If We Knew What It Was We Were Doing

That line has been living rent-free in my head for years Most people skip this — try not to..

You've seen it on mugs. Posters in university labs. LinkedIn posts from people who've never actually run an experiment. On top of that, "If we knew what it was we were doing, it would not be called research, would it? This leads to " — attributed to Einstein, though nobody can find the exact source. Doesn't matter. The sentiment hits harder than the citation No workaround needed..

Here's the thing: most people treat that quote as comfort food. A nice reminder that confusion is normal. But if you actually do research — or build products, or write code, or try to solve problems nobody has solved before — that quote isn't comforting. It's a job description That's the whole idea..

What Is Research, Actually

Strip away the lab coats and the grant proposals. Research is just structured uncertainty.

You have a question. If someone did, you'd just ask them and move on. You don't know the answer. Then you do it again. Nobody does. So you design a way to poke at the unknown, gather signal, and update your mental model. And again.

The difference between research and guessing

Guessing is "I think this might work." Research is "I think this might work because X, so I'll test Y and measure Z, and if Z happens then X gains weight, but if not then I need to reconsider X."

That structure — hypothesis, method, measurement, update — is the only thing separating science from superstition. It's also the only thing separating a startup from a hobby, a novel from a diary, a renovation from a disaster.

Basic vs. applied: a false dichotomy

People love categorizing research. Basic (curiosity-driven). On the flip side, applied (problem-driven). On top of that, translational (bridging the two). Which means in practice? The line blurs until it disappears.

CRISPR started as basic research on bacterial immune systems. In practice, the web started as a way for physicists to share documents. mRNA vaccines spent decades in "useless" basic science before saving millions of lives. In practice, you can't predict which rabbit holes lead to gold. That's not a bug — it's the whole point That's the part that actually makes a difference. That alone is useful..

Why This Matters More Than Ever

We're drowning in certainty theater Worth keeping that in mind..

Social media rewards confident takes. Investors want "de-risked" bets. Also, the result? Managers want predictable timelines. Plus, schools teach kids that every question has an answer in the back of the textbook. A culture that treats uncertainty as failure instead of the prerequisite for discovery.

The cost of pretending we know

When organizations pretend research is just execution with extra steps, three things happen:

1. People stop asking hard questions. If "I don't know" looks like incompetence, you'll get performative confidence instead of honest exploration.
2. Resources flow to the wrong places. Safe, incremental work gets funded. The weird, risky, potentially transformative stuff starves.
3. Failure becomes shameful instead of informative. A null result in a well-designed experiment is data. A null result in a performative "research" project is a career risk.

Look at the replication crisis in psychology. Or the Theranos saga. Or any number of "AI breakthroughs" that evaporate under scrutiny. That said, these aren't failures of intelligence. They're failures of culture — cultures that punished uncertainty until people stopped admitting it existed.

The personal stakes

This isn't just institutional. It's personal And that's really what it comes down to..

Every time you start a project that matters — writing a book, learning a language, building a product, fixing a relationship — you're doing research. You don't know what you're doing. If you did, you'd be done already The details matter here. No workaround needed..

The people who finish hard things aren't the ones who feel confident. They're the ones who built a process for moving forward without confidence. They treat "I don't know" as a starting signal, not a stop sign Surprisingly effective..

How It Works: The Loop Nobody Talks About

Textbooks show the scientific method as a clean cycle. Observe → Hypothesize → Experiment → Analyze → Conclude. Real research looks more like a plate of spaghetti thrown at a wall Worth keeping that in mind..

Phase 1: Finding the right question (harder than answering it)

Most failed projects die here. Not because the execution was bad, but because the question was wrong.

A good research question has three properties:

• Specific enough to test. "Why do people behave badly?" is philosophy. Practically speaking, "Does displaying social proof increase checkout completion on mobile? Because of that, " is research. That's why - **Connected to something that matters. Plus, ** If the answer changes nothing, why spend the time? - Survivable if the answer is "no." If a negative result destroys your thesis, your funding, or your ego, you've set yourself up to cheat.

It sounds simple, but the gap is usually here Worth keeping that in mind..

I've seen PhD candidates spend two years on questions their advisor knew were unanswerable. Still, i've seen startups build features nobody asked for because the founders fell in love with a solution before validating the problem. The pattern is always the same: they skipped the uncomfortable work of sharpening the question And that's really what it comes down to..

Phase 2: Designing tests that could actually fail

This is where most people cheat — often unconsciously.

They design experiments that can't falsify their hypothesis. They test on convenient populations. Now, they measure vanity metrics. They stop collecting data when the numbers look good and keep going when they don't Less friction, more output..

A real test has teeth. So naturally, it specifies in advance: what would change my mind? What result would make me say "huh, I was wrong"? But if that result is impossible by design, you're not researching. You're performing That's the whole idea..

Phase 3: Running the thing (where the magic isn't)

This phase is mostly boredom, logistics, and things breaking.

Equipment fails. Code has bugs. Your collaborator gets sick. In real terms, the reagent lot is bad. Participants ghost. Also, the weather ruins your field study. And the API changes. The IRB takes six months.

It's where the quote lives. If we knew what we were doing, it wouldn't be called research. You're not executing a plan. You're navigating a minefield in the dark, updating your map with every explosion.

