3 ms·

The problem is our obsession with positive results. Let's say there are 57 possible ways of doing something, only two of them work, but there is objectively not

The problem is our obsession with positive results. Let's say there are 57 possible ways of doing something, only two of them work, but there is objectively not enough information to decide which one and testing each of them requires decades of work. Out of 120 people who tried, three stumble upon the correct solution, those are seen as geniuses and heroes, while the others are seen as failures who devoted their lives to misguided ideas, even though the work they did was identical. Similar with experiments. Good experiments are those that produce expected results, but those teach you nothing. A well designed experiment is in fact the one whose results are as unpredictable as possible, as it has the highest chance of providing new information. While an experiment designed to achieve a predicted result can only teach you something new by sheer accident, when the prediction turns out to be wrong.

AnthonyMouse 8y ago
We celebrate the winners because it's hard to know how much of winning was chance and how much was skill. There were 57 possible ways of doing something, but the successful person picked the best one. Was their choice fully random, or did they see some specific potential in that option and purposely chose that one to try? What about the choice to try one of the 57 to begin with rather than any of the other infinitely many things they could have done with their time when there was no guarantee any of them would work? There being more ways than one individual can attempt also has a solution. Anyone who wants to can go in with the other scientists, have each one try one of the options but agree to share the spoils if any of them succeeds. Then you succeed if anyone does and the proceeds from winning are reduced but the probability increases. One reason not to do this is if you think one specific option is more likely than the others and you're willing to risk individual failure for individual success. But if you're correct to do that then you're correctly rewarded for doing it. And there are reasons the individual method can produce better results in addition to that -- if you're in a pool with many others and succeed when any of them do, you have a reduced incentive to be meticulous because your personal reward is no longer strongly correlated with your personal success or failure. Anyone can join the pool, slack off and hope to profit if someone else succeeds. Then many people do that and many advances are lost. See also employees of any large bureaucratic organization/government. So we want individuals to take good risks, but the only way to tell if they're good instead of just risks is when they succeed. And the reward has to be in proportion to the risk to get people to do it, which means attempts with a low probability of success require a large reward.

AnthonyMouse 8y ago

We celebrate the winners because it's hard to know how much of winning was chance and how much was skill. There were 57 possible ways of doing something, but the successful person picked the best one. Was their choice fully random, or did they see some specific potential in that option and purposely chose that one to try? What about the choice to try one of the 57 to begin with rather than any of the other infinitely many things they could have done with their time when there was no guarantee any of them would work? There being more ways than one individual can attempt also has a solution. Anyone who wants to can go in with the other scientists, have each one try one of the options but agree to share the spoils if any of them succeeds. Then you succeed if anyone does and the proceeds from winning are reduced but the probability increases. One reason not to do this is if you think one specific option is more likely than the others and you're willing to risk individual failure for individual success. But if you're correct to do that then you're correctly rewarded for doing it. And there are reasons the individual method can produce better results in addition to that -- if you're in a pool with many others and succeed when any of them do, you have a reduced incentive to be meticulous because your personal reward is no longer strongly correlated with your personal success or failure. Anyone can join the pool, slack off and hope to profit if someone else succeeds. Then many people do that and many advances are lost. See also employees of any large bureaucratic organization/government. So we want individuals to take good risks, but the only way to tell if they're good instead of just risks is when they succeed. And the reward has to be in proportion to the risk to get people to do it, which means attempts with a low probability of success require a large reward.