1. Structurally, COB needs to be thick enough to handle loads....A baseline is 300 mm (12"). It cannot take alot of compression load, as in high levels of snow and a heavy weight roof so it needs to get thick which adds cost to the foundation. 2-6 PSF compared to a 3000+ PSI OPC concrete wall ~20-50 PSF. In high winds and earthquakes it needs a support structure especially at top and corners, less weight at top to resist shear (take your hands and rub them together that is a shear plane, clays do not do that great in shear can be much lower than compression).... Rebar in part is used in concrete for this purpose and tension. Steel can take 80,000(construction grade) - 300,000 (stainless) psi of compression, bearing.
2. Physically, less is more. Finding the right thickness is difficult. Dry climates need less since clay moisture index or storage compacity requirement is less, that is given the same type of clay....If one used kaolin it needs to be VERY thick even in dry climates, magnesium bentonite less, sodium bentionite or "swelling clay's less". They are just as many horror stories of too much mass in the wrong places and wrong climate zones, over heating homes, or cold walls, high humidity, swelling and shrinkage from the wrong clay's, cracks, bad ratio's, stains, water damage, microbials, etc....same with straw and clay or lime renders.
The problem with "u-value" or "r-value" of just looking at conductance of heat transfer through a medium such as clay for thermal bridging, is it does not account for the other methods of heat transfer, convection, radiation, phase change (internal, hidden). Take a steady state source of radiant heat like a heat gun point it at the COB measure the temp on both sides you just conducted the u-value or r-value test, what is known as a 'hot box" test.....Now lets get real add some wind, rain, pressure gradients, freeze-thaw, humidity, pollution, etc....Now try and simulate that in all climates around the globe in a lab. Not going to happen! Now perhaps we see why International building code gives it(concrete) a simple r-value of 5-10 depending on thickness, because it is too complex otherwise. Ideally, it is best to have just enough thickness to get a small exchange of heat from one side of the wall to other to keep it dry, and phase change to create heat or cooling in the wall that only gets better in high humidity. We'll I have been apart of that hot box testing model( very expensive software in the tens of thousands per license and steep learning curves) and lab simulation in multi-million dollar labs, BSC (Building Science Corp has to a lesser extent). As a result, we now have what manufactures should put on labels "effective" in situ r-value that has dynamic affects. Knock down fiberglass value 10-40%, loose fill cellulose 10-20%, mineral wool 3- 5-%. That is what it takes experience wise to understand it. Now go out to forums get millions of opinions in super impressive big worded post on the subject now you have popcorn entertainment who needs redbox
I run the fluid dynamic models at my job since they can afford me, and my home designs because I can. Most homeowners cannot. So there are some steady state and "dynamic" models in this industry that are rarely accurate (WUFI, few others) Needs a good seat jockey, trash in trash out. I cannot predict this behavior without the proper tools(models, lab, etc) so how do people think they can guess? I tell you what, consider yourself very lucky if you guessed correctly. Most that put hundreds of thousands into their homes could end up being eaten alive by very uncomfortable sick homes, and to tear it out and start over very expensive.
COBs outer skin or surface is not the correct pore size to handle liquid water, it needs to be modified to make it more durable, especially when exposed to the elements of nature.
Ben, I do not see anything wrong with the insulation core, COB renders at varies thicknesses given it meets the two minimum requirements I noted above. One issue with cellulose blown is it need to have the right density ~ 3 lbs ft3 and it can settle over time, so can straw. The rigid mineral board I suggested has a compression value of 750-1200 PSI (5-10) PSF which is much better than foam ...also foam cracks @ 10% deflection, mineral board take 25%..so it can add to the COb composite flexure, compression, and shear strength better than the others you mentioned but it cost more up front. If you add the wood panels skins to mineral wool core all the mechanical strengths noted have improved especially if tied to a upper bond beam, corner and center post of the correct spans and size....again for large winds and earthquakes only not Wyoming persay.