This kind of thing could replace some uses of blockchain. The pattern of clouds on earth every day is unique, so anything that needs a timestamp with 1-day resolution could just tie it to a hash of today's cloud image pixels gathered from many receivers around the world, each taken at noon from the POV of a given geostationary satellite. That way one could prove that an event took place no earlier than today, especially if you hash each cloud photo with the hash of the one from the day before, which creates a cloudchain. The chain's ordering is verifiable from multiple sources and created wholly external from any human agency or pseudo-random number generator.
Speaking of which, you could also use the pixels (or some permutation of them) as a random number seed that changed every day.
Is it any better than any other source of random or extremely difficult to predict data? Seems like you could just use data from the stock market or some radio antennas or telescopes.
The cloud data is independently verifiable and somewhat globally accessible, which is the main benefit. In a court of law you can refer back to any number of facts to corroborate the data.
But yes you could use stock prices or sports scores to the same effect.
The stock market is human-generated so somebody with enough money could nudge it in one direction or another. Radio antennas or telescopes could work too; I'm just not aware of anything they see that meets some definition of "random" except for maybe the surface of the sun, and the pattern of clouds on the earth.
In this case you have to trust whatever oracle you are using. The images need to be verbatim so everyone would have to use the same centralized source, which means it is not the same as a more decentralized source.
Venusian clouds are easy to see with earth-based telescopes. Problems: There's not a lot of variation -- it's solid cloud cover. And Venus has phases and we never see it as a 100% disc (the Sun is in the way).
It's actually quite difficult to observe clouds on other planets from earth (or earth orbit) due to resolution limits of the combination of reasonably-sized telescopes and usable wavelengths.
Coincidentally I used weather information to generate a random number, I used {{PollenCount}}{{LowTempCelsius}}{{HighTempCelsius}}{{WindSpeedKph}}{{Humidity}}{{UvIndex}}{{SunriseAt}}
I know this would collide eventually but hey its just a hack for my IFTTT applet [1]
It would only work where there was no motivation to show an event happened earlier than it did, but rather as late as possible. Such as the proverbial kidnap photo where the victim is holding a copy of today's newspaper.
That's the point, it proves the day. I think there's still a challenge to solve which is mapping the points from which to construct the hash, because you need a low enough quality to agree on pixels but a high enough quality to get enough entropy to be strong as an time consensus authentication source.
Repeatable random simlations that cannot for some reason just check the date and use the date (and some salt) for the RND seed? What simulations are we talking about here? Must be some obscure things?
So essentially instead of hash-based computer proof of work, you're suggesting looking at cloud patterns (say, as the sun's proof of work)? Sounds like a great way to save energy, if we can ensure image authenticity/integrity.
The vulnerability is the one satellite everybody is looking at. But that could be mitigated by having a network of ground stations looking up and asking "is it cloudy overhead?" With enough of these sample points scattered around the planet and correlated with the satellite's imagery, one could make hacking the satellite impractical.
Why not taking the clouds of Jupiter? Then you could use telescopes and multiple satellites alike because that far away parallax shouldn’t change a lot.
If you do get through, this is much more interesting than the title makes it sound. It's not just the images, but how you can get them directly from the satellite with your own radio and antenna.
> Is there a word for the converse of "clickbait"?
The opposite of bait is repellent, so in that sense it would be "clickrepellent". However, I don't think that gets across the point that it is the inverse of "low value content with a misleadingly appealing headline", i.e. it is high value content with a headline that doesn't adequately reflect the value of the content. For this I was going to suggest "clickfish", as in the valuable objective of bait (assuming a fishing metaphor). However, a quick internet search reveals this is a widely used term already, e.g. for an internet surfer who falls for (and reposts) clickbait, or the name of anti-clickbait software. Then I thought of abstracting this a level with "clicksupper", as in fish supper, a wholesome meal made from fish and served with chips. Or how about "clicktreasure", thinking of buried treasure. But then again, how about simply "anti-clickbait"?
It's also possible to receive infrared images using the same technique, which can be used in part to visualize the water vapor content of the atmosphere:
Some parts of the imaging instrument (ABI) on GOES-17 responsible for keeping it cool failed shortly after launch, causing the temperature to vary over the course of the day:
This doesn't affect the "visible" light images too much, but can cause all sorts of problems for the infrared, which is used scientifically for various cloud measurements, sea surface temperature, land surface temperature, etc. It's still up in the air how useful we can make this data.
I don't quite understand it from skimming the article. How are you able to talk to the satellite on the protocol level? Is it continuously sending down the bits in a known format? Do you have to authenticate with it?
In this particular case, the author is using an SDR device to caputure data emitted by the satellite, and then decode it using software (goestools). There is no interaction with the satellite, it's just a device listening to a broadacst emission.
As mentioned in the article, there are many interesting related projects:
This is really cool! I keep meaning to try something like this just for the thrill of receiving the data myself.
Though if receiving the data with your own antenna and SDR isn’t exciting, and you just want to use the data, AWS hosts these (GOES-16 and 17) data on S3 for everyone to use: https://registry.opendata.aws/noaa-goes/
(Disclaimer: I work on the team that helps host these datasets)
I did wonder what full disk images of Earth might contain, considering various block formats to contain various data collected from the globe. Then I read the article.
You quite often see a bank of clocks for each time zone, it would be good to have a geostationary image like this, from several different weather satellites.
The rationale behind this was that, in this way, you can receive the stream with off-the-shelf satellite TV hardware and a PC, and without special radio operator knowledge. This was of course well before the advent of cheap SDR devices.
Speaking of which, you could also use the pixels (or some permutation of them) as a random number seed that changed every day.