The Scientist and The Alchemist
A Poem About Causality
Correlation is not Causation, said
the Scientist, of the Tribes' dance for rain;
She offered the gift of Experiment
in Observation's place.
Change One and not the Others, and then
in every case:
If Cause precedes Outcome, then
Dance precedes Rain.
The Alchemist is a Skeptic, are
two Things ever the Same?
For Same in Form and Substance,
is not Same in Time and Space.
The Laws of Physics and Reality
Orders each Atom; In plain,
the Average of a Flood
is not a Drop of Rain.
"All models are wrong-"; But
which models are useful?
If all Turtles are Maps, then
what even are they? And yet,
What has been Set in Motion,
in Motion it Remains: the Shepherd
cares to count his Sheep, the
Miller to weigh his Grain, and
the Watchmaker works to make the Watch,
which Sets his working day.
In Infinity so Partitioned,
to the Near and Far Away,
in this Universe, on this Earth,
in this Time, in this Place,
the Scientist carries out her Experiments,
the Alchemist his Arrays.
Note: This is a reworking of my own essay On Causality, which I thought had some good bits I really liked buried in too much other nonsense, so I wanted to pull the good bits out to stand on their own.
I was reminded of it in the first place because I recently watched this brilliant video on how Machine Learning works by partitioning possibility spaces, how each node in a single layer of a neural network corresponds to a folding of the possibility space, with each region partitioned by the folds corresponding to some outcome we care about. And how multiple layers of a neural network correspond to folding the same space repeatedly without unfolding, allowing for 2^n partitions per fold vs 2n partitions per fold.
Which in turn reminded me of this brilliant video on how it’s possible to test (with validity) more than one variable at time, by taking advantage of the fact that testing two variables at once is effectively the same as testing both variables singly (but at the same time), with the common idea being “clever operations you can do on possibility spaces, once you realise possibility spaces made of discrete possibility states are a thing, and not just a smear of probability.”