Yes let's get specific.

Suppose a satellite had a 100000 fps camera recording the intensity of Alpha Centauri A zooming in on the star such that the star filled an area averaging 50,000 pixels on a CCD with a total pixel count of 4,000,000 pixels. Assuming 24 bit color with 8 bit intensity value for each of the Red, Green and Blue channels, each bit of the display was determined to be "illuminated" or "not illuminated" or "undetermined" based upon a minimum average threshold value of 180 per channel for "illuminated", a maximum average threshold value of 20 for "not illuminated" and values between 21 and 179 for "undetermined". Suppose, computer software made such a recording for 100 days and determined that each image frame over the entire recording registered no less than 30,000 "illuminated" pixels and less than 2,000 "undetermined" pixels with an error rate of of less than 1 frame in 10 Trillion.

Now suppose that, one clear evening, the recording indicated that the light from the star was abruptly transitioning such that all pixels on the CCD were falling to an intensity lower than the maximum average threshold for "non illuminated". They also found that the recording indicated that the intensity would then abruptly transition back to its previously illuminated state. They called the period of time during the lowered "dim" and the period of time of normal illumination "bright".

Upon studying this "flickering" scientists found that the each dim period would last no less than 1/10000 +/- 1/100000 of a second as would each bright period. They further found that the transition between any dim to bright or bright to dim always occurred on 1/10000 second boundaries. As such, they could conveniently analyze each 1/10000 second consisting of 10 images to determine if such a time interval was dim or bright. They encoded the dim intervals as 0 and the bright intervals as 1.

Further analysis showed this, effective binary stream, would repeat itself every 2 hours yielding 4.5 MBytes of data. Recording this 4.5 MBytes of data over 100 repetitions they found that each bit of every repetition was consistent with every bit of every other repetition with an error of less than 1%. Hence, for every bit, 99 out of 100 trials gave the same result for any given bit, kind of like how they do error correction on noisy networks using repetition.

They saved the resulting 4.5 MBytes of binary data to a file on an Intel computer running Windows 10. They named it "signal.mpg". They double clicked on the file and their favorite MPEG viewer opened and started showing a 20 second clip. In the video, there was Alpha Centauri A right where it was supposed to be in relation to all the other stars in the video. They zoomed up to a planet orbiting the star and down to a real cool surface filled with buildings and bustling vehicles up to a little house and a yard where a 6 foot tall ET looking alien waved at the camera then reached up, grabbed it and stuck it in his pocket.

Question: Would people consider that EVIDENCE of ET?

Answer: Of course we would. 99% would. the 1% would be viewed as nut jobs.
It would be all over the news and billions of dollars would be funded to contact them along with billions of dollars to build better weapons just in case.

Yet still, its just a flickering light, thereby showing that we can infer intelligent causation from a little flicker of light.

Computers do not self assemble any more than that signal could "naturally" occur. End of proof.