Comment by strainer | Hacker News Reader

Comment by strainer | Hacker News Reader

Parent Full thread

Good point about solar just transferring the energy. Im not sure how practical it is to brighten so many constructed surfaces like pavements and roads it makes a big impact and is tolerable to live with. An estimate that only 2.7% of the worlds land is urbanised, and 2/3rds being ocean... increasing its albedo significantly could be difficult. An advantage of space shading is it doesn't rotate out of play and can be concentrated a bit away from the edges of the globe where light is a bit more likely to reflect away.

Agreed it would be difficult and heavy to add minature gyroscopic and solar cell, comms and control unit to the 16 trillion disks proposal, since they are just 60cms wide. That plan seems basically like throwing confetti into L1, where it might get drifted away by solar ejections, out of stable orbit, in matter of years.

Since each of those disks is 1 gram and about 0.25 square meters, that scheme involves total of 4 trillion square meter shades orientated randomly. If shades are squares for simplicity and 8 meters in diameter, so 64 meters square and similar material would be 256 grams, add some stiffening and a control unit for say an extra 150 grams. Maybe a clever unfolding feature like an insects wing. Its a bit more material to launch but probably less than double at some control/shade ratio, and the tracking shades will block light about 30% more effectively than randomly orientated ones. Works out around 50 billion of the 400gram 8 meter wide shades with mini solar-comms-gyro units to produce. So I'm still overconfident in the practically of this kind of scheme, especially since cost to space has been minaturised since they previously had expert attention.

I have to admit its just one of those hunches Im hawking of late, cheers for following :)

I re-read the paper in detail. There's a lot more going on than the commentary on this page assumes, with some impressive calculations, but lots of unaccounted-for (and un-invented) elements, and even more unanswered questions.

For example, the author does assume 1gram (2ft diam) shields, but also specifies the necessity of precise angular orientation, and relatively modest location control / steering. This is to be accomplished by an unspecified number of "control satellites" using aimable mirrors and passive radiative pressure, which keeps the discs simple and the weight at 1 gram. Required disc area is ~7x the required total shading area, due to indecent angle and required transparency and spacing. A new gps-like network of "navigation beacons" will also be required to maintain spatial reference. He then says that to track individual disk locations/ orientation, that each disc needs a GPS-like nav receiver, 2 cameras, some processing/ communication ability, and a power supply, presumably solar. But he doesn't account for (or even reconsider) that there will be added weight and cost due to this.

He also doesn't consider the cost of inventing, manufacturing, and deploying the control satellites or the new type of nav beacon network, or even estimate how many of each will be needed. (GPS tech won't work, but if he knows that, he doesn't state it.)

There's other questionable assumptions, like a launch cadence of every 5 minutes for 10 years to leo, and that cost of launch will roughly equal cost of fuel due to the scale.

He also estimates the method of transfer from leo to L1 will require delta-v of 1km/s and assumes it will be solar powered / ion propelled, but leaves it at that, also without scope or cost.

It's interesting, but less convincing than I expected, given all the missing details, and also makes me more convinced that risk assessment, management and mitigation haven't thoughtfully been considered. That said, the shading, geometric layout, and required material properties calculations are quite impressive.

Slightly tangent, but in talking about the risks and unintended consequences if such a project, lots of potential issues have been mentioned (which is my bigger overall concern), but given the specs, a new one occurred to me: this design starts to look a lot like what might be required for true singularity type AI.

+/- 16 trillion self-powered, locally communicating "cellular" elements, each with some minimal sensory and processing capability. Add to that the mobility presumed by the author, and the fact that it is literally designed to block the sun, and you've got everything you need for a matrix rewrite...

Thanks for the info on the paper - I haven't read it, just seen a number of reports of different approaches to space shielding, that seem to establish its a physical/industrialy modest possibility (for a global project). It would be very early stages for the details of any project, I expect a well funded competition for competing designs and to be surprised and educated by actual designs. It just seems the prelimary designs already discussed establish its physically not unrealistic if launch capability can be sorted. 20 years ago ideas of huge wind turbines planted at sea and even floating were considered by most to be impractical or unproven, but the basic facts supported their possibility as major power sources - the density of air, the intensity of wind, the existence of resilient ocean hardware like oil rigs, the existence of very large aircraft wings. There was no end of unresolved details that could be counted to dissuade, but the broad limits were already visible and attractive. Rather than a Manhattan project to race to those limits, industry has grown slowly to approach them with not that much special funding to help.

The control coordination of trillions of L1 satelites doesnt strike me as a huge challenge to people already working in similar fields. I do some work on simulations and read about related technics. Control units just get individual and group call codes that they can respond to, in timed windows if helpful. A number of manager satellites beam signals to sectors and listen to responses. They could likely often pinpoint individual transmissions themselves with modern radar style tech, although trillion is a large number, sheilds will rarely eclipse one and other because they would be very dispersed and relatively small. They dont have to relay all messages individually through neighbours but perhaps could. The options for implementation are extensive but I believe familiar to network designers. They don't really have AI processing to do, its just a matter of maybe reporting their neighbors numbers so position can be determined and relayed so they know where to 'light-sail' and can receive schedules for when to let light past - if that becomes necessary or is deemed advantageous. Im just rambling but there are loads of possibilities for how to organize them. As a multi-node processor it would be rather slow because of the average latency between nodes being many kilometers, rather than millimeters to meters in a supercomputing cluster.

Reply on news.ycombinator.com