3 ms·
I'm surprised there's no mention or discussion of the importance of open-source tooling for replicable science. Without seeing and reviewing the source, how can
by askvictor 9y ago
I'm surprised there's no mention or discussion of the importance of open-source tooling for replicable science. Without seeing and reviewing the source, how can you tell that a particular calculation is right? Also, relying on costly tools such as Mathematica cuts off a sizable amount of the population from being able to replicate or play with your findings on cost grounds alone.
- falkod 9y agoLong-term Mathematica user (physicist) here: I don't think the use of open source software would make most science -- maybe that does not apply to cs/datascience -- more replicable. Usually that takes an expert in the field. And usually these experts are employed at universities where Mathematica licenses are not the prime cost factor. That said, I am all for open source software. Although I would argue that probably trustable scientific results do not rely on the inner workings of e.g. Mathematica anyway, but use Mathematica as vehicle for say linear algebra or symbolic manipulation etc. While the inner workings of Mathematica may not be open source, in principle the relevant algorithms are not propriertary but usually well-known mathematical results and as such at least in principle easily reproducible outside of the ecosystem.
- jononor 9y agoOpen source brings with it a strong culture of publishing code openly. It seems that the amount of public Jupyter notebooks is already higher than that of Mathematica or similar, despite those tools having a 20 year lead?