Designing 150MB One Earth Message for NASA New Horizons 2 mission?

[Jarek]

NASA is planning New Horizons 2 mission leaving the Solar System, containing 150MB "One Earth Message": images, audio, 3D files ... instructions how to decode it:
http://oneearthmessage.org/
https://www.ips-planetarium.org/page/oneearth
https://www.centauri-dreams.org/2017...golden-record/

The question is how to encode it - with simple to present, universal decoding instructions.

I am in contact with Jon Lomberg who is director of this project, thought to try to discuss it here.
For color images RGB is quite subjective, it should be rather some hyperspectral image especially of the globe.
For reasonable compression ratios it should be lossy compression, but how to choose and universally present the process - transforms, quantization, entropy coding?
What hypothetical other civilization could expect there?

Any thoughts?

ps. If the message is a 2D pattern, we could make it visually containing the decoding instructions - for information steganographically hidden in the exact pattern ( https://ieeexplore.ieee.org/document/7348695 ).

Update - some materials:
1974: https://en.wikipedia.org/wiki/Arecibo_message
Use of color: https://pdfs.semanticscholar.org/9b4...3615bdae1e.pdf

Update - my bad, it didn't get to me, but this is still for New Horizons 1 - launched in 2006, passed Pluto in 2015 ...
It has a 4 GB memory chip and there is a plan to send and store there the 150MB message ...
But probably it is not the last chance, succeeding missions will be finally launched - with more such opportunities.

Lots of technical details about New Horizons: https://www.boulder.swri.edu/pkb/ssr/ssr-fountain.pdf
The download might be still possible till ~2030 if remains operational.
Uses rate 1/6 turbo coding: https://space.stackexchange.com/a/9924

[Shelwien]

I think that some simple compression methods can be used without explanation - like RLE, LZSS variants.
In fact detection and decoding of unknown LZ schemes is a regularly occuring task - just recently we had these: https://encode.su/threads/3147 https://encode.su/threads/3140
Something like QR-code with visual decoding description overlay is more "mathematically beautiful",
but is also much harder to reverse-engineer, even with visual hints.

Also I guess it would be a good idea to test whether humans can decode it before sending.

[Jarek]

While it is tempting, please let us avoid SF digressions, and try to focus on the problem itself - designing universally self-explanatory message, what is a very interesting question itself, which might bring some better understanding (...of "understanding"), maybe some new applications, and 150MB is a considerable amount to work with ... and a perspective of some participation in NASA space mission seems also quite interesting.

I got some details - there is 4GB chip, into which there will be later uploaded from distance these 150MB.
There might be possibility of printing some information/instructions e.g. on some surfaces, maybe 3D in layers of material - to be disassembled and read under microscope.
The chip contains some simple hardware error correction decoder, but adding hardware data compression decoder seems not an option.

Interesting possibility is trying to universally explain some minimalist programming language, then write decoders in this language ... and maybe other relevant programs.

The main interest is multimedia data, but maybe pure text is also worth to consider.
Notable income is from lossy compressors, but they are relatively complex ...

[Shelwien]

> some minimalist programming language, then write decoders in this language

Turing machine maybe?
Or some cellular automata? https://youtu.be/My8AsV7bA94 (Turning Machine built in Conway's Life)

For images fractal compression may be interesting in this case.

[Jarek]

Turing machine is extremely inefficient, but this might be an option.
I thought rather about using electronics as something universal - explain simplified transistor, build gates from it, simple programing language from it...

I don't think we could get a reasonable lossy image compression with cellular automata?
Fractal compression might be an option ...

[Shelwien]

> I don't think we could get a reasonable lossy image compression with cellular automata?

We probably can, for example a 3D CA based on neural network optimized to extract image features.

But my point was that CAs are simple to describe and turing-complete, so we could use them to extract something more complex.

As to electronics, what you're talking about (idealized virtual model of electronics) is imho a kind of CA (far from the simplest too),
while for real electronics it would be pretty hard to describe the components - like, are you sure you know how to build a transistor from atoms,
such that its properties would match ones of a transistor built with some specific litography process?

[Jarek]

Building real life calculations with CA seems extremely inefficient (and difficult to explain for more complex rules), but I haven't worked with that and might be wrong (ok, generally I don't get the CA excitement, e.g. building physics with it - where we know well the building material: fields, which can get localized solutions: solitons as particles, using topological we get quantization ...).

Electronics is more universal, necessary from practical perspective, and simplified transistor is just a cable which is controlled by another cable - a simple visual diagram should leave no doubts, then build higher blocks using this diagram...
Even better would be building "didactic electronics" presenting some algorithm: working, built in let say micrometer technology, such that you can take out the cover and look at electron flows e.g. with tunneling microscope (STM) - that's a truly universal presentation of an algorithm.

ps. I have just realized that it's still about New Horizon 1 (not 2) - which has passed Pluto in 2015, I don't know when the window of opportunity for sending information there ends (?) - I think it mainly depends on error correction, but it might be updatable e.g. with some interleaving (?) ... in 70s Voyagers have used extended binary Golay code, Mainer 1 Reed-Muller code, New Horizons probably uses some Turbo Codes (?)