02 July 2011

Book Reviewing

I am well calibrated to some movie reviewers, e.g. Roger Ebert, to the extent that I can usually predict my rating of a movie based on the review. I've not found the same with book reviewers. In fact I seem to be get a bum steer every time I buy something on the back of keen reviews.

The most recent example is Robopocalypse (robocalypse?) by Daniel Wilson, which we bought after a gushing pre-publication nod from Corey Doctorow (usually spot-on, will give him a pass on this one). Got that sinking feeling not long into it. As someone who loved Max Brooks' World War Z, I expected to enjoy it, but it seems way more contrived (yes, I know, more contrived than zombiegeddon) and excessively derivative. Not only did he follow the pattern by first publishing an analog of The Zombie Survival Guide in the form of How To Survive A Robot Uprising, but he goes on to seemingly replicate some of the World War Z characters as well (e.g. old Japanese social outcast who defends solitary kingdom, selfish teen who discovers bravery). The recurring characters seem crafted to sell movie rights (well, should have known, but at least this has potential to be quite an entertaining film). There is way too much "telling" and not enough "showing" -- that is, too much of the narrator telling me that something was really important or influential or that someone was a great leader, rather than letting the story show me someone demonstring leadersheip, or show me the impact of events or actions. (Even the overrated but not bad Philip Pullman "Northern Lights" trilogy suffers from this, with almost every character randomly ejaculating love and praise for the awesomeness of that "little girl".)

Another problem is the premise. Here's where I kick myself for listening to recommendations. Thirty years ago I remember the publications that would accept sci-fi stories practically begging for no more "computer-gone-crazy" stories. And appliances turning on people was an old gag even then. And, seriously, robots suck. I recently watched a demonstration of some robots making breakfast. It was sped up to keep people from falling asleep watching it, and the robots were crap at every skill. Industrial assembly robots are awesome, sure, but compared to advances in computing, robotics is young and stilted. The computer on my phone is powerful enough to kick my ass at chess, but the world's most sophisticated robots can get an egg out of the fridge and fry it about as well as a trained dog. So when Doctorow calls the book "gripping, utterly plausible, often terrifying" I'm left baffled in disagreement on all counts. Premise aside, even the dumbest premise can make for a good read, but this book just didn't work for me.

Justin Cronin's The Passage was a widely gushed-over book that I got upon reading reviews. The prose was solid, definitely good craftsmanship on display in the writing. But, again, there was a big outtrade between the recommendations and my reading. It was... ok. It badly needed editing (real editing, I'm not talking about proofreading). And much of it that seemed to be building up to really big plot points or big moments for characters that either never really happened or turned out to be disappointing or kind of nonsensical. Lots of comparisons to Stephen King. Maybe that should have been a tipoff. If you want an imagined Stephen King experience without reading Stephen King, I'd try The Unquiet by John Connolly.

So either I find a book reviewer I can calibrate to or just ignore book reviews in general and solely taking recommendations from people I know. In that vein, the book I just finished that I'd recommend (although it's almost written like a play): Explorers Of The New Century by Magnus Mills.

1 comment:

JustJoeP said...

Robot proficiency and capabilities are a direct result of how much time, money, and effort are spent on development and programming. Industrialized assembly robots are awesome at doing simple tasks, repetitively, with robust factors of safety and usually excellent precision. Nanobots, quantum computing, and AI have not yet turned the corner to the point where someone can employ the patience of a robotic AI developed advancement, combined with the practicality of mundane, everyday, widely varying tasks. Cooking eggs, dressing a small child, petting a cat, playing a drum solo... these things do not lend themselves to complex, non-linear, divergent programming and 6 axis (minimum) execution.

Yes, there have been multiple advances in all sorts of AI, nanobots and nanomotors/nanomachines, and quantum computing, but from what I've read about in the scientific literature, their all narrowly focused, on a small microcosm, or extremely limited task. The I-Robot scenario of thinking, "feeling", "sentient" AI I doubt will begin to be approachable in our life times.

I used to think porn / human lust / the sex industry would drive virtual reality advances and then AI advances to the point where "sentient", highly versatile, and unpredictable (so there's variety) 'robots' would first be used by very rich, very geeky, very undesirable men to fulfill their fantasies in ways humans cannot. For the first 10 years of the internet, the only thing that made money was porn. Sex sells. BUT, it hasn't happened yet. Maybe someday when we are in our 90s, they'll have Bladerunner-esque bots in the prototype stages. It faut voir.