Are there intermediate steps? In the creation of this data, yes. The monkeys bang out characters one after another rather than all at once.
You are being deliberately obtuse. Whether it takes a monkey a week to type 41 characters or 5 seconds, the data is examined only as a unit. But that is not the only way that your sequence can form.
What if there was another way. Say, one monkey types "question." right now. Two days later, another monkey types "To be or n" while a third monkey manages to get the sequence "ot to be, tha". A week down the road, the first monkey types out "t is the ".
These fragments are cast out into the big wide world and wander around and bang into each other. When the ends align, the fragments link up and your sentence is complete.
Those are the intermediate steps that are not accounted for. The author assumes that each trial is exactly the right length every time because the calculations cannot work for anything else. That is why the application of these calculations is incorrect, and your article proves exactly nothing.
Then you don't understand that the mathematical probability remains the same whether you flip ten coins in one group, 15 coins in another, and 16 coins in still another group for a total of 3 intermediate steps, or one coin 41 times.
The final probability for randomly hitting a pre-determined outcome of 41 units is precisely the same either way.
That's what the math addresses: the final probability of data in a chaotic, unintelligent environment correctly sequencing itself. Except in this case, we're dealing with much lower probabilities of getting each correct character out of far more possibilities than a mere two sided coin toss...