| 0.83 | |||||||
| A.I. Artificial Intelligence | 2001 | 5990 | 3.14 | 0.69 | |||
| I, Robot | 2004 | 6047 | 3.81 | 0.41 | 0.56 | ||
| Cyborg | 1989 | 2068 | 2.79 | 0.66 | |||
| Johnny Mnemonic | 1995 | 4186 | 3.03 | 0.52 | |||
| 2001: A Space Odyssey | 1968 | 5920 | 3.81 | -0.75 |
Four factors resulted, which may be considered four dimensions of variation across the films, or clusters that group together films with much in common. The number in the four factor columns are loadings, expressing how strongly each film represents the factor. Note the negative loading for 2001 in the last column, implying it has a quality opposite to that of the three other films in the category. The first factor seems to have clear meaning, grouping together very high quality, popular films with considerable intellectual depth. The fact that some clusters of films are hard to interpret suggests we may face a similar difficulty in the realm of politics. Note especially that the factors overlap, which is not common in factor analysis, but could easily be the case in a multi-party political system, in which the platforms of competing parties contained some of the same issues.
In earlier research, I had used the same methods to analyze science-fiction authors, finding not only that four dimensions of variation existed, but that each had extremely clear meaning (Bainbridge 1986*). The first group of authors wrote hard-science fiction very closely connected to physical science and technology, and filled with optimism. The second group wrote new-wave fiction, closer to the social sciences, pessimistic, and critical of contemporary society. The third group wrote a variety of kinds of fantasy that ignore real science and emphasize magic. The fourth group consisted of pioneer writers like Jules Verne and H. G. Wells; indeed the fourth dimension of science fiction is time.
If the units being rated were politicians, some of the factors would be clear, representing central individuals in well-established parties. Some would be more complex, perhaps connecting politicians of whatever party who had an anarchist or isolationist streak. Some factors might be based on race, gender, or geographical area, and a few factors might be quite indeterminate in meaning. But if we had a recommender system for politicians, a person who already liked politicians A, B, and C, could be advised what other politicians that person might like, providing links to their websites so the person can check out their platforms and other writings. If the units being rated were specific political positions, then each factor would represent a reasonably coherent platform of compatible positions.
Especially for local politics, it is important to build into the mix of methods some that enable formation of groups of ordinary citizens to tackle particular problems of interest to them. Massively Multiplayer Online (MMO) games do this all the time, and some have excellent systems for short-term team building. For example, World of Warcraft has a good system for assembling small teams of five players, or even raids as big as 40, linking themselves through a real-time communication system on the basis of a short-term practical goal (Bainbridge 2010*). When there are as many as 40 participants, the system is hierarchical, usually assembling individuals in realtime into 8 sub-groups of 5, with more intense communication among the leaders of these small groups, and from the momentary leadership down to ordinary members. The newer MMO, Rift, has successfully worked out methods to bring together similar numbers of people instantly, in a given local area under sudden attack, even without much communication, or at least exceedingly fluid communication.
Long-duration voluntary groups are also common, often called guilds. A guild called “Science” I created in April 2008 in World of Warcraft to organize the world’s first major scientific conference held inside an MMO, is still in existence three and a half years later (Bohannon 2008*; Bainbridge 2010a*). When I registered the guild inside World of Warcraft, I had the power to name each of a half dozen levels of membership, and use the guild-leader part of the interface to decide what powers each level would have. For example, I let all members above the first level recruit new members to the guild, but only very high-level members could promote someone to a higher rank. After the conference, reasonable discussions with the most active members transferred the leadership from me to a
