Climbing grade drift

Why does a climbing route rated hard in one decade quietly become a moderate one later without a single hold changing?

The rock has not moved. The holds are where they were, in the same order, at the same angles. Yet a route that was a serious undertaking in one guidebook is listed a grade or two lower in the next edition, and nobody committed a fraud to make that happen. So either the guidebook was wrong then, or it is wrong now — or a grade is not the kind of thing that can be wrong in that way at all.

What exactly is a grade measuring, if the thing it measures is unchanged?

Try the obvious reading first: a grade is a property of the rock, like its height. If that were true, drift would be measurement error, and errors should scatter in both directions. Do they? Overwhelmingly the drift on established routes runs downward. A systematic direction is a strong hint that we are not looking at noise around a fixed value.

So a grade is not a property of the rock alone. It is a relation — between the route and a climber attempting it. And relations move when either side moves.

Which side is moving? Consider what the first ascensionist actually did. They arrived at an unclimbed piece of rock, not knowing where the holds were, which foot went where, whether the crux was a lock-off or a swing, or even that the sequence existed. They solved it. Then they proposed a number by comparing the whole experience against other routes they had done.

Now consider the twentieth ascent. That climber knows the line, knows the holds, and knows the sequence, because it has been described, filmed, and marked with chalk. The rock is identical. The problem, however, has been solved already and the solution handed over free. This is the largest and most underrated component of drift: the route did not get easier, the puzzle got answered. Which suggests a prediction we will come back to.

Second, look at the scale itself. Climbing grades are ordinal — a rank order, with no unit and no physical reference standard. There is no kilogram of difficulty kept in a vault. A grade means "about as hard as these other routes", so the scale is anchored entirely to a set of reference climbs and, through them, to the technique, training and expectations of the people who climb them. Add fingerboards, indoor walls, professional climbers and forty years of accumulated technique, and the reference set is climbed by a different population. Meanwhile the top of the scale keeps extending upward as harder routes are established, which compresses everything below it in perception if not in definition.

Third, and more speculative, is who gets to have an opinion. A climber who fails on a route usually contributes nothing to its grade — the votes come from ascents. It is often argued that this biases consensus downward over time as traffic increases, though there is a counter-argument that climbers have every incentive to keep a route's grade high once they have ticked it, and I do not think the net direction of this particular effect is settled.

Finally, the question's own premise deserves a poke, because the rock does sometimes change — in both directions. Holds break and traffic polishes footholds smooth, both of which make a route harder, while chalk and boot rubber mark the useful holds, which makes route-finding easier. Physical change is real but is not the story, and is not even reliably downward.

That last observation is the test. If beta accumulation is really the dominant driver, then drift should be unequal across route types: routes whose sequence is obvious from the ground — a uniform crack, a bare slab, a pure endurance pitch — should hold their grades far better than routes with a hidden or counter-intuitive crux, where knowing what to do is most of the battle. If instead every route of a given vintage drifted down by about the same amount regardless of how much its difficulty was informational, then beta would not be doing the work and the re-anchoring of the whole scale would be the entire explanation. As far as I know this has never been measured properly — grade drift is documented through guidebook editions and argument, not through data — so treat the weighting between the two mechanisms as unresolved.

It is worth separating this from the more familiar drift in school grades, because the two look alike and work oppositely. Educational grades drift upward, driven by assessors under pressure from the people being assessed, applied to work that is not a fixed object. Climbing grades drift downward, applied to an object that is physically fixed, and the incentives run the other way: first ascensionists and repeaters alike generally resist downgrades, because a lower number diminishes what they did. The drift happens anyway, which tells you it is not being driven by anyone's interest. The fixed point of difference is the direction of pressure — in one case the measured party wants a better number, and in the other they want a bigger one and still do not get it.

• Why onsight and redpoint grading conventions exist, and how much of the beta problem they actually absorb.
• Whether indoor grading, where routes are reset constantly and the setter is known, drifts differently or at all.

Every term the collection defines is gathered in the glossary.