Monday, September 28, 2009


A Gate at the Stairs
Lorrie Moore

Tree of Smoke
Denis Johnson


Lorrie Moore and Denis Johnson are both novelists best known for their short story collections (Birds of America and Jesus' Son). While I enjoyed both of these novels, I don't think either comes off on the whole; along with Anne Enright's Gathering, they are good examples of what happens when novels with interesting material suffer from structural flaws; in each case, I think the flaws have to do with the fact that short stories (and esp. collections) require less careful arranging than novels. Each of these novels is, in its own way, clumsily contrived; they all feel over-planned, as is often the case with work that isn't planned carefully enough.

Johnson's 600-odd page book aims to be a grand panoramic novel about Vietnam; the theme might induce a sinkiang feeling but shouldn't, Johnson's very good on people in the more advanced stages of rage and fuckedupness. Tree of Smoke is filled with the usual Johnsonian menagerie of memorable characters: Kathy Jones, who grieves for her dead Adventist husband by having uncomfortable sex with the first white man she finds, and stalks leanly and hungrily through the novel; the Colonel, legendary and drunk, with his crackpot theories about the CIA's proper place; the Houstons, all id, whom the war neither kills nor strengthens but can still scare shitless; etc. All of this is very well done: so, on the whole, is the main character Skip Sands, who's the straight guy to Kathy and the Colonel until he unconvincingly goes apeshit and ends up running guns in Thailand. But all of this is spread out among reams of deadweight: local color, formulaic barroom scenes, a Vietnamese subplot involving a double agent (in general the Asian characters are lousy), and more than you ever wanted to hear about jungles and temples. The book is arranged chronologically, beginning (disastrously) with the JFK assassination and walking the characters through the major stages of the war; the interleaving of stories is often distracting, and generally makes the first quarter or so of the book quite painful. The second half is better and less portentous, but even here it's not entirely clear that the stories add up to more, interleaved, than they would as e.g. a book of 'Nam-themed short stories.

A Gate at the Stairs is a coming-of-age novel with 9/11 as backdrop; thankfully it doesn't have much to do with 9/11 but, once again, the Big Themes are too blatant to work. Unlike Tree of Smoke it has a plot, sort of: Tassie Keltjin is a "quasi Jew" from rural Wisconsin who goes to college in a random midwestern college town and gets enmeshed in the relatively complicated life of Sarah Brink, a middle-aged woman who tries to adopt a child with her sleazy husband and hires Tassie as the babysitter. The child they end up with is half-black: various cultural difficulties ensue, which end up with Sarah setting up a support group for minority parents to get drunk and blather -- unconvincingly, superficially, and at inordinate length -- about minority issues. Meanwhile Tassie has an affair with a "Brazilian" who turns out to be an Islamic terrorist. When Tassie goes home that winter she finds that her brother is about to enlist and her mother is insufferable. Everything eventually disintegrates at the end of the spring semester; Tassie goes home for the summer and her brother gets sent off to Afghanistan and dies. Tassie gets weepy and reflective, takes a semester off and then goes back to college.

Nevertheless, this novel has its merits; it's beautifully written in bits, and social -- and other -- events are sharply observed. Sarah Brink is worked out with extreme precision. There are two reasons that none of this really helps in the end. The first is that the plot is too obviously contrived to get the social (and intellectual, and whatever) commentary in, the deeper problem though is that the commentary is shallow and boring. Two examples: 1. Tassie supposedly has her mind aflame with Simone de Beauvoir, Chaucer(!), etc., from her (pre-plot) first semester, but her second-semester classes are almost entirely an excuse for Moore to satirize academic silliness and out-of-touchness. The satire isn't good enough to excuse the breach of character. 2. Tassie is paid to babysit the Brink baby and all the other multicultural babies while their parents go on about culture downstairs. From time to time someone uses an agricultural metaphor, and the narrator reminds you that she's from a farm and knows what a backhoe is. This gets substantially less endearing about the sixtieth time.

What these novels have in common is that they're too easy to categorize. They seem to have been written along the following lines: "let me write a novel of type X. Where shall I set it? What sort of characters do I need? What themes or battles should I include?" This is both overambitious -- both books would have been better if the authors knew their limits -- and lazy, because the framework of a book ought to be revised to deal with the obvious fact that some bits work a lot better than others, or have diverged from the main plot. This is also part of what was wrong with Anne Enright's The Gathering -- deadweight like the Lamb Nugent plot was left for the reader to sift through because taking it out would have involved rearranging the book, which would have been too much trouble.

