According to an unattributed Google AI post, “The scientific method is a systematic approach to gaining knowledge through observation, experimentation, and analysis. It involves making observations, asking questions, forming hypotheses, making predictions, testing those predictions through experiments, analyzing results, and drawing conclusions.”
Here’s a more detailed breakdown from that post:
So, what’s the problem with the scientific method as currently practiced? It could be argued that human progress has relied heavily on the successful application of this method to a myriad of human questions and problems. However, I believe there is a flaw in it that has resulted in countless hours of wasted time and needless conflict.
The problem is in Step 3. When someone proposes an explanation for an observed phenomenon, it is human nature that they will become biased toward their hypothesis. After all, that hypothesis is the product of their own intelligence and creativity. From that point on, in Steps 4 through 8, the scientist will have a predisposition, consciously or subconsciously, to skew the prediction, experimental design, data analysis, conclusion and subsequent iterations in the direction of supporting the original or perhaps slightly modified hypothesis. This becomes especially true once the results are presented as a formal peer-reviewed paper, where the scientist is defending the work against skeptics with their own biases and economic interests.
I therefore offer the following modest proposal (with apologies to Jonathan Swift), to modify the scientific method steps as follows:
1. Observation: Notice something in the world that you find interesting or puzzling.
2. Question: Formulate a question about the observation you’ve made.
3. Hypothesis: Collaborate with other independent-thinking peers to develop two or three testable explanations for your observation.
4. Prediction: Based on these hypotheses, make predictions about what you expect to observe in an experiment.
5. Experiment: Design and conduct an experiment or set of experiments to test the group’s predictions (minimum group size of two).
6. Analysis: Carefully analyze the data collected from your experiment.
7. Conclusion: Draw conclusions based on your analysis, determining how well each of the hypotheses are supported or refuted, and which correlated best with the experimental data.
8. Iteration: If none of the hypotheses are well supported by the experimental data, make revisions and repeat the process from step 4.
In the ideal case, the peer group would have the integrity and understanding to agree on how to present their data for peer review. Of course, that might require putting ego and economic self-interest aside, and therefore might not happen. But this process might improve the speed and efficiency of advancing knowledge as well as the robustness of experimental data.
In conclusion, I’m sure my proposal will be well received by scientists everywhere and will not be controversial.
James Harper
Cogitation Stew 