How do I justify a world where drones are commonplace but computers remain large?

Question

In those golden age sci-fi comic books, we often see the main character with some type of robotic companion (maybe controlled by an electronic computer or some organic "brain"), while computers remain large.

My understanding of robotics is that small, portable electronics are needed for them to exist. The same type of electronics that would enable smaller computers.

How could a world exist without desk-sized computers, but with some degree of viable robotics? Handwaving with non-scientific factors (like magic) is not an issue but I'd still like to know what exactly the "magic" would change in this world. Please let me know if more information is required.

Answer 1

Animal controls.

pigeon controlled bombs were a thing. They worked. they are however very limited in what they can do and of course each pigeon was single use.

Basically you have trained animal handling the controls. Electronics are way cheaper, lighter, and easier but some things can be done without them. For a willing amount of handwavium you could make animal controlled drones.

Answer 2

We're not actually that far off such a situation now. The push to run everything in "the cloud" really means "run it all in a giant datacenter somewhere you can't see it", and giant datacenters aren't all that different from your classic mainframe.

Most people still have a compact bit of personal compute power in the form of their phones, and what's left at the moment tends to be a laptop but even that doesn't need to be the case in your world... we already have stuff like office "productivity" applications like spreadsheets running in datacenters, and stuff like the late unlamented google stadia showed that games could be left there too.

Your world doesn't have intermediate-sized computers simply because they aren't needed... phones, terminals (which may be phones) and datacenters cover all the roles required.

There are technology limits for drones, of course, and having tiny capable drones with long battery lives does imply modern or future technical capabilities. The exact timescales and technological abilities of "golden age" settings were often a bit fuzzy, so as you haven't made any tighter restrictions on times and technologies, this seems like it would suit you.

Answer 3

Drones don't have to be small and highly computerized.

Way Back When Computers Were Huge, radio controlled airplanes existed. The control units were a bit large and heavy, and you needed line of sight to control them, but the highest tech electronics were transistors.

And as far back as the 1950s, when computers barely existed, the Ryan Firebee was buzzing around for the US military.

Answer 4

You can have drones with no electronics at all.

The Goliath Tracked Mine

These little machines are terrifying drones. They would crawl across No-Man's Land with 100kg of high explosives and blow up in the opposing trench. They did not have computers. Some had electric motors but most had gas engines. It was all mechanical.

You can get a fair bit done with clockwork and punchcard type memory. I like the idea of a robot buddy made from a decomissioned Goliath Tracked Mine, with the 100 kg of explosives removed to make room for a computer of sorts. And a voice simulator which is really a stack of 100 little phonograph records each with something different to say.

Answer 5

Biopunk!

Who says that drones have to be robotic? They can be organic instead! Either whole thing is organic, or just controling parts. A brain in a jar scenario, except that the brain doesn't exactly need to be natural, and can be grown in a vat. In our history we advanced microprocesors way faster than chemistry, biology, genetics and similar life sciences, but that doesn't mean the same must be true for your setting. If advancements in those fields were way ahead of computer knowledge, you could get scenarios where people could buy artificial servants and still have room-sized "personal" computers.

Answer 6

How about a breakthrough in analog electronics without a corresponding breakthrough in digital electronics? Analog electronics is great when you need a fast if somewhat less precise result. Digital electronics, however, provides extreme precision, but sometimes takes longer to obtain it. This is true even in the real world. A drone does not need extreme precision, and it is conceivable that one could run on an analog computer. It knows only that it must go approximately North for approximately 500 meters at an approximate altitude of 800 meters, so an analog computer may suffice for that. However, there are still large digital computers for doing precision work, such as calculating the 20,000th digit of pi.

Answer 7

QUANTUM COMPUTERS

You are comparing two different orders of magnitude in processing power.
Integrated Circuits (IC) may suffice a drone for sensors, attitude control, communications, etc. Even for basic mission fulfillment tasks.
Their processing usually requires little power, is small, lightweight, cheap.

Then you have proper computers, quantum computers (QC), homes to AIs. Mean beasts that may solve NP problems so blazing fast they may burn the fingers on your keyboard.
They drive billions of devices connected (aka IOT), set missions both to drones and human crews, assess sensor reports, analyze and upgrade themselves constantly. They run the world.
They run on the fastest possible hardware for the task: quantum computers. This technology just does not handle well miniaturization.

