Wednesday, 15 May 2002

Roger MacBride Allen: Isaac Asimov's Caliban (1992)

Edition: Orion, 1993 (Buy from Amazon)
Review number: 1091

In recent years, this kind of "collaboration", billed on the cover of Caliban as "unique", has become quite common. Basically, someone who is a relative newcomer as a writer (and almost anyone would have fallen into this category when compared to Asimov in the early nineties; Allen was a reasonably well established author) takes a classic piece of science fiction and writes a new novel or series based on it and under the supervision of the original author. The results are frequently surprisingly good; writers other than Asimov who have allowed their work to be used in this fashion include Anne McCaffrey. The benefit of course is that these novels have a ready-made appeal to fans of the original, and many authors of successful novels have a problem fulfilling the demands of their fans and publishers for more of the same. (They frequently want to move on to something related to current interests - and develop as writers.) Also, the younger writer may well have ideas which put a science fiction classic in a new light.

This is indeed what Allen has done with Asimov's robot stories, even though you might be forgiven for thinking that all the possible variations on stories based on the famous three laws of robotics have already been written. In fact, the scenario for this novel and its successors (it is the first of a trilogy) seems to tacitly agree with this, being concerned as it is with the development of a new set of laws to replace those which have been the basis of robot design for centuries. As in Robots of Dawn, the plot of Caliban is a mystery where a murder attempt has proved possible in a situation where the presence of robots should have made it impossible. The victim in this case survives the attack, but has no memory of it as she recovers. Her position as a developer of New Law robots, as one involved with a controversial terraforming project and the freeing of an experimental robot with no law constraints at all (enabling him to harm people and disobey orders) complicate matters. Allen's investigator, sheriff Alvar Kresh of the city of Hades on the Spacer planet of Inferno (marginally terraformed, as its name suggests), is made sufficiently an outsider by his job and his intelligence (the average citizen of Hades coming across as pretty obtuse) that he is in a similar position to Asimov's central character Elijah Baley. Baley works better as a character, because in his person he is a focus for the tension between Spacer and Earth human, and Allen has to import this to Inferno. (He does this by making the terraforming project run by Settlers, descendants of Earth people who began colonising space again after the Spacer embargo on this was lifted after Robots of Dawn.)

The major flaw in Caliban for me is the way that the research into replacing the Three Laws is described. It is said that the laws are impressed into the design of positronic brains at a fundamental level, with the result that new laws require the development of new hardware, the gravitronic brain. This is stressed throughout, but seems extremely unlikely to me, being based on identifying hardware and software in a way which has never been a big part of the design of the electronic computer. (It may come from ideas in some of Asimov's early stories, where computer Multivac is described in mechanical terms.) Even if the laws were partly encoded in hardware, it surely wouldn't be difficult to redesign the positronic brain either to move this encoding to software (as the most difficult part of getting a computer to follow the laws would be to provide sufficiently usable definitions of concepts such as "human", "harm" and so on) or to redesign the hardware to cope (working to redefine things in a familiar environment being far easier than at the same time having to work in a completely new background). Even the way that conflicts between the laws cause the robots to freeze up makes the whole setup seem more like software than hardware.

Since this part of the background is quite fundamental to the plot, it does have an effect on my willingness to accept Caliban (those who do not work as computer programmers may be happier with it). Ignoring the problem leaves a neat little detective story with an well realised if naturally not particularly original background. Best suited to its target audience of the fans of Asimov's robot stories (of which there are many), Caliban would nevertheless have something to offer a more casual reader.

No comments: