I’ve been thinking about this as well. You and Cedric made me think about confirmation bias more, and I ended up reading @Jared_Peterson 's substack post on biases not being a real thing.
To paraphrase (at my own peril), bias is a what, and we get in trouble when we treat it as a how. Quoting Jared (emphasis mine):
But most of it is post-hoc; identifying errors and then describing the error as a bias as opposed to focusing on the underlying process which caused the mistake
Treating confirmation bias as a how means you need to treat your existing knowledge as a liability that leads you astray. We see this in Furst’s advice to avoid having any hypothesis, which is at odds with both the data-frame theory research and empirical reality. Every successful case I’ve found where they practiced HOI’s deliberate innovation started with a product idea! A data-frame flavored version of confirmation bias as part of demand investigation might look something like:
A founder typically has a wealth of experience and intuition about their domain. This is good; they should listen to it. But they are usually novices at finding authentic demand. So they frequently use heuristics like “I believe my idea is great and just need to confirm it with a few other people” and “I can take people’s words at face value when I present them my idea”. Both of these heuristics can generate confirmation bias errors (which lead to frame fixation), but the processes that produces them are quite different.
We can replace them with better heuristics, like “initial ideas rarely find authentic demand so hold them loosely”; “people tend to say they love your idea just to be nice”; and the core idea from HOI “don’t be satisfied until you find a market that can NOT NOT have your solution”. Expert entrepreneurs and salespeople have already discarded the common (but flawed) heuristics above. By focusing on the actual flaws in the process, we retain the upside of the hard won existing knowledge.
To be fair, HOI mostly talks about the underlying hows and recommends starting in a domain you already have expertise in. So, I don’t think it’s using “confirmation bias” to throw the proverbial baby out with the bathwater. But it’s a neat illustration of how short the leap is between “confirmation bias as a how” and “systematically ignoring valuable experience”.