Wittgenstein, knowing no financial necessity, had no such creative dialogue between form and error.
The Craftsman (p. 258)
Much research today is done with the support of a grant and is expected to be completed within a specific period. These projects are subject to external constraints that do much to establish the "form" of today's research practices and their written expression. An important effect of these constraints has to do with how you react to error.
Sennett devotes part of The Craftsman to a comparison of two houses in Vienna: the house Ludwig Wittgenstein built for his sister in Kundmanngasse and the Villa Moller, built by Adolf Loos. "When the foundations were not laid as specified, [Loos] could not afford to dig them up and start again; instead, Loos thickened the form of one side wall to accomodate the mistake, making the thickened wall an emphatic side frame for the front" (p. 258). Wittgenstein, by contrast, was not just building a house: "I am not interested in erecting a building, but in ... presenting to myself the foundations of all possible buildings" (p. 254)*. (Yes, that looks a lot like Wittgenstein's early philosophical project, but more on that some other time.)* So, as Sennett notes, he found himself having to rebuild a part of the house in order to raise the ceiling about an inch. He was working without constraints because his client was his sister and both of them
werehad been heirs* to one of the largest fortunes in Europe at the time.
Loos had no such luxury, and it is his example, not Wittgenstein's, that we are therefore bound to follow. (In philosophy, I sometimes think, we are likewise bound to follow Heidegger's example rather than Wittgenstein's for the same reason. Again, more on that later.) Once the foundations of your project have been laid, you can't just tear them up if they don't give you what you need—or rather, what you thought you needed. You have to rethink the building on the foundations you happen to have; you can't keep tearing them up and starting over. You must establish a "creative dialogue between form and error".
*Do read Presskorn's very illuminating comments and corrections below.