Why does open source work?
Hed Gulley knows one way.
Mathworks runs a biannual programming contest in Matlab. Most programming contests pit solo contestants or teams in head to head competition, each contestant working alone and submitting their entries at the end, but this one works a little differently. Entries are entered in the open. Everyone get to see every entry and everyone knows what is the best entry at all times. More than that. You can submit an entry that is only a sinle line change from the previous best entry. If your entry is faster, you are the new leader.
The contest takes place over a week and the results can be plotted and analyzed. . There are two kinds of progress. Quantum jumps and the small tweaks. Looking at the graphs it appears to me that the majority of the progress comes from the quantum jumps, maybe 75%, but I can’t really tell from the data that the article shows, although it seems like Hed has access to the appropriate data.
It is also possible that the tweaks allow a give method to be pushed to its limit and show the necessity for a new approach. Without the openness, there are more possibilities for doubt about the limits.
It would be interesting to score the contest on the basis of contestants contributions to the incremental improvement.
Sorry, comments for this entry are closed at this time.