The skill here isn't brilliance. Can you distinguish "this approach is wrong" from "this implementation is broken"? Can you stay curious when everything is annoying? Worth adding: it's debugging stamina. Can you keep a lab notebook (literal or metaphorical) that lets you reconstruct your reasoning six months later when the reviewer asks?

Phase 4: Updating — the part everyone skips

You got a result. Now what?

Most people: "Great, it worked / didn't work. On to the next thing."

Real researchers: "Okay, why? What would I do differently if I started over? What does this actually tell me? What are the alternative explanations? What new questions does this raise?

The update phase is where learning compounds. Worth adding: skip it, and every project starts from zero. Do it rigorously, and you build a personal knowledge base that makes future projects faster — not because you know more answers, but because you ask better questions.

Short version: it depends. Long version — keep reading And that's really what it comes down to..

Common Mistakes: What Most People Get Wrong

Mistake 1: Confusing activity with progress

Running experiments feels productive. Writing code feels productive. Reading papers feels productive Simple, but easy to overlook..

But if those activities aren't tied to a decision — "this result will make me do X instead of Y" — they're just motion. Research is decision-driven. Every step should reduce uncertainty about a specific choice. If it doesn't, stop Small thing, real impact. That's the whole idea..

Mistake 2: Falling in love with the method

Mistake 2: Falling in love with the method

When a particular technique yields publishable numbers, it’s easy to convince yourself that the tool itself is the source of insight. The allure of a sleek model, a novel assay, or a cutting‑edge statistical pipeline can eclipse the question of whether it actually addresses the problem at hand.

Why this is dangerous:

• Tunnel vision. You begin to interpret every observation through the lens of the chosen method, forcing data to fit the framework rather than allowing the data to dictate the framework.
• Blind spots. Limitations inherent to the method — untested assumptions, restricted generalizability, or hidden biases — are glossed over because the excitement of “using something new” outweighs the rigor of critical appraisal.

How to stay grounded:

1. Pre‑register the analytical plan. Declare which models, transformations, or tests you will employ before looking at the results. If the data violate the prerequisites, you’ll be compelled to adjust the approach rather than rationalize the violation.
2. Run a “method‑agnostic” baseline. Implement a simple, well‑understood procedure (e.g., a linear regression, a control‑group comparison) alongside the sophisticated technique. If the complex method does not outperform the baseline in a meaningful way, the extra effort is unjustified.
3. Invite a skeptical reviewer. Ask a colleague who is unfamiliar with the method to critique the experimental design. Their fresh perspective often uncovers hidden dependencies on the technique itself.

Mistake 3: Ignoring the cost of uncertainty

Researchers frequently treat uncertainty as a nuisance to be minimized at any cost, yet every decision carries a trade‑off between precision and practicality Most people skip this — try not to..

• Over‑optimizing for confidence intervals can lead to painfully slow data collection, missed deadlines, and resource exhaustion.
• Paralysis by analysis occurs when you demand perfect statistical power before even forming a hypothesis, turning curiosity into a bureaucratic exercise.

Balancing act:

• Define a minimum viable level of certainty required to make a decision. Anything beyond that is optional and should be weighed against the marginal information gain.
• Use sequential analysis or adaptive designs that allow you to stop early if the effect size is clear, thereby preserving resources while still respecting uncertainty.

Mistake 4: Underestimating reproducibility

A result that cannot be reproduced is effectively invisible to the scientific community Which is the point..

• Hidden variables — such as subtle differences in protocol, software versions, or environmental conditions — can invalidate findings without obvious warning signs.
• One‑off successes give a false sense of validation; without independent replication, the community cannot assess the robustness of the claim.

Practical steps:

1. Document every deviation from the canonical procedure, no matter how trivial. A lab notebook entry that notes “used a fresh batch of reagent X on day 12” can become the key to tracing a reproducibility failure later.
2. Share code and data openly, preferably with version control and containerized environments that capture the exact computational context.
3. Encourage replication attempts as part of the publication process. Journals that require a replication checklist raise the overall standard of rigor.

Mistake 5: Neglecting the ethical dimension

Ethics is not a peripheral checklist; it shapes the very questions you are allowed to ask and the methods you may employ.

• Participant welfare must be weighed against scientific gain. Cutting corners on informed consent or compensation can compromise data integrity and erode public trust.
• Data stewardship involves safeguarding privacy, preventing misuse, and being transparent about limitations.

Embedding ethics into the workflow:

• Conduct an ethical impact assessment at the proposal stage, listing potential harms and mitigation strategies.
• Schedule periodic “ethical audits” during the project to verify that ongoing practices still align with the original commitments.

Conclusion

The research journey is a sequence of disciplined phases, each demanding its own blend of curiosity, rigor, and pragmatism. By constructing tests that can genuinely falsify hypotheses, embracing the inevitable chaos of execution, committing to systematic updates, and avoiding the pitfalls of activity‑driven progress, methodological infatuation, unchecked uncertainty, reproducibility neglect, and ethical oversight, scholars transform raw effort into cumulative knowledge Worth knowing..

When every step is anchored to a clear decision‑making purpose and every failure is treated as a source of information rather than a setback, the “magic” of research emerges not from sudden epiphanies but from a steadfast, iterative commitment to truth.