Talk:Action At A Distance Scales Exponentially

✏️ Draft: This page is a draft and is likely to be incomplete. You can help write it if you like.

Getting everyone to agree on anything significant is work. This is a view that most people find unobjectionable. Even exempting all the explanations about human nature, this is hard by default simply because transmitting information is hard and sometimes impossible. This is easily visible in hidden identity games where people are explicitly supposed to deceive the people they play with, but it's true in lots of cases. While some humans know this, since it's pretty fundamentally a part of contract theory, Humanity acts as if it does not.

Some times this is just because of physical limitations. The treaty that ended the war for American independence was pretty much ready months before the last naval engagement of the conflict simply because the simple short message saying "we're at peace" could not be communicated fast enough by the technology of the time. This sort of thing is not much of a problem anymore where wars can cease instantly if required, with radio being near universally available.

Other times it's simply because other parties are not you. You can be nominally their master, but each of them exercises independent power. Sir Walter Raleigh was pardoned and sent to find El Dorado with instructions not to attack the Spanish under penalty of death. What does his dear friend Lawrence Kemys do but immediately get embroiled in a battle with The Spanish, get Sir Raleigh's son killed, and then commit suicide. This would eventually lead to Raleigh's death by hanging.

Layering Increasing Misalignment Risk

You can push at any vertex to try to move the opposite side but you won't get that much movement in the direction you desire and lots of other things will move in directions you don't desire

It's not just a game of Telephone that causes misalignment due to tx/rx errors, but also that each layer has its own incentives and when trying to create change you are altering incentives and reward structures. But because of the self-similarity of each sub-organization with respect to its parent organization, this error is actually amplified exponentially. Unlike rigid structures with layers, each human organizational layer is also a joint which can articulate in many directions.

Unconsidered as a Cost

A common form of inherently flawed plan is the "if only we'd all agree". Stated in this form it's obviously idiotic, but people will frequently bring up the idea sufficiently obscured. A classic example from recent times has to be in pandemic response: If only the Republican states also agreed to do X we could have had fewer deaths. But often X is something that anyone could have predicted Republicans would not have gone along with like staying home for 6 months, not seeing one's family, etc.

Because Action At A Distance Scales Exponentially, making us all agree is a massive effort. That's where all the magic is.

Another common example is "If Elon Musk spent all his wealth he could solve X" but even that's untrue. He could not solve most of the world's problems this way. If he wanted to solve starvation, the coordination systems he'd devise would have to:

• be sufficiently able to be fraud tolerant that pushing in money at one end creates output at the other end
• successfully overcome opponents who benefit from the current system (including many who claim the opposite)
• fight the incentive by participants in the structure to create more structure

Most of these things end up being instead like pushing a structure of matchsticks around on a table. You get a lot of loss in the structure until the stuff reaches the end.

Costs are not even always