A cross-walk between rubrics, and where it breaks

You can map one tool's axes onto another's only by matching what each axis is described as scoring, never by its name, and only for direction and rank, never raw numbers. Two "presentation" axes can share nothing but the label. Past a certain mismatch in rubric shape, the honest move is to stop mapping.

Match by description, not by name

The label on an axis is a product decision, not a specification - the same word covers different things on different tools, so the name is the least reliable part of the match. Start instead from whatever description each tool gives its axis, if it gives one at all: what does the tool's own documentation, or the prose attached to a result, say the axis is actually scoring? Two axes with different names but overlapping descriptions - one tool's "definition" and another's "clarity," say - are a better candidate match than two axes sharing a name but described differently underneath.

Mark what doesn't match

Not every axis on one tool has a counterpart on the other, and forcing a match where none exists is worse than leaving a gap. If tool A has six axes and tool B has four, expect at most four honest pairings and two axes on tool A with nothing to map to. Write those down as unmatched rather than folding them into the nearest-sounding axis on the other side - a forced match produces a cross-walk that looks complete and is actually just wrong in two of its rows.

Never compare the numbers directly

Even a well-matched pair of axes should not have their raw scores compared to each other. Each axis carries its own internal calibration even within one tool, and that problem compounds across two tools that were built and tuned independently. A 7 on tool A's proportion axis and a 6 on tool B's matched axis are not evidence that tool B is stricter on that specific quality - they are two numbers from two different distributions that happen to use the same digit range. Clinical researchers hit the same trap with instruments meant to measure one quantity: Bland and Altman (1986) argued that agreement between two methods needs its own analysis, and that "the use of correlation is misleading" as a stand-in for it. What you can compare, cautiously, is direction and relative position: does the matched axis rank this submission similarly against your own other submissions on each tool, not what the raw number says.

Where the cross-walk breaks down entirely

Some pairs of rubrics resist this whole exercise. A tool with three broad axes and a tool with twelve narrow ones don't decompose into any clean one-to-one mapping - the twelve-axis tool has drawn distinctions the three-axis tool never made, and forcing a match collapses information rather than translating it. Past that point, the honest move is to stop trying to align axes and compare the tools on order alone: does the twelve-axis tool and the three-axis tool rank the same set of submissions the same way, leaving the axis-level detail unmapped.

Where this fits

A cross-walk is a structural exercise - it tells you which axes are worth comparing at all, before you get to the narrower question of what it means when two specific, matched axes disagree on one result. The naming problem underneath all of this - why the same word means different things on different tools - is worth reading on its own if the cross-walk keeps failing on vocabulary rather than substance. None of this applies to a measurement in the ordinary sense; a length taken with a tape doesn't need cross-walking against a rubric axis, because it isn't reading the same kind of judgement in the first place. The underlying models behind two different tools are often closer than their rubrics suggest - what the model is actually doing under the hood varies less across tools than the vocabulary layered on top of it does, which is part of why a careful cross-walk can find real matches even between tools that look unrelated on the surface. A human panel resists this exercise almost entirely, since what a written review is actually responding to rarely decomposes into fixed axes to begin with. Rate Cock's own six-axis breakdown is a reasonable anchor to cross-walk against precisely because it publishes both the axes and the results they apply to, which is the material a cross-walk needs and most tools don't supply.