In order to perform interferometry, either light has to be sent through a tunnel, or the phase of the light must be recorded.

When performing interferometry across the world, obviously the tunnel isn't an option.
And recording the phase of the light isn't possible for frequencies in the terahertz, as far as I know.

So, how about reducing the frequency?
Although frequency is a completely unalterable property of waves, it can be modulated with a new frequency by having it interfere with a different frequency.
The produced frequency can be as low as you like, and as long as you know the phase of one wave the phase of the produced wave will tell you the phase of the other wave.
So if you can produce a wave of a known phase, and that wave has a close enough frequency to the observed wave that you can record the phase of the produced wave, you know the phase of the observed wave.

If you can produce a wave of 10^8 Hz while knowing its phase, and it interferes with light of 10^14 Hz, then the resulting variation in amplitude is 10^6 Hz which would certainly be much more plausible to record the amplitude of.

Is there a scientific reason this isn't done?