Why Aspirants Misjudge Which DILR Sets Are Really Doable

DILR

Some candidates pick well and some do not, and the gap is not reasoning ability. A candidate who can solve every set given time will still lose the section if the sets they enter are the expensive ones, and they will usually conclude afterwards that the paper was hard.

Misjudging which sets are doable is a specific, diagnosable failure with four common causes, and each has a different fix. More importantly, it is measurable from your own practice, which means you can find out which cause is yours rather than resolving to choose better. This piece sets out the four and the drill that exposes them.

Doable and Cheap Are Different Questions

The root confusion sits right at the moment of selection, and everything else follows from it.

When candidates assess a set they are asking whether they can do it. That is a reasonable first filter and it screens out what is genuinely beyond reach. What it does not tell you is how long doing it will take, and those two properties are only loosely related.

A set can be conceptually simple and computationally enormous. It can be immediately clear and then require testing fourteen cases. The clarity you felt was about the first question and told you almost nothing about the second, which is why the overrun feels like a betrayal: the set kept its promise and you misread what the promise was.

Four Causes of Misjudgement

Surface Reading Is the Most Common

It is also the one whose main proxy runs backwards, which is what makes it so costly.

Candidates treat short sets as easy. But constraints are what let you deduce, so a set with many specific conditions has handed you many tools, while a four line set leaves the structure open and forces you to carry several configurations at once. The long set is frequently the cheap one.

Tidy presentation and a familiar-sounding scenario are equally unreliable. Presentation is a formatting choice and the scenario is decoration; neither says anything about how the conditions behave. Candidates use these proxies because they are visible in a second, which is exactly why they do not work.

Not Checking Whether Anything Is Forced

The second cause is the single most predictive signal, and it takes seconds to check.

Ask whether any first placement is forced: a condition, or a pair, that pins something down with no choice involved. A set offering that will usually cascade, because a forced placement tends to make the next one forced too.

If nothing is forced and every condition still permits several arrangements, you are in a branching set whatever its length or tidiness. Branching does not add cost, it multiplies it, because every new condition has to be tested against every open configuration and the bookkeeping starts consuming more attention than the reasoning.

Estimating From the Wrong Version of Yourself

The third cause is subtle and it affects strong candidates most.

Your sense of how long a set type takes comes largely from practice, and most practice is untimed or loosely timed, in a comfortable room, with no competing decisions. The number you carry is your untimed speed.

Inside a section you are slower: the clock is running, you are tired, you have decisions competing for attention, and working memory is degrading. A candidate estimating with untimed numbers will systematically under-price every set, and the error compounds because they enter more sets than they should.

Cause	What it looks like in your data	Fix
Surface reading	You skip long sets that turned out cheap	Read every set, judge on forcing not length
Missing branching	Entered sets that never closed	Check for a forced first placement
Speed optimism	Estimates consistently under actuals	Estimate from timed practice only
No feedback loop	You cannot say whether estimates are good	Record prediction and actual per set

The Fourth Cause Is Why the Others Persist

The first three are correctable and they do not correct themselves, because nothing tells you they are happening.

An estimate improves by being checked against reality repeatedly. If you never record what you predicted a set would cost and what it actually cost, no amount of practice calibrates the estimate, and a candidate can do hundreds of sets while their judgement stays exactly where it started.

That is why this cause underlies the rest. Install the loop and the other three surface on their own, because your own data shows you skipping the cheap sets or under-pricing everything by four minutes.

Log the Sets You Skipped

One column matters more than it looks and it is the one everyone omits.

Record whether you attempted each set, not just how the attempted ones went. The sets you skipped are the ones you have no outcome data on, so a systematic error in skipping is invisible without deliberately checking.

Periodically go back and solve a few sets you skipped, untimed, and see what they would have cost. Most candidates find one specific family they systematically avoid and systematically could have done, and correcting that single habit changes the section more than any technique.

What a Good Selection Read Contains

Since the causes are known, the read that avoids them is short and specific.

Sweep every set under a hard budget of under a minute each, asking three things of each one. Do I recognise the family and the structure it wants? Is any first placement forced by the conditions? Do the questions ask for one thing from the structure or several different things?

Those three answers give you a rough cost, and cost is what you rank by. Then start with the cheapest set you are confident about rather than the most interesting, because the interesting set is frequently expensive: what makes it interesting is usually an unusual structure, and unusual means no recognition to draw on.

What the Log Tells You After a Month

The drill is worth doing for its own sake, and the report it produces after ten or fifteen sections is where it pays properly.

Line up your predictions against your actuals and look for the shape of the error rather than its size. Estimates that are consistently short by a similar margin point at speed optimism, which is fixed by recalibrating from timed practice rather than by trying to be faster. Estimates that are accurate on some families and wildly wrong on others point at a recognition gap in those families specifically.

Then look at the sets you skipped. If the ones you later solved untimed turned out cheap, your threshold is set too conservatively and you are leaving workable sets unread. If they turned out expensive, your skipping is sound and the problem is elsewhere, which is useful to know because it stops you correcting something that was already working.

Most candidates find one dominant pattern rather than a general unreliability, and that narrowness is the good news. A single systematic bias, once named, is correctable in a fortnight of deliberate practice, while a vague sense of choosing badly can absorb months without improving.

The Hard Part Is Behavioural

The procedure is not complicated and executing it is, and the difficulty is not analytical.

Once you have read a set you can clearly do, it is almost physically difficult not to start it. The sweep collapses at the second set, and what you have then is not selection but acceptance of the first workable thing you met. Since there is no difficulty gradient, that is close to random.

Expect the sections where you install this to feel worse, because minutes go into reading rather than solving and the return arrives later. Judge those sections on how many sets you read rather than on the score, for the first three or four attempts.

The Summary

Misjudging doable sets is not a reasoning problem. It starts with confusing doable and cheap, since the selection read usually only asks whether you can do it and never asks what it will cost.

Four causes follow. Surface reading, where length and tidiness are used as proxies and the length proxy runs backwards. Missing the branching, which is the single most predictive signal and takes seconds to check. Optimism about your own speed, because the number you carry comes from untimed practice. And no feedback loop, which is why the other three persist.

Install the loop first: predict attempt-or-leave and a time per set, record the actual, and log the sets you skipped. That surfaces the other three from your own data. Then run a proper sweep, judging on whether a first placement is forced rather than on appearance, and start with the cheapest set you are confident about.

If you cannot say whether your cost estimates are accurate, no amount of further practice will calibrate them. A CAT preparation strategy review will identify which of the four causes is yours, and a personalised CAT preparation plan schedules the estimation drill alongside the solving.