"Unlike physics or biology, the social sciences have not demonstrated the capacity to produce a substantial body of useful, nonobvious and reliable predictivee rules about what they study"
Scientific experimentation:
talking
--> logic
--> repeated experiments (dropping heavy balls)
--> controlled experiments (scurvy)
-->random (disease and impact of genetics)
--> widely distributed (crime in different areas)
Ex: literacy program in one school may not be effective in another
"In short, no program within this universe of tests has ever demonstrated in replicated or multisite randomized experiments that it creates benefits in the excess of costs"
-Experimentation is key to success. Find what works.
"These days, experimentation is something that one assumes from a successful online commerce company"
3 Lessons:
1) Doubt effectiveness of random/replicated trials
--> trial / error better
2) Don't try to change people, try to change incentive
ex: cannot force people to stop commiting crimes. Counseling, transitional aid, boot camps do not work. But changing the criminal environment (clean up the area) does.
3) No magic; Programs that do work lead to small improvments in comparison to the problems they are trying to address
"Science may someday allow us to predict human behavior comprehensively and reliably. Until then, we need to keep stumbling forward with trial-and-error learning as best we can"
No comments:
Post a Comment