My understanding is that, per the Greek root, androids are artificial people, or at least possessing human features in appearance or behavior, though the behavior might be only programmatic. Materials are not especially relevant.
Robots are constructed automata (not drones) of any shape or behavior.
The two sets overlap.
I had always understood “android” to mean any human-shaped robot, regardless of whether or not it could pass as human. Thus, for instance, Daneel Olivaw is an android, but so is C3PO.
Meanwhile, I would not call a humanoid biological construct an android, since that’s not a robot. The best term for such an entity would be just “human”. Or, if one insists upon the distinction, “synthetic human” (not “artificial”).
Speaking of the Star Wars droids, meanwhile, there’s also at least one use of “cyborg”: C3PO introduces himself as being in “Human-cyborg relations”. Presumably, this is just another word for droids, since a human with some mechanical parts (the usual meaning of “cyborg”) shouldn’t need an interpreter to relate to humans, and all the entities Threepio is shown relating with are droids.
You obviously don’t live in Seattle. 
If you read the rest of my post you can get the context, too.
Huh? I did read it. I’m disagreeing (as Chronos appears to be) that the classification you describe is clear-cut.
Then we’re stupid and we’ll die.
And I’m saying that this was the classification scheme during Hamilton’s career. He died in 1977. It’s not surprising that, without him keeping discipline the meaning has once agaiun broadened.
I think the classification I describe is more disciplined, and more meaningful. Why should what they’re made of be so important, overriding general morphology, ability, and behavior?
Also, I am not convinced that Hamilton’s convention was ever a standard across all science fiction.
You can claim that your classification is whatever you want.
In the time I started reading SF, Hamilton’s convention was kept pretty well. It had been from the 30s or 40s on, and it seemed to stay that way until his death. Certainly it swayed Nicholls and Clute’s influential Encyclopedia of Science Fiction to adhere to the same cdefinition when it came out in 1979, and even into its second edition. And even in its current online incarnation:
I’m not sure that “common usage” is really meaningful, here: The only common usage of the term “android” is to mean “a mobile operating system developed by Google”, and before that, the term wasn’t commonly used at all. When the word is uncommonly used, it’s by SF authors and fans, and in that case, it’s usually in reference to the works of a particular canon, each of which will have its own terms, definitions, and distinctions.
All of the robots in Star Wars are called “droids”, regardless of whether they’re at all humanoid, and are treated in essentially the same way. In Asimov’s works, they’re all called “robots”, and there’s again no particular distinction made between the ones that can pass for human, like Daneel, human-shaped but metallic robots, like Giskard, and completely nonhumanoid ones like The Brain. In Blade Runner, organic synthetic humans exist, but are called “replicants”, not “androids”, and mechanical robots are never referred to. In Heinlein’s Friday, there are genetically-engineered “Artificial Persons”, and again, no mechanical robots of any note.
It seems to me that the only context in which one can meaningfully say that “Androids are this, and robots are that” is in the context of a work which has beings referred to by both names, and where the distinction between the two is considered significant. And I strongly suspect that such works are both rare, and not very influential.
Well, I think that by sparing Deckard’s life, Roy showed he was more humane than the humans who created him.
But thanks to you I did a little checking and found that my understanding of the term “tragic hero” has been skewed, so I take your point.
Or in a series of works that have one or the other. Like those that cumulatively gave me the impression I have.
But if you want an example with both terms, Star Trek: The Next Generation, “Deja Q.” Data, a non-biological artificial person, says explicitly that he is “an android, not a robot.” (There are not many “robots” in Star Trek, but when they appear they look like machines, not people.)
I reckon that influenced more perceptions of the difference than any given armful of Silver Age comic books. And it was unsurprising to me, since it meshed perfectly with the understanding I already had by 1990.
I’ve heard this said before and I’m always a bit puzzled by it; humans spare the lives of their enemies and of dangerous animals etc regularly. Not always, but regularly. Roy did no more. He is distinctly inhumane to some. He showed that he was at least as humane as humans on at least one occasion. That is all.
Maybe Roy just wanted someone to be with him when he died.
As a combat model, Roy likely wasn’t designed to be humane. In sparing Deckard, he transcended his design.
Agreed as to both.
He showed something that resembled grief when Zhora and Leon were “retired”, and a genuine sense of affection and loss when he found Pris’ body – the others may have been more fear than grief, fear that Deckard was closing in on him, but with Pris it looked different. So he sought to punish Deckard, but at the end, he found his humanity when he lifted Deckard to the roof and chose to expire in front of him rather than kill him.
It showed that Tyrrelco’s notion that giving them short lifespans was both correct and wrong. I think the underlying message was that deep genetic/epigenetic manipulation of humanoids to create physical castes was (would be) an ill-conceived strategy, because you cannot eliminate the individual’s drive to grow outside of their defined role, and to do so in the first place would lead to the greater stagnation of the human race.
The original story was based on Dick’s research for another story, during which he read documents and records from third reich archives and was struck by one officer’s complaint, “we are kept awake at night by the cries of the starving children”. Hence, he imagined people (or convincing simulacra) devoid of feelings. The writers of Blade Runner never read Dick’s actual book, but they seemed to capture its fundamental theme quite well.
Or, it could be that Deckard himself was a replicant and that is why Roy was reluctant to actually kill him.
I think we can all agree, at this point, that there is absolutely no question whether or not Deckard was a replicant.
Yup, that’s at least as settled as balrog wings and boxers vs. briefs.
What’s wrong with “skin job”? “Synthetic human” is maybe too close to “synthetic person”, which was Bishop’s preferred term and related to human-appearing machines.