Interview
Travis Oliphant: NumPy, SciPy, Anaconda, Python & Scientific Programming | Lex Fridman Podcast #224
- The creator of NumPy, SciPy, and Anaconda is expected to continue empowering scientists and engineers across big companies, small companies, and open-source communities to solve difficult problems through programming.
- Python is predicted to maintain significant market inertia for many years, making it "very hard to push away" despite the existence of other tools.
- Julia is forecasted to struggle against the established dominance and inertia of Python's ecosystem.
- Python's accessibility is expected to endure, allowing users, including those without deep expertise, to solve problems without needing to become language experts or rely on a heavy "translation layer."
- The language is projected to evolve based on user and pioneer requests, maintaining a focus on readability and practicality even while retaining some known limitations.
- NumPy is anticipated to continue inheriting concepts like rank, broadcasting, methods, and reduction from its predecessors, specifically Numeric and APL traditions.
- A specific timeframe of "2015-17" is cited as a period when NumPy was the current array standard, despite the emergence of numerous alternatives in the "past two or three years" relative to that context.
- The scientific and technical worlds are predicted to face a loss of genius from non-English speakers due to the dominance of English, which the speaker fears will obscure significant potential contributions.
- The speaker predicts that over time, speaking specific languages will transform individuals into "different human beings," citing the intranslatability of concepts like Russian poetry or the emotional swings inherent in those cultures.
- Non-Latin-based languages like Chinese and Japanese are expected to face difficulties if programming syntax does not leverage the user's native language center, potentially making these languages harder to learn compared to Latin-based ones.
- Microsoft is predicted to have historically struggled with understanding array-based programming, though the company is expected to have improved in this area.
- The speaker notes that array-based programming is currently underappreciated by those in systems or web programming cultures who may not understand n-dimensional arrays, yet it offers a liberating perspective on data as objects.
- Cultural expectations of compactness in languages like Perl are predicted to lead to code that becomes increasingly unreadable over time, contrasting with Python's emphasis on longevity and readability.