Note that current quantum processors are still a far cry from the marvels of your future.

In 2019, Sycamore completed a task in 200 seconds that Google claimed, in a Nature paper, would take a state-of-the-art supercomputer 10,000 years to finish. Sycamore processor - wikipedia

But Sycamore has a measle, pitiful, laughable 53 qubits.
QC processing power is measured in Giga qubits, a whole other world.
They need a complex apparatus to work, a controlled environment, are very expensive and they are large.
Most of all they need cold.

You see traditional racks with 1000W processors can be quite a challenge too cool.
But in quantum chips power consumption is very, very low.
There is a catch though, quantum processors need to be kept at a very low, very steady temperature. Even small temperature (10 millikelvin!) increases can render the entire system unworkable.

To keep systems in a quantum state, designers have to minimize the risk of anything disrupting the fragile position. The slightest temperature increase can mean that atoms and molecules move around too much, potentially causing a quantum bit (qubit)'s voltage to spike, and flip from one quantum state to another.

See Cooling Quantum Computers

They also need to be shielded by external factors that may induce decoherence, the information in the quantum system can become randomized or totally erased. That is in general referred to as noise.

Noise refers to the multiple factors that can affect the accuracy of the calculations a quantum computer performs. Quantum computers are susceptible to noise from various sources, like disturbances in Earth’s magnetic field, local radiation from Wi-Fi or mobile phones, cosmic rays, and even the influence that neighboring qubits –the building blocks of a quantum computer– exert on each other by mere proximity. These disruptions cause the information an idle qubit holds to fade away.
Noise in quantum computing

Quantum error correction (QEC) deals with noise reduction but it is necessary to have the processors in a controlled environment.

In short the need for extremely low, strictly controlled temperatures and a safely controlled environment prevent miniaturization of QC.

But they are great, every mad scientist wet dream.
And are sold with all sorts of blinking leds for the retro afficionados.

For reference:
Quantum Computing

Answer 8

Money.

Small electronics components are more expensive, especially if those components work in a different way (RAM vs hard drives come to mind, where the former is typically measured in GB, and the latter in TB... or SSD vs HDD).

It's also certainly plausible that AI could use different electronics from traditional computers, that would allow it to be far smaller, but would also come with a substantial price tag (or perhaps it just isn't able to perform the functions of a typical computer, much like a human wouldn't be a great substitute for a computer).

Convenience and necessity.

This is closely related to price.

A personal drone needs to be small, but a personal home computer does not. That alone could be a significant driver for the sizes of each.

The sizes and prices of these would still be correlated, but it's certainly plausible for small-ish drones and large-ish computers to co-exist.

As a real-world example, consider that mobile phones and laptops co-exist, along with desktop computers (although those are getting far less common).

Although if you have a world where traditional mobile devices don't exist, this alone probably doesn't explain that. On the other hand, drones are sometimes presented as a replacement for traditional mobile devices, so that may explain why they don't exist or why they're uncommon.

Power.

In the modern day, AI is one of the more processor-heavy things people use computers for. Although it's also worth noting that training AI tends to be much more processor-heavy than using AI (and training would likely happen elsewhere, rather than on the drone itself).

I can certainly imagine a world where the primary use for computers tends towards even more processor-heavy (or memory-heavy) tasks, such that many people would just need rather bulky computers to do what they need or want to do.

Answer 9

There is a large computational difference between training a model and deploying a model. Here are a few examples.

1. Self driving algorithms are trained on millions of hours of data, requiring the training software to comb through all of that, which takes a very long time with very high powered machines that require a lot of cooling. But when done, they have some matrices that can be deployed into cars with much less compute capacity.

2. A human adult (the biological kind as well as potential robotic kinds) has undergone years of training to recognize and react to many different events in its environment. Half a lifetime of experience has been encoded into neural circuits that can respond to future events in real time at low cost.

Your drones could just be low-power implementations of a given weighted training matrix that's already been generated by the large computers. All the drones need are the weights encoded in the matrix. And the constant learning and adapting going on in the large computers could be sent periodically to the drones as updates to their models. But figuring out what the weights should be, and whether they should change to adapt to a changing world, takes real processing in a cooled server farm.

Answer 10

You could follow the Star Wars model (or my 100%-unresearched idea of it, anyway):

In the Distant Past, some Unspeakable Horror came about due to unrestricted use of computers, so now the only "computers" are either 1) extremely specialized non-turing-complete machines, or 2) exclusively controlled by a sentient intelligence (droids).

