitemap checkers, and build scripts behind this site are Python, and our advice matches our practice. Choose Python when the problem is exploratory, data-shaped, or glue-shaped, or when the team's time matters more than the CPU's, which covers most problems. Reach for something else when you need predictable low latency on a hot path, tight memory, or the compile-time guarantees a large team relies on, and even then Python is often the right orchestrator around that faster core. If you are starting from zero, start here: the official tutorial¹⁶ plus a small real project beats any course. And keep an eye on free-threaded Python; if it matures, the language's oldest caveat quietly goes away.