The Scientific Method Doesn’t Exist
I started this exploration wanting to answer what I thought was a simple question: "What is the scientific method?" It wasn't a simple question. This whole journey made me question ideas I had accumulated throughout my education. In this post, I document my journey.
A Neat Illusion
Textbooks make the scientific method look simple, obvious, and easy to understand. The model of a scientific method presented by most textbooks is this.
The scientific method is a systematic process for acquiring knowledge through observation, formulating a testable hypothesis, conducting experiments to test the hypothesis, analyzing the results, and drawing conclusions to refine or reject the hypothesis.
Here are excerpts from some sources around the world.
Each of these descriptions presents a model of the scientific method that is clean, step-wise, and easy to remember - first, you do this, then that, and so on. It's clean, it's logical, it's imaginable, it's teachable. This model works if you aim to understand the definition, to explain the meaning to others, to score marks in your school, or to score points with your boss.
But does it work for practicing science?
Messy Realities
For most working scientists, the purpose of scientific inquiry is to get as close as possible to the truth and to form a model of reality you can reliably act on. The path of seeking the truth is not the straight-line path that the textbooks describe.
In 1928, Alexander Fleming, a professor of bacteriology, returned from a vacation to notice that mold had grown in one of his bacterial culture plates and all the bacteria in it were dead. He isolated the mold and tested its juices on another bacterial culture, and they killed the bacterial culture. The juices were later named "penicillin".
The experiment happened by chance. He formed the hypothesis after the fact and then he tried testing it to confirm it. Was this any less scientific, or any less true?
Philosophy Meets Science
Who decides what is and is not science? Who are the people who study this? What do they have to say about the topic? In my exploration, I landed on the domain of "Philosophy of Science". Here I learned that we don’t even have a fixed definition of science, let alone the method. I would recommend avoiding the "Philosophy of Science" rabbit hole, but to give you a taste of it, here are some disagreements that philosophers of science have.
- How do we draw a line between real science and pseudo-science? (The Problem of Demarcation)
- Can we justify using past observations to predict future events? (Problem of Induction)
- Do science’s “unseen” entities exist, or are they handy fictions? Do electrons exist or are they a model that helps us predict how wires carry electricity? (Realism vs. Anti-realism)
- Can everything be fully explained by what it's made of, or do complex systems have their own rules? (Reductionism vs. Emergence)
- Do scientists uncover facts already out there, or do lab debates, funding pressures, and publication politics forge them? (Objective Facts vs. Social Construction)
One of these disagreements is "Methodological Monism vs. Pluralism" - must there be one strict “scientific method,” or is it better to allow several methods? This is an open question. We don't have a consensus on this. That essentially means we don't have a consensus on "What is the scientific method?" At this point, I don't even know whether to frame the question with "a scientific method" or "the scientific method".
Finding Common Ground
How do we know we are doing science? Well, we agree on answers to some of the questions of science. Here are the 6 major ones.
- What ultimately decides between rival claims?
- Observational/Experimental Evidence
- Example: Take your thermometer's reading instead of guessing you have a fever.
- Are scientific conclusions final?
- No. All knowledge is provisional and fallible.
- Example: Newtonian mechanics worked for centuries until relativity corrected it.
- Must scientific results be reproducible?
- Yes. Independent replication is necessary. Where direct reruns are impossible - say, a supernova - scientists demand reproducible analysis pipelines instead.
- Example: Chemists publish full protocols so any competent lab can re-run the synthesis.
- What makes a model worth keeping?
- The ability to predict and/or explain phenomena beyond what inspired the original model.
- Example: Mendel's genetics predicted 3:1 ratios in pea traits he had not yet counted.
- Should scientific inquiry be subjective or objective?
- Objective - free from personal or cultural bias.
- Example: Double-blind drug trials strip away doctors' and patients' expectations.
- Can causes be supernatural?
- No. Science brackets the supernatural by methodological choice, not because it can disprove it.
- Example: Meteorologists model floods with fluid dynamics, not divine wrath.
