Many moons ago, I did a post on ways to make your sci-fi technology more realistic. My day job is in aerospace engineering and am I also a serial nitpicker with more pet peeves than you can shake a stick at, so if you thought that post was the end of my ranting – well, it very much wasn’t.

I want to open with the same caveat as last time: sci-fi doesn’t have to be realistic to be good. Fiction is a playground, and the whole point of speculative genres like sci-fi is to speculate – to dream about what might one day be possible, regardless of whether it makes sense right now or not. Sci-fi authors were writing books about space travel back when the common view was that rocket science was a crackpot field that wasn’t going anywhere. When you write sci-fi, you risk engaging in the long tradition of accidentally predicting things that your descendants will look back on as either prophetic genius or quaintly, hilariously, embarrassingly wrong, and both options are beautiful and valid.

However.

Sometimes, you would like things to hold a little more water. Maybe you’re writing about modern-day technology, and you need a little background on the current state of the art. Or maybe you’re writing hard sci-fi – a subgenre that puts an emphasis on scientific realism and explaining the details of the technology in your world. Maybe you just feel like it!

So whatever the reason, here are a few more science and engineering things you might want to know.

Your inner ear is a lying idiot

The first thing you learn when you work in aviation is that the human vestibular sense is pile of steaming hot garbage. At ground level, it’s so reliable you’re unlikely to notice anything strange, but the moment your plane takes off, it all goes to hell. Your sense of balance is basically an educated guess by a brain that will continue to think it is sitting in a chair on the ground at sea level despite copious evidence to the contrary. I guess we can give it some grace since nothing in our evolutionary history could possibly have prepared us for a world full of flying machines, but still – it’s annoying.

Understandably, this is kind of a big deal for pilots and astronauts. If you are sitting in a bird-shaped piece of metal 30,000 feet in the air, or rocketing through space in a glorified tin can propelled by explosives, and your brain still thinks you’re on your living room couch, you absolutely cannot trust yourself to steer the aforementioned metal bird or tin can based on inputs from that brain. Pilots and astronauts often feel sensations that are at odds with what is physically happening to them, and they are trained to largely ignore those sensations and fly based on the data on the instrument panel in front of them instead. This can create problems, of course, if those instruments are inaccurate for whatever reason (hint, hint – you’re a writer, you can use this to make a mess for your characters!). But most of the time, they trust their instruments, which take good care of them in return.

So if you’re describing what at character is physically experiencing during a spaceship chase scene or as an airplane careens toward the ground, consider doing some research or adding subject matter experts to your beta reader pool to verify that your sensory descriptions really make physical sense. You could show a passenger freaking out and causing trouble because they’re wrong about what’s happening, or ramp up the tension as the pilot momentarily puts too much trust in their sense of balance rather than their instruments and makes a catastrophic mistake. Or it could be just a fun way to add some color and interest to the description of your scene. Of course, the average person isn’t going to care much, but if that was enough to stop you from delving into this, you wouldn’t still be reading this post, would you?

Machine learning only as good as its training data

Recently there was a story in the news about a group of US Marines tasked with sneaking up on an AI whose algorithm was trained to detect people. It was, in fact, trained on six days’ worth of images of those exact soldiers walking around. And therein lies the key to what followed: while teaching the AI what a person looked like, they only taught it what person looks like what they are acting like a normal person. When the soldiers successfully snuck past it, they did so with tricks like somersaulting towards it, pretending to be a tree, or hiding under a cardboard box (according to the article, “You could hear them giggling the whole time”). The AI saw them, but it refused to admit they were people, because the only information it had about people was what they look like walking around normally.

I’ve worked with machine learning algorithms, and can confirm this is how it always tends to go. When you’re trying to diagnose a problem with a machine learning system – or figure out how to break/evade/undermine it – the surest way is to examine the data it was trained on and ask yourself, what has it not seen? This is made even more convenient by the fact that a lot of algorithms out there right now are purpose-built for very specific applications; take them out of those applications and they largely won’t be able to infer what they’re looking at. Not only are these algorithms unable to infer new information about something most of the time, but they’re incapable of even conceiving that there exists information they do not know.

Of course, this could change. AI technology is developing rapidly, and we’re likely to see it advance by leaps and bounds over the next few decades. But if you’re an author and you think you’ve written your characters into a corner because the evil AI they’re up against is all-seeing and all-knowing, then I have happy news for you. It doesn’t know everything – but it probably thinks it does, to an almost ridiculous degree, and you can use that.

And hey, you know what else has a surprisingly limited body of knowledge? People! Which brings us to our next item – one which is far less likely to change.

Specializations are narrow

I am hardly the first person to say this, but complaining is my specialization so that’s not going to stop me.

Your Martian geology expert is probably not going to be able to place the strange bacteria that show up in their core sample in the appropriate phylum. They will need to call in an evolutionary microbiologist who specializes in extraterrestrial life. That person will be perfectly free to come and help, because they will not be busy dealing with a zombie plague on Earth – that’s being handled by the virologist next door. And when the spaceship bringing this Martian microbiology A-team (who are definitely not infected with a zombie plague) to Mars is mysteriously sabotaged, that microbiologist is going to be as useless to the process of preventing a reactor meltdown as a professional chef. For that, they’ll need a nuclear engineer, probably one with experience in the specific area of space propulsion. And when the team phones home to relay their discovery that the Martian bacteria are behind that zombie plague back on Earth, the virologist is going to have to call a coworker with a different specialization, because their research is on viruses and this is not that.

The point, if you don’t see it by now, is that “scientist” is an extremely broad category. Every STEM field is a massive entanglement of hopelessly narrow and specific areas of study, packed to the brim with people who are really good at understanding That One Specific Thing and almost nothing else.

Putting a limit on the things your scientist characters are good at does two very cool things for you. First, it generates conflict: to solve the problem, they need to figure something out that they don’t have any background in, or get help from someone else. And because conflict is neat, that person may be, say, very far away, or perhaps detest them for some dramatic reason. Congratulations, now your book has even more tension, and you barely even had to think that hard! Second, it gives you an excuse to expand your cast of characters, which opens up avenues for even more fun things, like building out additional subplots, or highlighting different aspects of your themes or worldbuilding.

This also heavily informs my next point, which is…

Science is collaborative

How many times have you read about a character who works as some kind of technician or mechanic (why is it always a technician or a mechanic?) and therefore is an antisocial hermit with no friends who doesn’t play well with others and has to learn a valuable lesson about How Human Emotions Work and the Power of Teamwork in order to make it to the end of the book? Have you ever looked at that character and wondered how the hell that person made it through their education? Yeah, me too. I’m glad I didn’t have to write lab reports with any of them, that’s all I’m saying.

If I could wave a magic wand and get rid of one stupid sci-fi trope once and for all, it would be this nonsense. Nothing gets done in STEM without a team. Nothing. We brainstorm in groups, build things in groups, test things in groups, operate things in groups, fix things in groups, write reports in groups… I could go on. I once went to a welcome reception at an engineering conference where the party was still in full swing as the venue was attempting to clean up around us. I watched the bar being wheeled out of the room as we all stood chatting. (They tried to take the desert table too, but there was an uprising.) Yes, a lot of us are introverts, but the thing about introverts is that when we find our people, we go absolutely buck wild.

In closing: give your surly, brooding starship technician character some friends. Please.

Now I want to hear from you – what’s something about your field that fiction tends to get wrong? If there’s something you wish other authors knew about your day job, leave a comment here or over on Instagram so we can all take your advice. (And maybe I’ll do a post compiling some of them in the future!)

Happy writing!