I commented this earlier and it was instantly downvoted for some reason, but I totally agree with both points; I also think they are intrinsically connected in many ways because using customer-development to understand and build an initial audience often helps identify distribution channels. Here are a few examples I have found

1) I made a product to help people manage their diets; crowded space so it was hard to find users, following Steve Blank's customer-development process I found that the product should have been made for Dietitians not for end users. This may have been obvious to most but it was but to me at the time.

2) I made a product for Medicare/Medicaid but it was hard to get my foot in the door to even be considered for any of the Gov. contracts, so following the same model again I found that if I used politicians that had made promises during their campaign that my product could help them to fulfill then I could use them as distribution channels.

One thing I disagree with or don't understand is

> ship earlier now (often free and open source) to learn faster, but it doesn’t change the attention dynamics much.

I have found the opposite, it seems like the bar is much higher now and not even the boost in productivity that I have gotten from using AI has allowed me to ship secure/high quality products as fast as I would like to; perhaps I am just too insecure about my work but its all new I guess.