I came here to write something similar: the article does not pay enough attention to changing environments. The dotcom bubble ideas that are now successful do n

I came here to write something similar: the article does not pay enough attention to changing environments. The dotcom bubble ideas that are now successful do not share the same background environment of internet availability, bandwidth, networks of datacenters, storage, powerful portable computers, and all around higher digital literacy and people used to interacting with computers. If these things hadn't changed then trying and trying again would have continued to fail with high probability. As written, the statement by Ford is wrong, it is important to keep a kind of Tabu list detailing what didn't work in the absence of gradient information. What experts get wrong is they fail to be sensitive to dependent parameters that are not static (I know the article says something like this too but it sort of glosses over it). The problem is many of these are hard to say outside of hindsight. Perhaps listing the requirements before launching and listing the state of those requirements after failure, in addition to the idea might form a basis for what to retry. An expert might say something like "A website for dog food and toys will never work, we already did that 20 years ago and it failed spectacularly", when they should be saying "A Website for dog food did not work 20 years ago, here were our requirements and environment, has anything changed?". But even that isn't perfect so the process can't be deterministic.