> I wouldn't put Cixin Liu's trilogy as hard hard sci-fi, as it invents a few more, but all in all, there aren't that many "fantasy devices" there. It definitely has a flavour of hardness.

Even that much is being overly kind (at least to the first book - I gave up after that). Hard SF introduces a small number of well-defined ideas and then works through their logical consequences. Science fantasy introduces magical devices with particular consequences as and when the plot demands. The magical VR system and magical particle computer - essentially all the SF elements of Three Body Problem - were examples of the latter.

The second two books definitely approach it more closely.

I guess the piercing of my suspension of hard sci-fi disbelief happened in the 2nd book(?), where post-Fall they rebuild into a militarily space capable society with relatively advanced weaponry.

If you're going to take 1,800 pages (English translation) to lay out your series, and you can't spare a few for 'well, now we're building giant fusion-powered warship fleets'? That's not anything approaching hard sci-fi in my book.

Yes, all the technologies used on the ships are extrapolated and believable, but the realignment of society, economics, and the organization of such a construction effort boggles the mind. And to say 'it just happens'? That's kind of like going from WWII to 'and so we landed on the moon' in a single page.

Fair. The time jumps were a bit jarring, though I guess I got used to those after a similar one pulled in Seveneves. I wouldn't mind if Cixin Liu shortened some parts of his book and used the reclaimed space to fill in those time jumps a bit, though.

They get to invent as many "impossible" things as they want, really. But they all need to have plausible-sounding explanations (given our current level of scientific understanding), and they need to act in a consistent, predictable manner, including consistency with known laws of nature - e.g. if you have antigravity, the default assumption is that it obeys the inverse square power law, and if it doesn't, that needs to be called out and have some explanation of its own.

I'd hew closer to TeMPOraL's definition.

On the one hand, one obviously needs some technological leeway to write a fun novel set in the future.

On the other, if every roadblock is responded to by suspending scientific treatment, it doesn't seem like that's a good faith "hard" effort.

I think Ringworld is instructive. The Wikipedia article specifically calls out technical inaccuracies in the original... inaccuracies Niven specifically made an effort to patch up in later books.

So the count of intentional lapses seems as decent a proxy for hardness as anything else.

I would consider Ringworld to be borderline hard sci-fi, but precisely because it has some reasonably sounding explanation for his technomagic. E.g. the General Product hulls - them being a single macro-molecule is a reasonable pop-sci explanation for their properties, even if the book doesn't explain how such a molecule is made. Does that count as an intentional lapse?

What are your thoughts on Diaspora? (Greg Egan)