Hi, this is Ray.

I want to start with a confession about my own learning tendencies that took me embarrassingly long to correct. For most of my adult life, when faced with something new I wanted to learn, my default approach was to read about it exhaustively before actually trying it. If I wanted to learn a new tool, I would read the documentation cover to cover before touching it. If I wanted to learn a new skill, I would consume every article, video, and book I could find before attempting anything practical. This felt disciplined. It felt thorough. I told myself I was preparing to do things right rather than jumping in blindly.

The problem with this approach, which I couldn't see for years, is that it produced a specific pattern of shallow expertise. I knew a lot about many things without being genuinely capable at most of them. I could discuss techniques I'd never used, describe processes I'd never executed, and explain concepts I'd never actually applied. When I finally did attempt the practical work, I would discover that most of what I'd read hadn't stuck in a usable way, and I still had to learn the actual doing through actual practice… now with the added handicap of accumulated preconceptions about how things "should" work that often didn't survive contact with reality.

The turning point came when I started doing the opposite. I would learn the absolute minimum needed to start experimenting, then start experimenting. When I got stuck, I'd learn what I needed to unstick, then keep experimenting. This felt sloppy. It felt wrong. It also produced dramatically better learning than my previous approach. Within months of switching modes, I'd built more actual capability in whatever I was learning than I'd built in years of the reading-first approach. Not because the reading was useless… it wasn't. But because reading is a different mode of learning than experimentation, and each produces different kinds of understanding, and both are needed but I'd been over-relying on one and under-using the other.

Today's newsletter is about that. What the research actually shows about experimentation as a learning mode, why it produces things that pure study can't, and how to actually structure your learning to include the specific benefits of trial-and-error engagement. This is one of those topics where the popular framing (experimenters are just impulsive, careful learners plan first) gets the science genuinely backwards. Let's get into it.

Ready to transform artificial intelligence from a buzzword into your personal revenue generator?

HubSpot’s groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.

Inside you'll discover:

Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.

Get The Free Guide

Let me start with what the science actually shows, because the foundational research is more relevant than most learners realize.

The classic research on trial-and-error learning comes from Edward Thorndike in the early 20th century. According to summaries of his work, Thorndike's experiments with animals demonstrated that learning occurs through a process of attempting various behaviors until the successful ones are reinforced. This foundational research highlights the essence of learning through persistence and adaptability. This might sound obvious now, but the finding was substantial. Learning happens through action-outcome loops, not just through observation. Doing something, seeing what happens, adjusting, doing again… this cycle is a fundamental mechanism by which biological organisms actually acquire capabilities.

The specific finding I want to highlight comes from research on early exploration and later performance. According to a study of mice navigating spatial tasks, Solutions varied among mice but were predictable based on individual early trial-and-error patterns observed in Test 1: mice that had initially explored more extensively found better solutions. Finally, when the barriers were removed, all mice reverted to the best solution after active exploration. Thus, early active exploration helps mice to develop optimal strategies. Read this carefully. The mice that did more early experimentation found better solutions to later problems. Not because experimentation is inherently virtuous. Because the experimentation was building the specific cognitive maps that enabled optimal problem-solving later. Same principle applies to human learners. Early experimentation builds capacity that later study can't replicate.

The research on exploratory learning specifically has a useful framing. According to research in this area, exploratory learning emphasizes learners' active exploration, discovery, and reflection to acquire new knowledge and solve problems. It involves iterative learning processes, deep cognitive engagement, and the integration of real-world contexts and social interactions, aiming to cultivate learners' problem-solving skills and critical learning abilities. The specific characteristics matter. Not passive absorption. Active exploration. Discovery. Reflection. Iteration. Real-world context. These features distinguish experimentation from other modes of learning, and they produce specific benefits that non-experimental learning doesn't produce.

The reinforcement learning research from AI provides another useful frame. According to research on exploration versus exploitation trade-offs, effective learning systems balance two modes: exploration (trying new things to gather information) and exploitation (using what's already known to produce results). Systems that only exploit stagnate because they never learn new possibilities. Systems that only explore never consolidate their learning into useful capabilities. Human learners face the same trade-off. The learner who only studies existing knowledge (pure exploitation of others' understanding) never explores enough to develop their own capabilities. The learner who only experiments without consolidating learns slowly. Both modes matter, and knowing when to shift between them is a genuine skill.

