[Absolute0]

Again you miss my point. Consider Machine Learning. There is an amazing open source software called as WEKA. however contributing to WEKA isn't as much significant as say presenting a significant work in ICML or NIPS or writing a paper in JMLR.

Open Source is amazing for programmers, but for someone who wants to develop say a novel recommendation system or special Computer Vision algorithm. Open Source participation is secondary. Also in many case ML algorithms tend to be trivial in their implementation.

Also my time is limited, I could try getting involved in Weka and implement recently discovered algorithms in it Or I can read papers, explore different datasets and try to come up with new ideas. Second case makes more sense to me.

All I am saying is that it makes more sense for me to display my capabilities in developing better algorithms rather than ironing out their implementations.