I find some of what makes it into CACM is just absolute dreck. To be honest I think they are having trouble finding authors for articles. I can't tell you how many times I've seen articles from Kate Matsudaira prattling on about managing a software team (which, based on conversations with my wife, who is an actual director of software development for a large team, seem to convey naive entry-level management insights at best) or endless articles from Ivar Jacobson trying to evangelize the Next Big Thing to Replace Agile for software engineering.
At the same time there are usually 2-4 articles per issue that are very good. CACM also presents curated lists of interesting research articles, so (as practitioners) we do not need to sift through every article from every conference proceeding or journal publication talking about how some squeezed an extra tiny incremental fraction of x out of an O(n^x) algorithm or incremental papers discussing how compiler type inference algorithms could be improved this tiny little bit with this one weird trick.
And we occasionally get really insightful practitioner articles as well; for example, some interesting bits of how Google manages their code base, or what an incredibly difficult problems Coverity faces when statically-analyzing (supposedly)-compliant C/C++ code.
I think that's not in opposition to what the article says. The main point is that computer science largely dismisses social sciences, hitting realizations and creating problems which social scientists have long considered and solved. To fix this, more exchange of knowledge must happen between computer science and social sciences. In order for social sciences to embrace computing, computing must realize that social science is a necessary part of their work.
What social scientists consider "solved" does not usually jive with reality. That attitude in itself shows the field still needs to mature. They are also notorious for being more "enlightened" from my experience interacting with them at University.
Don't you think connecting the field more closely with computer science would help it mature, though? Both sides could benefit from this: Both get methods, challenges and insight from each other. It might bring both a bit closer down to earth - computer science gets a better idea of possible social challenges their work could face or produce; social science gains new perspectives from more technical and practical students.
At the same time there are usually 2-4 articles per issue that are very good. CACM also presents curated lists of interesting research articles, so (as practitioners) we do not need to sift through every article from every conference proceeding or journal publication talking about how some squeezed an extra tiny incremental fraction of x out of an O(n^x) algorithm or incremental papers discussing how compiler type inference algorithms could be improved this tiny little bit with this one weird trick.
And we occasionally get really insightful practitioner articles as well; for example, some interesting bits of how Google manages their code base, or what an incredibly difficult problems Coverity faces when statically-analyzing (supposedly)-compliant C/C++ code.