Thanks again for your comment, as you replied both here and on the blog, I'll repost my reply here:
John, thanks for your detailed reply. Sorry I could not respond sooner. I realise that my writing was imprecise in places, so I've edited the story very slightly to describe the situation better.
To clarify: the thatcher in the story did actually do some investigation upfront. I've changed the wording slightly to emphasise this. The problem in general, though, is how do you know when you've done enough risk assessment? The only way to eliminate _all_ risk is to actually do the project, but that is clearly way beyond the point of diminishing returns! But doing _no_ upfront inspection is flagrantly reckless.
Also to clarify: the thatcher did in fact work on a time and materials basis. However, I only vaguely alluded to it, and I've improved the wording again. It's interesting though that you assumed it was done fixed-cost-and-scope - I suspect that is the default situation people assume.
I'm interested by your suggestion of using professional indemnity insurance - how would the thatcher have used this? Can it be done in a win-win, none-confrontational way?
Finally, while I've included enough detail to describe a "full" project, the problem I wanted to emphasise was how (given both parties realise they are partially through a project of questionable value for either side) do they to resolve it in a mutually satisfactory way?
No matter how experienced the thatcher, or how carefully he manages the risk, there will always be a project lurking out there he had no way to prepare for. Improving at your job means being caught out by more difficult problems, as the easy ones you handle almost intuitively.