Let me name the specific mechanisms by which experimentation produces learning that pure study can't.

Direct causal knowledge. When you experiment, you directly observe cause and effect. Not what someone else told you would happen. What actually happened when you did this specific thing in this specific context. This first-person causal knowledge is different from received knowledge. It's tied to your own experience in ways that make it more usable, more retained, and more transferable.

Encountering the actual problems. Reading about how to do something usually presents an idealized version of the process. Experimenting produces the actual problems… the specific ways things go wrong, the specific issues you encounter, the specific difficulties you have to solve. These actual problems teach you things the idealized description couldn't have.

Building tacit knowledge. Some knowledge can't be transmitted through language. Michael Polanyi called this tacit knowledge… the kind of knowing that lives in your body, your intuitions, your reflexes. You can only build tacit knowledge through direct practice. Reading about how to ride a bike doesn't produce the ability to ride a bike. Only riding does. Same principle applies to any skill… there's always a tacit component that only experimentation can develop.

Debugging your own understanding. When you experiment and something doesn't work, you discover exactly where your understanding is wrong. This is diagnostic information you can't get from pure study because pure study doesn't produce specific failures at specific points. In Metroid Prime terms, the scan visor gives you information about the world, but you don't really understand a specific enemy until you've actually fought them and seen how your ideas about them work in practice.

Building intuition through feedback. Each experiment provides feedback that adjusts your intuitions. Over many experiments, your intuitions become more accurate. This intuitive knowledge is different from explicit knowledge… it's the felt sense of "this is likely to work" or "this probably won't" that guides expert performance. It develops only through repeated experimental exposure to a domain.

Discovering things nobody told you. Sometimes experimentation reveals things that aren't in the received knowledge because nobody has articulated them yet. Or things that are missing from the standard sources for reasons of oversimplification. Or connections between things that people who wrote separately about each hadn't noticed. Experimentation is a mechanism for original discovery, even in domains where "everything is already known."

Let me be specific about the pattern I fell into, because I think a lot of learners fall into it.

The pattern goes like this. You want to learn something. You start by reading about it. You read one book. Then another. Then another. You watch tutorials. You take courses. You accumulate information. You feel like you're preparing to become capable. But you never quite start actually doing the thing, because there's always more to read first, and you'd hate to start incorrectly, and you want to have a solid foundation before you jump in.

Months or years pass. You know a lot about your subject. You can discuss it fluently. You've absorbed the vocabulary. You feel like an educated observer of the field. But when you finally try to actually do something, you discover that most of what you learned hasn't translated into capability. You still have to build the actual skills through actual practice, and now you've spent a lot of time on preparation that's produced surprisingly little foundation for the actual work.

This pattern is common enough that it has various names. Analysis paralysis. Reading instead of doing. Perpetual student mode. Whatever you call it, the pattern shares a specific structure. Study is being used as a substitute for experimentation rather than as a supplement to it. The learner is optimizing for feeling prepared rather than for actually being capable.

The intervention isn't to stop studying. It's to stop using study as an excuse to avoid experimentation. Study a bit. Experiment. Study more when you hit specific questions the experimentation raised. Experiment more. Alternate the modes. The alternation is what produces actual capability. Pure study produces knowledge about capability without producing capability itself.

Okay, the practical part. Based on the research and my own experience, here's how to actually incorporate experimentation into your learning.

Start experimenting before you feel ready. This is the single most important move. Don't wait until you've read enough. Start doing things before your studying is complete. The experiments themselves will teach you what you need to study next, which will be more useful than trying to guess what you need to know before starting.

Ask "what can I try today?" every day. Even during heavy study phases, ask what small experiment you could run today that would produce feedback. Not massive projects. Small trials. What if I try this specific thing and see what happens? This question keeps the experimental mode active alongside whatever else you're doing.

Prefer minimum viable experiments. Instead of planning elaborate multi-step projects that require extensive preparation, look for the smallest possible experiment you could actually run this week. What's the minimum version of the thing that would give you real feedback? Start there. Learn from it. Iterate.

Track what actually happens, not what you expected. When you experiment, deliberately notice the gap between what you expected and what actually occurred. This gap is diagnostic. It reveals where your understanding was wrong. Ignoring it and just doing more experiments won't help. Noticing it and updating your model is where the learning happens.