Wednesday, September 16, 2009

Mere Integration

After that last, rather hastily written post (I've been swamped with work lately), I realized I should have read the SEP article on Parfit's repugnant conclusion before blogging about it. The provenance of the repugnant conclusion is different from what I'd expected: Parfit was apparently led to it by the puzzle of what it means to call a state of the universe better or worse than another if they have different people in them. (This is called the non-identity problem: not related to Quine's maxim "no entity without identity," which a rudimentary acquaintance with Actual People will discredit.) One could argue that two worlds with different sets of people in them are simply incommensurate -- this is one of the ways around the repugnant conclusion -- because "it would be better" is meaningless except as "it would be better for X," where X is a specified actual entity. I'm somewhat sympathetic to this dodge, but it seems overly strong because as phrased it doesn't allow quantifiers. (It's clearly sensible to state that nuking Dresden in 2045 would be bad for anybody living in Dresden in 2045.)

The situation is a little different for the intertemporal version of Parfit's paradox: here, one's measuring the same person's utility at different times. The form of the paradox is identical, however: you have a 50-year high-quality lifespan, ceteris paribus it's good to add 25 years of acceptable but unstellar life, it can't be bad if you even out your quality of life over the 75-year lifespan, rinse wash repeat. One could argue that my self at 50 is different from my self now and I'm within my rights to be a total asshole to my future self -- in which case the good-for-whom point still works -- but I don't buy it. That said, regret is an odd business: gratifications delayed are often gratifications forgone; gratifications indulged lead to syphilis. And it might be true that different parts of one's life are more different from one another than the lives of two different people of the same age, though I wouldn't really buy that either.

Of the eight responses that the SEP offers to the mere addition paradox, about half -- average utilitarianism, steplike or branched utility functions, and the claim that there are no lives remotely worth living -- work pretty much identically in the temporal context. The claim that no lives are worth living is, in fact, on safer ground in this version because it's less susceptible to objections of the happy-space-alien variety. The dumb response, viz. that transitivity doesn't hold, remains dumb. The responses denying that it makes sense to compare two worlds with different people, however, are unintelligible in this context; one is led to believe that they were never particularly good dodges, though some of them might be useful for other purposes. It might be that other dodges along similar lines can be found, but I'm dubious.

Monday, September 14, 2009


Generally rather be caught dead than link to lesswrong; however, this post on the "Lifespan dilemma" is sort of interesting if you can grit your teeth and deal with its spectacularly awful prose. I haven't read the post esp. carefully but it seems like there are two aspects to the idea: (1) a lot of math that seems basically irrelevant, (2) the clever idea of reformulating Parfit's paradox in terms of years of life rather than numbers of people. Personally I don't think Parfit's paradox changes when you formulate it in terms of lifespans than numbers of people -- the correct unit is clearly people-years, rather than people or years -- but I guess this might not be widely accepted.

I'm an average utilitarian* with respect to Parfit's paradox, despite the somewhat compelling counter-argument that this should lead one to prefer having thousands of people mildly tortured to having one waterboarded. I would like to reconcile this with a system where, like Parfit, you have a floor of utility for a life to be "worth living," and (unlike him) adopt average utilitarianism for people above the floor and stoatal utilitarianism for those beneath it. There seems to me no a priori reason why one shouldn't use different metrics for lives that are worth living and lives that are not, and this metric matches my intuitions best, even if it seems inelegant.

And I think I'm also an average utilitarian* with respect to timescales, which sort of explains why I hold the views on longevity (i.e., that there's no intrinsic merit to it) that I do. I should emphasise that these actually are my intuitions on the matter; I'm not trying to be contrarian, I just actually disagree that longer lives are better. The appropriate measure is utility per second averaged over one's life; a longer life is better to the extent that one can expect this to increase -- e.g., as an adolescent you'll probably increase your average happiness by staying alive a little longer; this also accounts for situations when you're 65, dying, and want to be kept alive long enough to e.g. check out an impending grandchild before you kick the bucket. One could also argue that as you grow older you've read and experienced more and therefore find life richer; therefore there's a steady upward trend to one's utility per second. I don't think this is true beyond a point, though, and there's a tradeoff here which Wordsworth's one good poem is very good on: one gets contemptible, stiff, and dull with age. Some of this has to do with the physiological processes that cause aging, but I feel like some of it also has to do with just having been around for a long time.

* I'm not a true utilitarian. However, like most people I'm a utilitarian most of the time, so it's meaningful to ask whether I count average or stoatal utility.