Topic: Server Sky - Space based mining? (Read 1490 times)

Not sure the relationship between kHz and heat production in this context, but based on the BFL BitForce stats it looks like 1 watt = 1 Ghash/s (or 1Ghash/J).

Keith estimates we can radiate 100-200 milliwatts per square centimeter.

With a 16 cm^2 chip-space configuration, this should net 1.6-3.2 Ghash per thinsat.

Note that he's talking about spreading the chip area around in hundreds of pieces on the same thinsat. Each of those would be in the Mhash (mW) range. Come to think of it that might be kHz since it is parallel processing (I guess 1000 processors at 1 kHz = 1 Mhash/s).

Cost-wise this seems like it compares well against BitForce ($274 per 5 Ghash/s), as long as we are talking in the range of $100 or less per thinsat.

To make this work you need to use the electricity on Earth. Heat dissipation would be impossible for the chips people are using today. Perhaps you could run an ASIC in the KHz range.

EDIT: A great place for sci-fi trying to become science fact is Centauri Dreams from the Tau Zero foundation, run by an ex-NASA engineer.

You've been polite so far, and I appreciate honest skepticism as a counterpoint to natural enthusiasm. 

I stand corrected on the cost. The Ariane 5ECA launch cost is $120M. On the other hand, SpaceX Dragon 9 is roughly comparable at $54M.

So, more realistically: If we can launch 3 million thinsats producing 4 usable watts each for $60M, that's $20 per thinsat or $5/watt. Assuming a rate of 10 cents per kWh,  that takes about 5 years to pay for itself. (That's way off from what I quoted -- I think Keith was actually intending to make a case with regards to theoretical limits of rocket launching tech, not near-term costs.)

To achieve something more like $1-$5 per thinsat would require better economies of scale for launching. SpaceX is already getting really good at this kind of thing, but more demand would improve matters. So the case is stronger if mining becomes really big business (which it could, if bitcoin prices get high enough), or if bundled in with other significant revenue sources (e.g. providing cheap broadband internet for the developing world).

Another way to bring down the cost is to make much thinner sats, but combine them with space debris already in orbit. (Their minimum size is partly due to need for ballast.) Alternatively, sand launched from the lunar surface could be used. (Unlike the earth, electromagnetic launchers are actually physically plausible from the lunar surface.)


Heat naturally converts to infrared radiation. Areas of the craft not facing the sun or any other nearby radiation source can be cooled to around 2.7 Kelvin, although the speed at which they are cooled depends on emissivity of the surface and how hot they are running.

http://server-sky.com/cooling
https://en.wikipedia.org/wiki/Black-body_radiation


The main reason to prefer an earth orbit is the low latency from the speed of light. Orbiting the sun could be useful for an interplanetary civilization, or if you just want huge amounts of raw processing power to do stuff that isn't timing-critical, like molecular simulations. However for bitcoin purposes you need to be able to hash a recent enough copy of the blockchain, so anything past 10 light-minutes would probably be out of the question. OTOH, we are only 8 light-minutes from the sun, so a swarm in the inner solar system could be mostly within useful range.

What an awesome concept.  You use the electricity in space, don't try to beam it back. 

I've been thinking about a self-contained unit that runs straight of a solar panel, turning the electricity straight into bitcoins without inverters or batteries or wires.  This takes that concept to a whole new level. 

Is there any reason they need to orbit earth?  Why not put them in orbit around the sun?  They can form their own mesh communication networks and work together as a phased antennae array for long distance communication.

I don't understand how you could cool a computer chip in outer space. Theoretically, the solar panels collect the light and maintain a supply of electricity. The chips use the electricity, but expel most of the energy in the form of heat. Where does that heat go? How do you dissipate that much heat in a vacuum?

Hmm, not sure if I want to get into a slanging match with a space cadet, but I'd like to point out that your link quotes the payback in energy cost of the fuel used to to launch the satellites. Not so much an apples to pears comparison as a grapefruit to raisin. Fuel cost is a tiny proportion of the cost of a launch. Must try harder.

Go away, troll.


This is not science fiction. The wiki has analysis of launch costs for thinsats. Payback time is about 3 weeks. http://server-sky.com/LaunchEnergy

I wouldn't go quite so far. NASA did a study back in the 1970/80s (following Gerard K O'Neil's High Frontier book). and came to the conclusion that Space Power generation just did not have a ROI. They used pretty optimistic projections for launch costs too (if anything its more expensive now than it was in those days!)

All this "Server Sky" is suggesting is that there is no need to beam the power back to earth, just use it in space. It certainly reduces the technical requirements, and would perhaps halve the cost. But half of outrageously expensive is still outrageously expensive.

Though looking further into the future the "Singulantarians" predict enclosing the sun in a Dyson Sphere of pure computronium, and transferring human existence into the cloud. Charlie Stross did a nice take on it in Accelerando. I haven't read his Singularity Sky yet, but I'm guessing this is where the term "Server Sky" originates.

TL;DR Sci-fi is fun, but don't confuse with reality.

This is fucking retarded even for the speculation forum.

Did you read the wiki yet? 

The major awesome thing about this is that the thinsats are extremely thin, and thus cheap to launch per unit of processing power. The heat output would be radiated into space, dramatic surface area works to our advantage there. Solar energy provides ongoing power at no cost, so as long as the thinsat lasts you would get at least some hashes (even if not as many as the next-gen thinsat provides).

Also, there is the matter of scalability... Satellites scale remarkably well compared to anything ground-based. At some point we start hitting the theoretical limits of ASICs, and the only way to compete is get bigger. And bigger means the cooling surface area needs to be bigger. At that point, space based (or perhaps lunar based) methods start looking much more attractive -- and they are certainly better for the environment.

Bitcoin mining takes up way too much power and as a results outputs way too much heat. Beside that issue, considering lead times and how long it would take to get those up there, by that them the hashing power would probably already be obsolete.

Keith Lofstrom has been working on an idea for a while now to put wafer-thin satellites in organized arrays in orbit. They would gather solar energy and cool by black-body radiation into space.

He's currently looking for people to get involved (tech geeks, enthusiasts, critics who know their stuff), and funding sources. Well, it made me think of bitcoin, since mining is an obvious funding source, and we have our share of software and hardware experts, and enthusiasts too.

It's a rather detailed proposal. Read the wiki if you have questions about how it would work: Server Sky - internet and computation in orbit

My thought is that an array of these for a reasonable price per unit could be crowd-funded, with a good ASIC chip embedded in the design. It would keep generating hashes for as long as the satellite lasts, with no ongoing electricity cost. You could own your very own satellite, or perhaps hundreds of them.