Every revision guide ends at the same place. Sit the past paper, then mark it against the scheme.
That second instruction is treated as the trivial bit. It is not. It is a genuinely difficult judgement task, performed under the worst possible conditions, by the person with the strongest reason to get it wrong, and it is where a large amount of revision quietly stops working.
Worth being precise about why, because the usual explanation is wrong.
It is not about being soft on yourself
The standard advice is “be strict when you mark your own work”, which assumes the problem is willpower. Some students are generous with themselves, certainly. But the main failure survives complete honesty, and it happens to conscientious students marking carefully.
The problem is the order of information.
To mark your work you have to open the mark scheme. The moment you do, you know the answer. And you now have to answer a question you can no longer answer honestly: would I have got there?
This is hindsight bias, and it is one of the most robust findings in psychology. Once you know an outcome, you cannot reconstruct your uncertainty before you knew it. The answer looks obvious in a way it demonstrably was not thirty minutes ago, when you were staring at the question with no idea what to do.
So the student thinking “yes, that is what I meant” is not usually lying. They are experiencing something that feels exactly like recognition and is indistinguishable, from the inside, from having known it.
Where it bites hardest
The damage is not evenly spread. It concentrates in specific places.
Method marks. This is the worst one. The scheme says M1 for a complete method to find the surface area. You look at your half-finished working. Is that a complete method, or the beginning of one that would have gone wrong? You genuinely do not know, and the scheme cannot tell you, because it is describing a category rather than showing you a script. Most people award themselves the mark. Across a paper, that is several marks of drift.
Anything marked in bands. Extended answers are placed in a level by an examiner who reads hundreds of them and has a calibrated sense of what a middle band answer looks like. You have read one, yours. You have no comparison set. Self-marking a six mark question is not hard because the criteria are unclear, it is hard because you have nothing to compare against.
“Or equivalent”. Schemes accept alternatives, and deciding whether your different-looking answer is equivalent or merely adjacent requires exactly the subject knowledge you are trying to test.
The near miss. You wrote 3.47 and the scheme says 3.5. Sometimes that is full marks and sometimes it is zero, and which one depends on notation elsewhere in the scheme that most students do not read.
What it costs
A self mark that is a few marks generous per paper does not just produce a slightly wrong number. It produces the wrong revision plan, which is more expensive.
If your self-marked scores keep landing near a grade 6, you stop worrying about the topics inside them. The marks you awarded yourself for methods you might not have completed are invisible in the total, and they are exactly the marks that will not appear in June. The paper told you the truth and the marking stage removed it.
That is the real cost. Not the inflated score. The problem you did not go and fix.
Five things that make self-marking meaningfully better
None of these are hard. All of them work by interrupting the order of information.
1. Mark blind first. Before opening the scheme, go through the paper and write down what you think each question earned, and why. Then open the scheme and mark it properly. The gap between your two judgements is the most useful thing on the page. It tells you where your sense of what a good answer looks like is miscalibrated, which is a skill you can improve and which will be running in the actual exam.
2. Leave a gap. Mark the paper a day later, not immediately. You will have forgotten what you were thinking, which sounds like a loss and is a gain: you will read your own working the way a stranger would, which is how it will actually be read.
3. Judge the page, not the memory. The only legitimate question is “is this on the paper?” Not “did I know this?” If your reasoning was in your head, it scored nothing, and marking it as though it counted is training yourself to keep leaving it there.
4. Swap with someone. Marking a classmate’s script is dramatically easier than marking your own, because you have no memory of their intentions to contaminate the judgement. You also learn more than they do from it, because comparing an unfamiliar answer against a scheme is exactly the calibration exercise you are missing. Swap, then mark, then compare how you each marked.
5. Record the cause, not the score. For every mark you did not award, write one word: unknown, misread, rushed, not written down, out of time. That list survives contact with hindsight bias in a way a total does not.
What self-marking is still good for
None of this means do not do it. Self-marking has a real benefit that has nothing to do with the accuracy of the number.
Reading a mark scheme carefully teaches you what is being rewarded. A student who has marked ten of their own scripts against real schemes writes differently in an exam, because they have seen ten times over that the credit sits in the substitution line and the connective and the unit. That learning survives even when the score attached to it is generous.
So the honest position is: self-mark for the calibration, not for the number. Treat the total as soft. Treat the list of lost marks as the output.
Where an accurate second opinion changes things
The one thing self-marking cannot give you is a mark produced by someone who did not watch you write it. That is what a teacher’s marking is for, and the reason students do not get more of it is not that teachers do not want to give it. It is that marking a class set to question level takes hours that do not exist, which is why most practice papers get a total and a comment rather than a mark by mark breakdown.
That is the gap Marky was built for: a class set marked against the official scheme with every mark accounted for individually, so the breakdown exists for every student rather than only for the ones who got a detailed conversation. It is not a substitute for a teacher’s judgement, and the teacher can change any mark it awards. It just means the first pass exists at all.
We are fairly blunt about the limits of that elsewhere, including how to audit it on your own scripts before trusting it.
Where Marky fits, and where it does not
Being straight about this, because the article above is partly an argument for something we sell.
Marky does not solve hindsight bias. Nothing does. What it changes is the availability of a mark that was produced without watching you write it: a class set marked against the official scheme with every mark accounted for separately, which is the thing teachers would give more of if the hours existed. It is a first pass, not a verdict, and any mark it awards can be changed by the teacher and is recorded as theirs.
It also cannot mark a paper without the official scheme, so it is no help at all on a script you have just sat and whose scheme has not been published yet. If you are weighing it against anything else on the market, the questions that actually separate these products are set out here, ours included.
If you do one thing with this article, though, make it the blind mark. It costs one lesson and it needs no software.