mirrorandloom

I’ve often thought of this as:
- limit of 1.01^n as n goes to ♾️ goes to ♾️ 
- limit of 0.99^n as n goes to ♾️ goes to 0

first order and second order effects causing more good than bad (I’ll settle for a Kantian Duty & Universality morals in this context to keep it simple) because in the long run, if it’s 51% it will compound overtime to infinity

Small increments in improvement generate massive results. If you can improve a model by even a small percentage, over time it will pay massive dividends.