For years I considered only social network optimization, not as any kind of social engineering but as a tool or technique offered to those who wanted it, to improve their own lives.
A problem has arisen, however. The latest algorithm, Recursive Exhaustion, has the potential to make society much worse. It could be used for social network pessimization, (pessimization is a relatively new term, used largely in the C++ community). Indeed, it could be used for social engineering in the worst of senses.
In a novel I have been working on, I address this problem through the means of a society of people dedicated to fighting the people who use it for evil. In theory it could be used by governments instead, who could make its abuse illegal for anyone but law enforcement, but I do not trust governments. Neither do the “good guys” in my novel.
See my account of how writing this book led me to my own (re?)discovery of this algorithm, with its potential for good or evil.
With reference to the diagram in the website header, which compares two ways of linking up three people, the red and black arrows refer to compatibility, which I described here in a 2004 chart.
These levels are based on the idea that the chances of finding a compatible person depends on the size of the pool of candidates, and my theory that like many aspects of human perception, a logarithmic scale is appropriated. The best match out of 200 people will not be perceived as twice as compatible as the best match out of 100. But the best match out of 10,000 might seem twice as compatible, being a Level 4 match, compared to a Level 2 one.
This diagram is taken from an old hand-coded page full of them, some of which will appear on this site in other posts. That page, written in HTML, dates from 2004.
It illustrates how a poorly linked social network can be improved using an optimization algorithm. The width of the lines represent the level of compatibility and effectiveness of communication between the individuals. The Before diagram shows a network with missing links which isolate some groups of people from the rest. None of those links are as good as those in the second diagram, obtained by just changing who is connected to whom.