OK, i can't believe i actually did this. This time i'll use some visual support to explain it and not going past this because it would require one of these things:

a) advanced numerical methods, which is boring as hell and takes alot of time to do.

b) coding it...but i don't code anything for at least 5 years so it also may end taking way more time than it should.

Anyway, neither of above options are worth doing just to brag rights so, instead, i made easy-to-understand graphs (that are very crappy so pardon me if not everything is aligned or keeping the right proportions...)

Here we have the graph. At X is the input. The number of symbols or dice. Y is the result, what you get from that input.

When we do symbol input we have this:

It reflects the linear growth of it...1 sword will always result in 1 sword, 2 swords in 2 swords and so on. This, though, cannot be used on die. When we make a graph for dice we may see something like this:

Just imagine that it's a spreadsheet of results. The size of dots reflect the number of results you got from a sample at that input. The bigger the dot the more often that result happened. So when we calculate the average result and make a graph of dice it would be like this:

The pointy corners reflect the variable aspect of it. So, now we placing both together we have:

Notice that in the end the results are pretty much the same. The sample is what defines how "pointy" the dice graph is...the bigger the sample the closer it gets to symbol graph.

But now the question...and what wyld cardburn do to it? It does this:

In this image B is the biggest distance it can get from symbol graph, which represents the overburn (all dice were burned with Wyld so instead of "0" the symbol graph now is the minimum result dice roll can offer).

Notice that, though, as you add more non-burned dice the graph slowly gets closer to symbol again because the impact the cardburn causes when compared to the number of dice starts dimishing. When you burn 1 Wyld and have 1 die you actually caused overburn...the minimum result is the symbol graph. If toy have 2 dice, the same carburn would affect only 50% of the dice...if you have 10 dice it would affect only 10% of them. That's why it moves back closer to symbol graph.

If you feel like testing it yourself, just get a paper and, couple dice and write down the results as follow:

a) 1, 2 = adds +0 to total

b) 3, 4, 5 = adds +1 to total

c) 6 = adds +1 to total and roll that die again.

You'll notice that the more you roll for that input the closer it gets to symbol result (number of input = number of result, like rolling 2 dice ending with 2 symbol) because every time you roll you're increasing the sample that will be used to calculate the average point.

Now believe me if you wish or try it yourself and you'll see it's correct. As said at start i'm not going any deeper in this subject than this and also not really try to convince anyone that it's correct anymore.