Analogue broacasts of TV and radio are fairly easy to decode intuitively - For radio, it’s an oscillating signal that represents the sound wave, in the case of TV, it’s raster data with repeating components - seems like it’s not a massive leap of intuition to interpret that as images.
But what about digital broadcasts? Those are just going to look like noise, unless you already know the codec, aren’t they? How complex a task would it be to reverse-engineer, say, MPEG2, based only on a collection of examples of the encoded signal, and without knowledge of the end product?
I would challenge your premise. Analog TV signals are not exactly simple. A lot of prior knowledge is required to decode NTSC or PAL into a useable image.
Even audio radio signals aren’t plainly obvious. Assuming you figure out AM and FM, you still need to use a lot of trial-and-error to find the frequency boundaries.
But there are discernable patterns in analogue TV signals, both spatial and temporal, that should allow intelligent aliens to eventually reconstruct the picture. I know, I saw them do it in Contact ;).
The thing about compression schemes is that they are trying to throw away redundant, repeated information, to maximise the entropy in other words. A perfect compression scheme, in this sense, would yield data that looks random. MPEG2 doesn’t try to achieve that kind of perfection, of course.
It is more complicated that you think Mangetout. You don’t just send the bits for the MPEG2 stream in order they are used. Before you get to the point of reverse engineering MPEG2. You have to reverse engineer the error coding and interleaving. You can look an analog TV signal and see repeating patterns. I am not familiar with digital TV but digital cellular is very complicated and the repeating things tend to be long sequences of tones.
But Earth isn’t starting out with digital signals at point 0. We’ve broadcast analog signals for decades, and since digital signals are limited also by the speed of light, they can’t overtake the older signals. So the aliens and their equivalent to SETI would first notice the analog signals, decode them, learn our languages, watch the TV and listen to radio to learn our history, …
and once they understand English and get all TV (since all signals are broadcast spherically), they can learn most of the technical stuff from there. There are geek and technical shows on TV, after all, and news shows cover the switchover to digital.
That part, at least, shouldn’t be too hard. They probably use some sort of error-correcting encoding in their own communications, and the math behind that is the same everywhere. Once you get the idea to look for them, those patterns in the bits should be fairly easy to spot. Of course, that won’t tell you anything about what the actual data stream is, though.
I think if intelligent aliens sampled a black-and-white NTSC or PAL analog signal for a good long time (say a few days), and if they had a sense of vision as we do, they’d probably figure out the encoding without much trouble. They’d also know the correct vertical orientation of the picture by watching for falling objects, and the correct aspect ratio by looking for things that are presumably circles (like planets, or bubbles).
Figuring out the correct horizontal orientation would be trickier. The signal would have to include at some point an image of a well known object that looks the same to us as it does to them. Perhaps an image of the Andromeda Galaxy, for example.
For a color analog signal, I expect they’d overlook or dismiss the IQ/UV color information, chalking it up to noise — especially if they didn’t have color vision themselves, or if they had designed their own color TVs “sensibly”, and not in the backward-compatible way we did. They could still extract a coherent B&W picture though.
Obviously this is all speculative. (I don’t know any aliens I can ask.)
There are many error coding ideas that are currently in use. Turbo codes, convolutional codes elliptical codes block codes.
Lets take turbo codes. Turbo codes have three sorts of parameters to them. The convolutional code you use for the taps. The interleaving method and the code rate you send. The interleaving is basically changing the order of how you send the bits through the convolutional coder. For a well performing code you want the bit order to be random and as such there are billions and billions of different ways of doing it with each being just as good as the next. Without knowing the interleaving pattern you basically have random bits.
I think you are vastly underestimating the issue with reverse engineering the error coding. The thing about error coding is there are so many ways to do more or less the same thing it is a stretch to say they do the similar math on planet X.
I’m under no illusions regarding the difficulty of decoding a contextless signal - it’s hard enough even when the signal is purpose-designed to be self-unpacking.
But an analogue tv broadcast that contains some rectilinear graphics that stay on screen for a few frames - or better still, a test card- should contain enough clues to offer a fighting chance of being decoded by an intelligent agent seeing it completely afresh.
The point is that (as I understand it) the task of decoding compressed video devoid of context is orders of magnitude more difficult than analogue.
Surely they could translate some nouns, at least? Example: they repeatedly hear sounds similar to “Obama” simultaneously with images of some kind of organism, which close study reveals to be the same organism. Conclusion: “Obama” identifies this individual.
Although I’m probably glossing over the difficulties of isolating words from continuous speech, especially for beings who have never encountered human speech.
Of course, the shows themselves would be cultural context. I mean, ultimately, each and every one of us has managed to decipher a language without any prior context.
The better a data compression scheme is, the more its output resembles random noise.
The electromagnetic spectrum is a finite resource. So any civilization using it as a communications medium will, over time, employ better and better data compression algorithms to make the best use of the limited bandwidth.
This means that the probability of any technological civilization decoding a signal from another technological civilization is very low. There will be a brief period after radio is discovered where a civilization’s transmissions will have some discernible structure, but within a few centuries the imperative of data compression will make every signal sound like static.
A signal from an alien race will either be a deliberate attempt at contact, or an incomprehensible flood of white noise from an unexpected source.
Well, the more efficient the compression is, maybe, but “efficient” is not necessarily synonymous with “better”. With most practical compression schemes in actual use, after they take out all of the redundancy they can find, they go and put some redundancy back in, to allow for error correction or at least detection (depending on the application). A signal which was genuinely indistinguishable from noise wouldn’t be of much practical value.