Jason Voorhees
๐ธ๐๐๐๐๐๐๐๐ ๐ฎ๐๐๐ โข ๐๐๐๐๐ฅ
- Joined
- May 15, 2020
- Posts
- 96,410
- Reputation
- 293,136
In operations research some of the first set of algorithms you get introduced to is Marknov Chain Monte Carlo. Also called MCMC. Which is a complicated mathemetical way of saying. You start at some point. You make tiny steps one at a time sometimes they get accepted sometimes rejected but over thousands or millions of iterations the chain is supposed to forget where it began and settle into the truth
In theory atleast on paper it explores everything that is possible but the most complex environment is always the real world and it is found that in hgh dimensional or multimodal landscapes that is the real world, it gets stuck in local modes for a very long time. This is a known phenomenon called mode trapping. The chain keeps moving, but it isnโt going anywhere new. The samples look different only because the noise is still running.
You can apply the same metropolis random walk algorithm to humans aswell. Early conditions set the starting state. Later choices are just banter steps. Most trajectories stay trapped in the same low reward regions and the rare escapes get over weighted. That is what is the illusion of productivity
Occasionally, there is a large jump. Like a new environment, a new career, a new relationship, a new city butare relatively rare. So the real question isn't Am I moving. It should be am i exploring enough because systems can keep moving forward but barely exploring.
In theory atleast on paper it explores everything that is possible but the most complex environment is always the real world and it is found that in hgh dimensional or multimodal landscapes that is the real world, it gets stuck in local modes for a very long time. This is a known phenomenon called mode trapping. The chain keeps moving, but it isnโt going anywhere new. The samples look different only because the noise is still running.
You can apply the same metropolis random walk algorithm to humans aswell. Early conditions set the starting state. Later choices are just banter steps. Most trajectories stay trapped in the same low reward regions and the rare escapes get over weighted. That is what is the illusion of productivity
Occasionally, there is a large jump. Like a new environment, a new career, a new relationship, a new city butare relatively rare. So the real question isn't Am I moving. It should be am i exploring enough because systems can keep moving forward but barely exploring.
Last edited:
