David Kipping

that can look for weird behavior.

We wrote a code called the weird detector, for instance, that was just the most generic thing possible.

Don't assume anything about the signal shape.

Just look for anything that repeats.

The signal shape can be anything.

Can we kind of learn the template of the signal from the data itself?

And then it's like a template matching filter to see if that repeats many, many times in the data.

And so we actually applied that and found a bunch of interesting stuff, but we didn't see...

anything that was the prime number sequence, at least on the Kepler data.

That's 200,000 stars, which sounds like a lot, but compared to 200 billion stars in the Milky Way, it's really just scratching the surface.

Yeah.

I don't work in compression algorithms, but I would imagine the more you compress your signal, the more assumptions that go on behalf of the decoder, the more skilled they really have to be.

A prime-number sequence is completely unencoded information, essentially.

But if you look at the Arecibo message,

they were fairly careful with their pixelation of this simple image they sent to try and make it as interpretable as possible to be that even a dumb alien would be able to figure out what we're trying to show them here.

Because there's all sorts of conventions and rules that are built in that we tend to presume when we design our messages.

And so if your message is assuming they know how to do an MP3 decoder, particularly a compression algorithm, I'm sure they could eventually reverse engineer it and figure it out, but you're making it harder for them to get to that point.

So maybe, I always think you probably would have a two-tier system, right?

You'd probably have some lower tier key system, and then maybe beneath that, you'd have a deeper compressed layer of more in-depth information.

Yeah, I mean, that's kind of the beauty of the field of techno signatures and looking for life is...