Those numbers come from a peer-reviewed paper and it refers to the data they found in their specific experiment. It is not argued that those specific numbers ap

Those numbers come from a peer-reviewed paper and it refers to the data they found in their specific experiment. It is not argued that those specific numbers appear in all cases. However, that data happens to be consistent with several other similar experiments done over the past 20+ years. The rest of the short article builds an arguments based on the assumption that at least 50% of the development time is spent figuring the system out. If you find this assumption to be off base, please disregard the rest of it. In any case, I'd encourage you to ask peer developers if they find that they spend 50% or more of their time reading textual artifacts as a way to figure the system out (code, logs etc).