Whatever the underlying event/reason/whatever, this rule is so absolutely universally accepted that no enforcement mechanism is needed - it's simply Not Done.

Answer 11

Radiation-Induced Bit Errors

Your world could have high background radiation, enough to cause bit flips in small portable machines that try to solve complex tasks. The radiation could come from natural minerals like uranium or thorium, fallout from an ancient nuclear war, or astronomical sources like Jupiter's magnetic field ion trap or a nearby excitable star.

The only place you could run real applications would be on underground servers, in shielded chambers made from carefully purified materials. Mobile devices could follow instructions sent by the servers, but not do much local compute before they locked up or started giving bad answers.

As an example, a typical phone today has a billion transistors, which is only feasible because the error rate on those transistors is incredibly low. Even today we can't hit a thousand qubits, because the error rate is too high.

(Biological systems can deal with surprisingly high radiation, like how the bacterium deinococcus radiodurans can re-knit its chromosomes back together after radiation induced double strand breaks. Biology that evolved on a world with high background radiation would naturally develop this sort of resistance, like we evolved to deal with our corrosive oxygen atmosphere.)

Answer 12

Have some sort of disaster happen in the distant past that wiped out a lot of equipment and knowledge. Like a nuclear war. The computer chips for the drones could be specialized chips for drones built in one of the very few factories that survived the war. They don't have the equipment or knowledge to build other computer chips.

Answer 13

It could be that silicon doesn't exist in your world, or that it is very low on certain conductors like gold or copper.

Most common computer technology is based on the ability to make small programmable electronics. If the materials we use are not readily available, different methods would need to be found. If there was also a lot of background EM or magnetic storms, then all those thin, fine wires in a microchip would act as antennae and shunt power in to circuits where it was not expected to be. It may be possible in your world that a move away from silicon and copper electronics is necessary.

One option is organic computing. While organic molecules can't handle extremely energetic particles, they are fairly unaffected by magnetism and lower power EM. Nuclear radiation more or less does to organic molecules what random EM does to electronics - it puts signals where they shouldn't be and alters the encoded instructions. If we're talking EM and not nuclear radiation, organic computing may be useful. Organic computing is very wide, but very, very slow. You can do tons of operations at the same time and store massive amounts of information, but doing those computations takes much longer and retrieving the information is not easy. It is also, well, organic, so it is wet, heavy and big. Your big computers might be organic units.

Another option is crystalline structure. It is tricky, but it is possible to form a multifacted crystal that has a large number of reflective planes. As you twist it under light, the various planes can be separated by fractions of degrees and can each reflect something different. Retrieving doped crystalline-coded information is easy, but forming the crystal and 'programming' it is fairly difficult. Changing the information on the crystal once formed is practically impossible as well. It would be a storage medium only, however it could hold the basic elements/instructions/pattern of machine sentience. It could be the core to the AI. It could be light, small and fairly resilient and could hold quite a bit of information. They also don't require all that much power since they rely only on light hitting them from various angles. You do need very accurate gimballing and rotation mechanisms to hit the various planes exactly. Since altering a reflective plane would be practically impossible without destroying much of the crystal, a doped crystalline core would be something you could rely on, preventing robots from changing their programming. They would be essentially golems, with their instructions etched in crystal in their head.

If you had some organic long-term programmable storage and a crystalline core, so long as you have energy solved, cybernetic robot/drones might be workable. Like any economy of scale, once the parameters are determined, mass production wouldn't be far behind. While designing an AI core might be a Herculean task, once the matrix necessary is known, creating them wholesale might be simpler.

Answer 14

The first drones capable of powered flight predated electronic computers and the Wright Flyer

While not exactly commonplace at the time, Samuel Pierpont Langley's series of Aerodromes (one pictured above was photographed in flight by someone named Alexander Graham Bell) flew in the last decade of the 1800s. One might imagine a world in which someone like Langley managed to improve the reliability and usefulness of the Aerodrome.

https://siarchives.si.edu/collections/siris_sic_7894

Answer 15

Different Technology in both of them. Personal computing has advanced to a new tier of technology that cannot be shrunk down to drone like sizes. But the operation of a drone doesn't require the type of processing power now available. so they continue to use traditional microprocessors, but humanity would rather not be bottlenecked by old technology.