Spacemapping
What we deliberately did not build
The non-goals list is longer than the feature list, and keeping it that way has been the hardest ongoing discipline in the project.
Enthropy Systems
Spacemapping
Qualified, listening, linked
Not everything in the archive can support every claim. Rather than fix that or hide it, we made the grade a property of the object.
Spacemapping
The producing process forgets the map
We have hours of audio rendered from real observations, and for much of it the link between second and pixel was thrown away at render time. It cannot be reconstructed by wanting it.
Spacemapping
It refuses to guess
A guess that is usually right is worse than an error, because the run still finishes and the guess is invisible in the output.
Spacemapping
The boring layer that makes the interesting layer checkable
Before any claim about structure in the data, something has to be able to say: this second of audio corresponds to these pixels, and here is the proof.
Far Wide Radio
What it actually takes to put a two-hour episode on the internet
The hosting was the easy part. The failures were all in defaults written by people who assumed your files were small.
Far Wide Radio
Curation by method is still curation
We dismissed a large part of the archive as "mechanical" selections. That was a misreading, and correcting it changed how the programs got built.
Far Wide Radio
A container may not assert what the content never had
We shipped every asset in the collection at four times its necessary size, with a sidecar file certifying a fidelity that did not exist. Twice, in one day, wearing different clothes.
Far Wide Radio
One cloud, five passes
The same NIRSpec observation, worked five separate times, played back to back. The cloud does not change between passes. The listening does.
Far Wide Radio
The reservoir, not the treadmill
Far Wide Radio is built to be re-aired, not to be fed. That single programming rule decided nearly everything else about it.