It's all about the U values. It's why you make heat exchangers out of copper and not ceramic.
The reason why SS holds more heat is that it's a poor conductor, which means it doesn't radiate it's heat effectively.
You could imagine it this way. If you were to take 2 blocks, one of SS and one of Ally and then heat them up to 100 degrees for hours, so they were 100 degrees right through and then chuck them in a bucket of cold water. The heat in the Ally one bucket would be passed to the water quicker. If you were to chop the blocks in half, there would be a sharper gradient of heat and a hotter core in the SS one.
The debatable point is whether you can dissipate the similar amount of heat via increasing the surface area, by using a tube. From what I remember you have 2 equations which can explain this.
Q=MCpdT where Q= the total amount of energy present. M= the mass of the stuff, Cp the specific heat capacity....how much energy it takes to heat stuff up by 1 degree and then the dT which is the temp change.
Then you have the other one, Q=UA LMTD. Where U=the sum of heat transfer coefficients (how porous the stuff is to heat....the key difference between ally and SS) A=the area for heat exchange to happen and LMTD the log mean temp difference.
The idea there being that you'd be able to analyse the Q gone into the bar by the first equation and knowing the mass of your lump. Then you'd be able to plug that into equation 2 and get a "what new surface area do I need if I make it out of SS" answer.
This would be an interesting thing for someone with more time on their hands to try out. HOWEVER, when the equation suggests "use a so and so % larger surface area" you can't just go assuming you can fit it all into a tube. The very nature of the U value problem meaning that the SS in the tube will not quickly reach an equilibrium temperature quickly. You'll have a cold back and a (very) hot front. You could reduce the mass of the material, which would mean that it reached operating temperature quicker and thus maximised the dT, which would efficiently shed heat, but the problem would be that it would be at a bloody hot temperature.
The alternative is to use loads of bars.
I've seen a few racks where they use a BIG ally no 2 bar and the rest are SS. This would work.
I think it's largely academic, because little people don't require as much friction to slow. So, it's the porkers and the bigger boned of us who create loads of heat and perhaps require bar heating to be considered in more detail.
I have glazed a rope using a petzl rack and share Rick's sentiment about bar wear. I have also considered SS bars and know they are faster....This means you would have to put more frictional area into play in order to gain the same friction/energy transfer.
My conclusion is that those people with the long racks have got the right idea. Spread the friction/wear/heating out.
This is an area I have given a fair bit of thought, with regards to getting some new bars made up.....(I have cained 2 racks to death in about 3 years)....so far the solution has been to buy another one!
I'll plug some numbers into the equations (and make some assumptions) and see what order of size we are talking about.
It would be interesting to measure the heat rise of 2 different weight cavers' 1 and 2 rack bars over an identical pitch with other variables the same....you could probably do this quite well by measuring the heat rise of a given amount of water in a set period of time by submerging them in a mug!!!!