Alternate exploration and consolidation. Don't just experiment continuously. After periods of experimentation, take time to consolidate what you learned. Write it down. Think about the patterns. Integrate the experimental findings with your existing understanding. Then return to experimentation with better foundations. This alternation is more productive than either pure experimentation or pure consolidation.

Use experiments to identify what to study. When you hit a specific problem in your experimentation (something you can't figure out through further trial and error), that's exactly when studying becomes maximally useful. You now know specifically what you need to learn, which makes the studying targeted and efficient. This is dramatically better than trying to guess what you'll need to know in advance.

Design experiments to test specific hypotheses. The most productive experiments have specific questions attached. Not "let me just try things." "If I do X, then Y should happen because of Z. Let me test that." When Y doesn't happen, you've learned that your model was wrong in a specific way. When Y does happen, you've confirmed something about your understanding. Both outcomes are useful.

Accept that most experiments will fail. According to research on this, the acceptance of failure as a learning tool encourages a growth mindset. In educational settings, students are more likely to engage with challenging material when they perceive failure as a pathway to knowledge rather than a definitive end. Failed experiments aren't setbacks… they're what you were trying to produce. The information from a failed experiment is often more valuable than the information from a successful one. This isn't a consolation. It's how experimental learning actually works.

Keep an experiment log. Not detailed. Just a record of what you tried, what you expected, what happened, and what you learned. This log becomes invaluable over time. It reveals patterns in your learning that any single experiment can't. And it gives you specific historical evidence that experimentation is producing capability, which supports continued experimental engagement.

Some honest caveats.

Not everything can be safely experimented with. Some domains have consequences too severe to learn through pure trial-and-error. Medical procedures. Legal advice. Financial decisions with your life savings. High-stakes physical activities. In these domains, more study before action is genuinely appropriate. Not everything is a low-stakes trial.

Not all learning benefits from experimentation. Some material is genuinely abstract enough that experimentation doesn't apply well. Certain kinds of theoretical study, historical understanding, or purely conceptual learning may not have obvious experimental applications. Match the mode to the material.

Experimentation without reflection is just repetition. As I've covered in the previous newsletter about repetition, experimenting without meaningful engagement can produce very little learning. The value isn't in the actions themselves… it's in the observation, reflection, and integration that follows each experiment. Without those, you're just doing things without learning from them.

Some experiments require substantial preparation. For genuinely complex projects, some preparation is legitimate before starting. The point isn't to skip all preparation… it's to not use preparation as an infinite delay tactic. In Chrono Trigger, the party doesn't fight Lavos without any preparation. But they also don't wait until they have every possible item before starting the journey. Reasonable preparation, then action, then more preparation as needed.

Individual variation is real. Different learners have different optimal balances between study and experimentation. Some benefit from more preparation; some benefit from earlier action. Pay attention to what actually works for you rather than assuming universal ratios apply.

Here's what I want you to take from all this. The cultural default for many educated adults is to over-study and under-experiment. This produces the specific failure mode of knowing a lot about things without being genuinely capable at them. The intervention isn't to stop studying… it's to stop treating study as a substitute for the experimentation that would actually build capability.

If you've been in the "reading more before starting" pattern for something you want to learn, please consider that this pattern might be actively preventing the learning you're trying to produce. The next book won't give you what actually doing the thing would. The next tutorial won't replace the feedback you'd get from a first attempt. Start experimenting. Even before you feel ready. Especially before you feel ready.

The specific practical move I'd suggest: for whatever you've been studying and not doing, identify the smallest possible experiment you could run this week. Something small enough that failure is fine, that requires minimal preparation, and that would give you actual feedback about what you've been learning. Do it. Notice what you learn from actually doing the thing versus what you'd been learning from reading about it. Most learners who try this discover a substantial gap between the two, and that gap points at what experimentation specifically produces.

In One Punch Man, Saitama's ridiculous training regimen isn't about studying the theory of getting stronger. It's about actually doing the exercises, day after day, producing the specific feedback that let his body actually change. Your learning works similarly for anything that requires actual capability rather than just theoretical understanding. Study is the map. Experimentation is the territory. You can't get to a destination by studying the map more thoroughly. Eventually you have to actually travel. Start today. Small experiments count. The learning that follows will surprise you.

Keep learning (and keep experimenting),

Ray