Originally Posted by gfk_velo

Part of this is down to what you find to be an acceptable level of shifting and part of it is down to possible longer-term damage to components.

The older Chorus RDs from 2009-2015 (which it sounds as if you have) were not designed to mount a medium cage. In fact, it's not widely known that within the first 3 months or so of production, the specification was changed, anyway, to allow the use of a 29T biggest sprocket. Early examples were only capable of taking a 27T and a small modification was needed to those derailleurs to allow the use of a 29.

Yes, the medium cage for the later Chorus 11s RDs will "fit", technically - the assembly is the same on all versions, CH, RE and SR, pre and post 2015. The changes that were made on the HO version of the RD post 2018, when Campagnolo started to offer a medium cage version were:

1. The composite of the upper knuckle of the derailleur was changed to make it more resistant to twisting loads induced by a long cage, in a fully-crossed shift. You can see the difference if a pre-and a post 2018 Chorus RD are sat side by side.

Pre 2015 11s RDs all had alloy upper knuckles, not initially tested for these loads (the 11s Athena triple RDs did have medium and long cages but we never queried whether any changes were made to accommodate those new parts at the time Athena Triple was launched). The carbon parallelogram plate, though, has "ears" that fit over the knuckles - how resistant they'd be to twist loads I don't know. Again, not tested as far as the SCs were made aware.

2. Between pre 2015 and post 2015 rear parallelogram plates, changes were made to stiffen and lighten, to increase accuracy of shift. The plates were stiffened again when HO was introduced, which would suggest that the very earliest 11s RDs where probably not felt to be stiff enough to give good shifting with a medium cage if two steps of modification were made on the way to making HO ... I do know that the rear parallelogram plate was changed between Athena 11s (standard cage) and the Triple versions.

So, how much difference does all that make? It probably affects speed and accuracy of shift, certainly under load - but is it a deal-breaker? Maybe not. Wear and tear would probably have at least as much influence on shifting fluidity over time but of course, if you start from a better place, if degradation is linear, you should "end" at a better place - service life might be longer, because it will take longer before shift quality breaks down to the point where it becomes totally unacceptable.

The next factor is tracking angle. In an "old" RD pre 2015, the top jockey essentially follows, when viewed from the back, a near straight line across the width of the cassette. Because the parallelogram pivots are angled relative to the derailleur path, as the H screw setting is changed and so the upper body angle changes relative to the dropout, the angle of that line changes - broadly, the wider the range of the cassette, the "steeper" the angle of the line as the top jockey wheel is moved down and back by changing the relative tensions of the upper and lower pivot springs (that's what the H screw does). That means, in the middle of the cassette, where maximum accuracy is needed, the top jockey tends to be further away from the sprockets on a wide range cassette because the "cross-section" of the cassette gets gradually more concave as the ratios get wider (look at an 11-23 and then at an 11-32 sideways on, to see what I mean).

The 2015 RDs were designed to "follow" the shape of the cassette more accurately - the HOs used the same mechanism.

Why does this matter in the context of a pre 2015 RD?
The accuracy will degrade in the middle of the cassette with chain, sprocket and jockey-wheel wear (not noticeable, on say, a 12-25 but more so on a 12-29).
Add to that the fact that the mech is being forced further away from the wheel centre to accommodate the 32 but the angle of the line that the mech follows stays the same - so the top jockey is even further away mid cassette and further away on the small sprockets, so chain control is compromised again.

If you add that to the Shimano 11s cassette, starting on an 11T sprocket also being 0.3mm narrower than the corresponding Campagnolo cassette (at the 11-12-13T end, Campagnolo have slightly wider spacing, Shimano's is constant), you'll have the possibility - which we investigated trying various tricks to widen the range of the EPS RD, basically a 2011-geometry RD, on the teams - of the chain overshooting the top sprocket because it's not as well controlled as it should be, and jamming between frame and top sprocket.

Each of these things taken in isolation is pretty minimally important - add them all up and you might be generating a longer term problem, although I would say, a good care and maintenance regime will control most of that. As a rider, one gets used to slowly degrading quality of shift and it's only when everything is put back to "as new" that you realise how bad it had become (hence riders often saying to me after we've done a full service "it feels like a different bike") ... regular and careful servicing acts as a brake on the degradation.

Campagnolo, like Shimano and SRAM, when making recommendations, assume a certain level of mechanic skill in assembly and set up, they assume a certain level of maintenance (and make recommendations about service intervals based on that) and they control what they can control - hence advising against mixing and matching, because they are not in control of another manufacturer's standards or tolerances. They'd advise against what you have done not necessarily because of any safety issue - but because they can't guarantee the behaviour of something that they haven't tested in the lab / in the field, using parts over which they have no manufacturing control.

I doubt the breakage you had was directly because of an out of spec usage, btw - I've never seen it happen "quite" that way. More probably, it was an ageing effect.

Sorry, that's another really long and complicated answer and not an attempt to blind anyone with science - but in the increasingly complicated world of groupset specification & with increasing customer expectations, simple answers to apparently simple questions are getting harder and harder to come by!