How do we do science? By using tools such as these, to form theories and to establish the probability of their truth, but there is NO fixed path to getting there. Some rely on tools such as gathering data, making hypotheses, running experiments, proposing theories, objectivity, reproducibility, etc. A scientist may use some or all. But, no single process or method connects them. You don't have to gather data or observe a phenomenon to form a hypothesis; you don't need to propose theories after experimentation. You can pick tools in a different order, and you can use tools more than once.
The scientific process is fluid. If your theories are open to change, other scientists can replicate your experiments, your data is not intentionally biased, and so on, you are doing science. Some theories demand massive datasets - like vaccine trials with thousands of participants - while others hinge on a single, precise observation - like the 1919 eclipse that showed starlight bending around the sun, confirming general relativity. Some need statistical methods, some need laboratory methods. Some experiments require precise conditions, while you can run others in the wild.
Some organizations publish field-specific guides:
- NIH on human trials
- OECD on social-science protocols
- CONSORT for randomised-trial reporting
- APA ethics, ASTM materials testing, FAIR data principles
Why Textbooks Lie
Then why are textbooks the way they are? Why do they teach a linear, unambiguous scientific method?
Ideas in textbooks work on the principle of scaffolding - simpler models are often easier to understand and provide a foundation for learning more complex concepts. We intentionally reduce the complexity to make the real models understandable. It's the same reason we still teach the Newtonian theory of gravity, instead of Einstein's - it provides a simpler model that works well for most real-world situations until we hit a wall, question the model, and graduate to a better model.
Teaching the scientific method in this way has a mnemonic value - it helps people remember scientific tools by presenting a single linear narrative. Linear narratives and causalities are easier for humans to follow. The simple scientific method is a scaffolding for this real scientific method. It's not the actual thing.
This approach to scaffolding complex ideas works because our education/workforce system works with the assumption that ONLY experts will use these tools in situations of consequence. You would need to go away from Newton's model of gravity to Einstein's model if you were a rocket scientist. And if you are a rocket scientist, you would know NOT to apply the simpler model because you learned about the nuance, first, in your higher education and, then, in the initial years of your job.
Similarly, for the scientific method, the system assumes that people will learn the nuance when they do actual science and until then a complex model is not needed.
Some textbooks come with a warning about this reduction of complexity.
When Simplicity Fails
Only experts will use these tools in situations of consequence
But does this assumption hold in the real world?
Take the example of people who vote for rent control for apartments. They use the standard model of supply and demand. In the rental market, the price is set to the point where how many people want apartments (demand) meets how many landlords offer them (supply). According to them, if you fix the maximum rent of an apartment it ensures that landlords can't exploit the tenants and more people can afford housing. And, since rent control often exempts luxury properties, those stay out of the purview of rent control. Logical? Yes. Does it work? No.
This is what happens. Landlords convert the apartments into condos so that they are out of the definition of rent control. In San Francisco, expanding control in 1994 prompted landlords to convert buildings to condos or neglect maintenance, shrinking the rental stock by up to 15 percentage points and attracting higher-income newcomers, thereby accelerating gentrification.
Non-experts can and do apply these basic models. They do it in policy, in government, in technology, in business. All ideas are not equal. Scaffolding works for physics because physics research is largely confined to trained physicists, whereas supply-and-demand curves or "the scientific method" are used by a much wider public. The "scientific method" shapes decisions in medicine, policy, and technology. That scaffolding breaks when non-experts wield simplified models in policy.
People who encounter a simple five-step cartoon either treat it as childishly simplistic or, paradoxically, reject it as over-engineered jargon; in both cases, they ignore the real, flexible practice of science. In both cases, it's a net loss for everyone.
Intellectual Humility
What do we do? I have a simple answer: intellectual humility. Intellectual humility is knowing that your knowledge has limits. It's the willingness to say, "I don't know" or "I may be wrong". Treat the textbook diagram as a mnemonic, not a map, and stay curious enough to know when you’ve left the paved road.