Here’s a question worth sitting with for a minute.
You spent thirteen years getting good at exams. How much of what you crammed for last year’s finals could you actually use today, without looking it up?
For most people the honest answer is: not much. And that’s not a personal failing. It’s the predictable result of a method. The way most of us were taught to study is very good at one thing, passing a test on a known date, and close to useless at another, becoming genuinely good at something hard. Engineering needs the second thing. So it’s worth understanding why the first thing doesn’t get you there, and what does.
I lead programme design at Kalvium, which means I read the research on how people learn and then have to make it operational. What follows is the short version of what actually holds up.
Cramming works. That’s the trap.
The reason cramming feels effective is that it works, briefly.
In 1885 a German psychologist named Hermann Ebbinghaus mapped how fast we forget. His finding, replicated many times since, is that forgetting is steep and roughly exponential when nothing intervenes. Learn something once, do nothing with it, and most of it is gone within days. I’ve written about the forgetting curve and what it means for how a programme should be designed at more length, but the student-level takeaway is blunt.
The night-before cram builds a kind of memory that’s designed to evaporate. It’s enough to recognise the right answer on the paper the next morning. It is not enough to reach for that idea, cold, eight months later, when a real problem needs it. Exams forgive this. They test recall of specific content on a date you know in advance. Engineering does not forgive it, because the work is applying half-remembered ideas to problems you’ve never seen.
So the goal isn’t to study more. It’s to study in a way that leaves something behind.
Retrieval beats re-reading
Here’s the single most useful finding, and almost nobody acts on it.
Henry Roediger and Jeffrey Karpicke ran a series of studies starting in 2006. The setup was simple. Students learned some material, then either re-read it or were tested on it. The re-readers felt more confident. They reported feeling like they’d learned more. When everyone was tested a week later, the students who’d been tested the first time, not the re-readers, remembered more.
This is the testing effect, and it’s held up across dozens of studies. Pulling an idea out of your memory strengthens it. Putting it back in by re-reading barely does. The catch is that retrieval feels worse. It’s effortful and a little uncomfortable, and re-reading feels smooth and productive, so students pick the thing that feels good and learn less.
What this means in practice is small and specific. Close the book and try to reproduce the idea from memory. Attempt the problem before you look at the solution. Explain the concept out loud to nobody, and notice exactly where you go vague. The struggle to recall is not a sign it isn’t working. The struggle is the thing that works.
Make it harder on purpose
Robert Bjork, a psychologist at UCLA, spent decades on a connected idea he called desirable difficulties.
His argument is that certain things which make learning feel harder in the moment make it stick better in the long run. Spacing your practice out instead of cramming it together. Mixing different kinds of problems instead of doing twenty of the same kind in a row. Testing yourself while you still feel underprepared. Each of these lowers your performance today and raises your retention next month.
That trade is deeply counterintuitive, which is why so few students make it. Doing twenty near-identical problems back to back feels great, because you get fast and it feels like mastery. Then a mixed set the following week wipes that feeling out, because you were building fluency, not durable skill. The discomfort of mixing and spacing is the price of learning that lasts. Bjork’s work is decades old, and most study advice still ignores it.
Deliberate practice, not just practice
The last piece answers a question people rarely ask: why do some people who practise for years never get much better?
K. Anders Ericsson spent around thirty years studying expert performers, and his answer was that the amount of practice matters far less than the kind. He called the effective kind deliberate practice. It isn’t mindless repetition. It’s practice with a specific goal, aimed at the edge of what you can currently do, with immediate feedback, repeated on the exact weak spot until it improves.
The distinction is everything. Solving fifty problems you already find easy is practice, and it builds almost nothing. Finding the one kind of problem you keep getting wrong, and drilling it with feedback until you don’t, is deliberate practice, and it’s worth ten times as much per hour. Comfort is the enemy here. If the practice feels easy, it’s probably maintaining a skill rather than building one.
Why engineering rewards a different kind of learning
Put those four together and a pattern shows up. The methods that build durable, usable skill all share one property: they feel harder while you’re doing them.
Exams let you get away with avoiding that discomfort, because they mostly test recognition of specific content you were told to prepare. Engineering doesn’t let you get away with it. The job is transfer, taking a pattern you learned in one place and applying it to a problem that looks nothing like where you learned it. Recognition memory is useless for that. What you need is the deep, retrievable, well-practised understanding that only the effortful methods produce.
That’s the real case against studying for exams and calling it learning. It isn’t that exams are evil. It’s that optimising for them trains exactly the wrong muscle for the work you actually want to do.
Why Year 1 here is retrieval and shipping, not lectures
This is the part I get to make real rather than just write about.
If the research says durable skill comes from retrieval, spacing, desirable difficulties, and deliberate practice, then a first year built mostly of lectures and one exam per semester is designed against the evidence. So Kalvium’s first year is built the other way around.
There’s a Learning How to Learn course in the very first semester, because the meta-skill is treated as a skill, not something students are assumed to already have. The DOJO runs daily coding practice, six days a week, which is retrieval and deliberate practice turned into a habit rather than an event. Students ship real, working software from Year 1, and building is a sustained retrieval exercise, because you can’t build without pulling earlier ideas back out and applying them to something new. Capstone-style projects start in Semester 2, not Year 4, so material keeps coming back in harder forms instead of being crammed once and abandoned. And the HEROS system tracks what each student is actually struggling with, so practice can be spaced and pointed where it’s needed rather than sprayed evenly.
None of this is exotic. Every piece of it sits in research that’s been public for years. What’s uncommon is arranging a programme so the research, rather than the exam calendar, decides the shape of the week.
Where the evidence gets thin
Intellectual honesty requires saying what isn’t settled.
The basic findings are solid. The forgetting curve, the testing effect, the general value of spacing, these are among the better-established results in the field. But the details are genuinely open. The optimal spacing interval isn’t a solved problem, and anyone selling you a precise schedule that “just works” is overselling it. Whether these effects transfer cleanly to something as complex as engineering, rather than to word lists and facts, is still being worked out. We design as if they do, because the underlying mechanism, effortful retrieval, plausibly applies, but we hold that with some humility. The strongest honest claim is that the approach is consistent with the best available research, not that it’s proven optimal.
My advice, if you remember one thing
Stop measuring your studying by how smooth it feels.
Smooth is the warning sign. Re-reading feels smooth. Highlighting feels smooth. Watching one more explanation feels smooth. The methods that actually build durable, transferable skill, closing the book and recalling, spacing your practice, mixing your problems, drilling the exact thing you’re worst at, all feel effortful and slightly unpleasant. That feeling is not the problem. It’s the receipt. Learn to trust it, and you’ll get good at hard things far faster than the people around you who are still optimising for comfort.
This piece sits under a broader argument: that a college should build an adult, not just award a degree. For the fuller version of the learning science and what it means for programme design, see why most engineering programmes don’t produce engineers. For the durable skills all of this is ultimately in service of, there’s the skills AI can’t replace that degrees still skip. And for families weighing programmes, the framework for choosing a B.Tech CSE programme and the complete guide to what Kalvium involves turn all of this into a practical decision.
Arvind is Head of Programme Design and Delivery at Kalvium. He writes about the cognitive science of learning and how it shows up in the design of a live programme, grounded in named research and honest about where the evidence is still being worked out. Read more from Arvind or browse the B.Tech category.