Comment by disgruntledphd2
4 years ago
Yeah, he was totally wrong on Python.
That being said, I'm not sure he would have predicted that data science would be what pushed Python into a better place. Very few people would have, certainly around 2004 or earlier when he was writing it.
Data science + the return of machine learning.
I suspect this is entirely due to the fact that Python looks like pseudo code and has a claim to be beginner friendly. Beyond that, it's baffling that the data science/ML crowd picked Python of all things. Tearing apart CSV files and hammering fuzzy text data into shape is Perl's forte. So I'm surprised Perl didn't see a revival rather than Python. It's possible the entire field doesn't know there is something better, if all they know is one language I suppose. But I'm sure it's part cargo cult too. Everyone uses React because everyone uses React kind of thing. Same with Rails.
It was pandas, unfortunately. R had the stats/ML crowd for a long time, and data frames are super hard to abandon. Python had a library for them, and was perceived as being simple so it won.
R still has most of the tribal knowledge about stats/modelling though, so many really good DS's end up having to learn python to communicate with software engineers.
> around 2004 or earlier when he was writing it.
That’s the year Ruby on Rails was released and exploded in popularity. Python wasn’t gone but it suddenly didn’t seem like the rocket ship waiting to launch like it had. I was surprised and delighted with how it has grown in popularity since but it didn’t look good in the mid 2000s.
Agreed. If I remember correctly, the infighting because of the 2 vs 3 debacle didn't help much.
Luckily, the 2 vs 3 debacle started to sort itself out just in time for the explosion of data science and machine learning software, almost all based on python APIs.
I think teaching made Python what it is today. First, so many introductory courses have been using Python, and the more advanced courses (outside CS) followed. Second, I also think that there are (many) more people doing mundane stuff, like web development, than data science.
I think both of these things are a direct consequence of the natural-language-like syntax, libraries, and lack of boilerplate.