# every pull a lesson: moral history in a multi-turn trolley game

2026-08-18 · Veit Heller · https://cyberwitchery.com/log/cw001-every-pull-a-lesson.html

we’re starting a paper series. it’s indexed at
[cyberwitchery.com/papers](https://cyberwitchery.com/papers.html), each note is
a pdf, and this is the first one:
[`CW:001`, every pull a lesson](https://cyberwitchery.com/papers/cw001-every-pull-a-lesson.pdf).

## why?

some of our work is paper-shaped. it’s long, dense, written material. it looks
like a paper, but we’re not interested in journals or publishing, so the work
remains paper-shaped only. they have not gone through peer review, and we won’t
pretend they did.

the question that started this one: would your trolley choice change if you knew
what the person on the side track had chosen when they previously held the lever?

still: why? simple: the lab publishes on what it finds interesting. this happens
to be a trolley on a side track rather than a network fabric.

## the papers

let people rotate between the lever and the tracks, and let their earlier choices
be remembered. one extra turn is enough to turn a thought experiment into a game,
because a choice now has an actor who can be treated for it later.

the trolley is a coarse instrument, so you cannot punish or protect one person
without also deciding the fate of everyone sharing their track. collateral harm
becomes part of the game, and a private grudge can branch into a public one.

the note develops the two-turn case, derives a threshold for collateral
retaliation, compares simple moral rules in mixed populations, and specifies a
finite-agent model for the part that doesn’t collapse into closed form. the
simulation, its tests, and the program that checks the analytical identities ship
alongside it.

it is exploratory, and, once again, it is not peer reviewed. read it with that
in mind, and [argue with us](mailto:contact@cyberwitchery.com?subject=papers) if
you think it’s wrong.

