The problem with much of sci fi is not with the ludicrousness of some of its speculations, but the failure to grasp emotional significance of situations and the failure to recognize that texts and words can convey the things. At heart the problem of science fiction is that focuses not on self-expression or introspection but on human potential (with all its promises and pitfalls).(Note: I haven't read much Lem [bad me], so some of what I have to say here may not be accurate in his case. I've read nearly everything Jim Kelly has ever published.)
Lem is probably not a representative example of sci fi, and in fact, just last Saturday, I listened to a remarkable story by Jim Kelley (Bernardo's House) which explorers an android's emotional connection to her owner. Not only does Bernardo's House adopt the point of view of the house/female android, it also conveys perceptions as they would be experienced by a nonhuman (an interesting notion by the way). Lem is interested in warning us about the future and human limitations in an impersonal way; Jim Kelley is more interested in showing how attachments could exist between humans and nonhumans (and pondering the social implications of this fact). He is interested in having the reader compare the emotional needs of the android protagonist to his own (in addition to experiencing the sensual pleasures of conjuring an emotionally/sexually available female android). Jim Kelley's story is appealing to our humanity.
Still, Kelley's story boils down to "Gee whiz, wouldn't it be neat if we could have sex with androids!" The problem is not that this is a ludicrous fantasy or that it is titillating; the problem of this story (and science fiction in general) is: what happens when having sex with androids become a reality? Would Kelley's story be still worth reading?
"At heart the problem of science fiction is that [it] focuses not on self-expression or introspection but on human potential (with all its promises and pitfalls)." This can, indeed, be a problem (just as stories that focus on self-expression or introspection have the problem of tending toward narcissism). The wide-angle tendency is an obstacle every writer and reader of traditional, genre SF faces, because it can be tempting to tackle the entire history of the universe (think Olaf Stapledon), which can produce some interesting thought-experiments, but seldom provides satisfying character development, something many readers desire, and easily leads to stories that inadvertently simplify complex concepts.
Because it often seems like science fiction is about the ideas that drive the plot, and because many mediocre writers actually try to write that way, I can understand where R.J. gets the (I think mistaken) idea that "Kelley's [sic] story boils down to 'Gee whiz, wouldn't it be neat if we could have sex with androids!'" Many SF stories do boil down to their surface concepts. But the good stories are more complex than that, and so they avoid the datedness that, frankly, the majority of science fiction has always faced. ("Bernardo's House" is, to me, one of those stories, because it would never occur to me to think it's just a story about sex with machines -- I found the character interactions involving regardless of the characters' species, and don't think technological change would affect my fondness for the story any more than it does my fondness for certain Theodore Sturgeon stories published fifty years ago.)
Stories with an inherent "sell-by" date can be interesting, but they are a different thing from stories that explore abstract, general, or ostensibly universal themes -- compare, for instance, many of the stories in 1950s issues of Astounding to the early stories of Philip K. Dick, and it's clear we're dealing with two quite different tendencies, because Dick never seemed to have much interest in science, and this forced him to write stories propelled by strange and/or goofy philosophical speculations dressed up in the typical science fictional gear of the day (rocket ships, robots, etc.) Plenty of stories published in the 1950s that were important to readers then and got people thinking about technology or science in some new and different way are now forgotten and unreadable and horribly dated, while even Dick's worst stories (and there were plenty) still have the ability to make us think. The various movies made from his work have updated the tech and added chase scenes, but each makes at least a passing nod at the central questions that so obsessed PKD: "What is human? What is reality?"
This is not to slight the forgotten or dated stories, some of which gave pleasure to a lot of people when they were published. This is simply to say there is a difference, just as there is a difference between a novel like The Jungle, which was an important piece of muckraking but hardly a good example of the novel as a form of art, and Ethan Frome, published a few years later, which didn't have nearly as palpable an effect on society, but is vastly more satisfying artistically than The Jungle.
At her blog, Farah Mendelsohn is exploring definitions of SF as well as the effect of reading experiences on different people (particularly children, the focus of her current research -- which reminds me, I need to do her questionnaire). Various of her concepts overlap at least partially with this discussion, for instance her statement that "For those of us who are fans, sf is a work of realism. The reading protocols of sf demand that we take things literally. To probably mis-summarize many critics, if we live in a world were someone can 'give' me their hand, then what we are suspending, is not a faith in realism but faith in metaphor."
One of the appeals of any form of fantasy fiction is its ability to make what would be metaphorical in one context into literal reality within the context of the story. There's a difference between the allegorical or symbolic literalization within the work of, for instance, Kafka and Borges, and the absolute, take-it-for-what-it-is literalization of most science fiction, which is one reason people who can read Kafka with no problem get stuck when trying to read genre SF. (The danger for people who read nothing but SF is a tendency to become dulled to metaphor as metaphor, because SF readers become so adept at suspending the faith in metaphor that Farah mentions. But that's another subject entirely...)
Sometimes a rocket ship is just a rocket ship. The trouble is, if a story is primarily concerned with wowing readers with a concept -- that literal rocket ship, for instance -- the story loses its value if the concept gets overtaken by the wonders of reality. Hugo Gernsback's novel Ralph 124C 41+ was first published in 1926, the same year that Kafka's The Castle was first printed. Gernsback's novel is the purest of science fiction, offering almost nothing to a reader other than technological speculation. Kafka's novel is an unsettling fantasy that bears little literal relationship to the world as we know it, but a lot of relationship to the world as it is emotionally and intellectually experienced by anyone who ever encounters, for instance, a bureacracy. Kafka benefitted from the literalizing tendency of fantasy that pounds similes into basic metaphors, allowing the Castle to be not just like something we desire entering, but something K. truly does want to enter. But Kafka's book has lasted through the years, and communicated so well to so many people, because while its metaphors get literalized within the story, they also maintain some of their abstract qualities. A reader who approached The Castle simply looking for a fantasy story would might find the book interesting, but would certainly miss many of its pleasures, which lie in the border between the fantasy of the story in the reader's own reality. A reader who approached Ralph 124C 41+ with no interest in technology as technology (and not a metaphor for anything) would, even in 1926, have been profoundly bored, if not completely perplexed, by the book.
Comparing one of the greatest and most influential books of the 20th century to an esoteric oddity is obviously unfair, but thinking about such vastly different books may be able to help us understand the assumptions we carry into any reading, and the better we get at noting our assumptions, the better able we will be to communicate about what we read.
Update 1/23: Cheryl adds some thoughts and thinks I've tackled the subjects here without thinking them through enough. That may be true, but I should make clear here that my intention was not to offer a thorough analysis of R.J.'s post, but rather to look at a couple of its ideas and link them to other things going on out there. Nor did I intend to suggest, if I did somehow, that I think "Bernardo's House" is doomed to timeliness. I thought I said exactly the opposite in a parenthetical remark comparing the story to some of Sturgeon's best. If you are in the habit of not reading parenthetical remarks because they seem parenthetical, then please reread the above with parentheticals included.