A total and a breakdown are not the same information

A total score tells you where a judgement landed; component scores tell you how it got there, and only a breakdown shows which part moved. Two tools can both give a submission 7.2 and disagree about almost everything underneath it.

What a total keeps

A total is good for exactly one thing: comparing across a large population fast. It compresses proportion, symmetry, presentation and whatever else a rubric covers into a single ordered value, and an ordered value is easy to rank, sort and share. That is the whole case for it. Everyone understands a 7.2, nobody has to read a table to understand it, and a leaderboard of totals is legible in a way a leaderboard of six-axis tables is not.

What a total throws away

Everything else. A total cannot tell you which axis moved between two submissions, because the six numbers that produced it have already been folded into one and the fold is not reversible. It cannot tell you whether a low score reflects one weak axis dragging down five strong ones, or six mediocre axes agreeing. Those are structurally different results that a total prints identically.

It also cannot be audited from the outside. Change one variable, resubmit, and the total moves - but you cannot tell whether it moved because the variable you changed affected the axis you expected, or because it affected a different one, or because two axes moved in opposite directions and partly cancelled. A total that changed by three tenths could be one axis moving by two points or three axes moving by a tenth each in different directions. From the total alone, those are indistinguishable.

What a breakdown keeps that a total cannot

A breakdown preserves the shape of the judgement, not just its endpoint. Change the same variable and a breakdown will usually show which axis actually responded, which lets you attribute a result to a cause instead of just observing that a number moved. It also lets you spot a rubric problem a total hides completely: if every axis rises and falls together across a set of your own submissions, the tool has one real judgement wearing six labels, and a total would never have let you see that, because a total is what a fake breakdown collapses back into anyway.

The same total, two different submissions

Here is the case that makes the difference concrete. Two submissions, six axes each, both totalling 7.2 under a plain average.

Both submissions round to a 7.2 total. One of them is a genuinely balanced result with every axis clustered near the middle. The other has one axis well above average and one well below, and the total is hiding a spread the reader would probably want to know about before drawing any conclusion. Statistics treats that spread as its own quantity, not a detail of the average: NIST's Engineering Statistics Handbook calls characterising the spread, or variability, of a data set "a fundamental task in many statistical analyses". A tool that only shows the total presents these two submissions as identical. They are not identical - they are two different shapes that happen to average to the same place, and averaging is a lossy operation by design.

What this costs you as a reader

A breakdown is harder to brag about, which is exactly why plenty of tools do not offer one. A single number is a cleaner share card and a simpler product to explain, and turning six judgements into one is a legitimate design choice for a tool built around comparison rather than diagnosis. But it is a choice, and it has a cost, and the cost is specifically the ability to tell why a result looks the way it does. Rate Cock is one example of the other option: it reports all six axes on public entries rather than a bare total, so a viewer can check the shape underneath a number rather than trusting the number alone.

Whether a tool's weights turn six sub-scores into a plausible total, or into something a plain average would not predict, is a separate question from whether it shows a breakdown at all - how the combination actually happens is worth its own look. Human review sidesteps this trade-off in a different direction: a person writing a response can name what stood out in prose without ever forcing it through a numeric axis, which is neither a total nor a breakdown but a third shape entirely. Measurement sidesteps it differently again - a length in centimetres is neither a total nor a breakdown, it is a single reported value from a stated method, and it does not compress several judgements into one because it was never several judgements to begin with. And the underlying model producing any of these axes in the first place is a separate layer worth understanding on its own terms, since a breakdown is only as informative as the independence of the numbers feeding it.