Hacker News Reader

Comment by wheaties | Hacker News Reader

And I expect to see large improvements in Pandas as it is an active and vibrant project. This will be hard to replicate in Julia, not for lack of trying by the community, but rather due to the head start of the Python community.

To be honest, if Julia wanted to really captivate an audience it would need to come out with a package or library which did something needed by the community in a way that Python didn't. Then you could conceivably see larger adoption rates.

View on HN

Wut? Python is going to get better because it has a head start? What's the logic in that? Surely the incumbent is less likely to improve than the challenger because it's already been polished. IOW it's usually the new guy that has room to improve.

I never said Python is going to be better. I merely stated that with a head start in terms of functionality you get from Pandas that Python will have the upper hand.

Also, if history is any guide, the "new" and the "hot" tend not to beat an incumbent just because they're new and hot. Having the most room to improve is not a blessing. You have to improve at a rate that is faster and greater than the incumbent can improve. And if they can't improve at a faster rate, libraries and ecosystems win by offering something compelling that is either hard or impossible to replicate in the incumbent.

Your argument assumes that improvement is equally easy in both systems, which is not the case. The Python ecosystem has a substantial head start, but Julia makes library development much easier – I would hazzard a 10x factor, but it's really hard to quantify. The result is that developer effort is amplified significantly, making it plausible for the newer system to "catch up" faster than you'd expect.

I don't really like to look at it so competitively, however. Open source is notoriously not a zero-sum game – everyone wins when there are more good options, whether they are in Python or Julia.

Reply on news.ycombinator.com