Spelunking CACM (1976)

 

Above are two tables from a report on production and employment of CS Ph.D.s in the U.S. I fear those ratios have not changed as much as they should in the intervening half century. One or two new Black CS PhDs a year. Typically single digits for women. And, of course, it was the 1970s, so the categories of minorities of interest were Females, Blacks, and Foreigners; my apologies to those who find the categorization offensive, which most people will by modern standards.  I have heard that the percentage of women PhDs was higher in 1980 than today, but it was clearly bad in the early 1970s. Pursuing a more complete comparison is a task for another time, but just a few months later CACM had an additional article on the status of women and minorities in CS. Read them both.

Allen and Cocke talked about directed graphs as representations of control flow and possible data modification in programs. I don’t think this is the introduction of the concept of data flow; certainly control flow and flow charts are much older than the mid-1970s. The paper does list work by Kennedy, Aho and Ullman on liveness and data flow. I don’t quite understand how they break loops and how they define “intervals”, which are roughly akin to small functions with a single entry point. But this is an important concept in understanding "liveness" of data, dependency in computation, and how to schedule operations. Without work like this, concepts as varied as garbage collection, caching and out-of-order instruction execution have no theoretical foundation.

Oliver Smoot presented a report on a WIPO meeting on protecting software. It wasn't the first such meeting; the UN authorized a group back in 1971.

Lampson and Sturgis have a beautiful (of course) paper, a retrospective on an OS designed at Berkeley. Of course they had the benefit of hindsight, but as always, they clearly laid out goals for the system, a philosophy, an analysis of what they thought was possible, then the system itself, and importantly, provide extensive discussion of what went wrong both technically and in project management. This paper alone could be a great model for how to write about computer systems.

Keller talked about formal proofs of parallel programs, two years before Hoare's CSP.

Perhaps the highest impact of year was Metcalfe and Boggs on a little idea called Ethernet. They cite an earlier paper by Farber in the Feb. 1975 issue of Datamation magazine.

The number of papers I want to talk about each year continues to grow. I'll have to be more selective, so I can go back to doing deeper dives.