Because products almost always fail regardless of whether they were tried by devs or non-devs. I think in reality, a lot more devs tried it than those who'd admit it - i tried too, and failed - several times. But more importantly, i did custom coding for 100+ startups and they
all failed. Over 90% without ever getting the thing out of the door - just by realising that the idea was lunatic halfway through the process, or by running out of money, or by switching to the next shiny object, and the remaining by ending up with zero (literally not a single sale or even a single free user, ever), or extremely little demand. No one ever even repaid back development costs, and the only two who came somewhat close, were spam apps launched on jvzoo.com - a vehicle for aggressive spam marketing where "JVs" push useless nothing mostly to grandmas and children.
It's a struggle because it is meant to be a struggle and i think actual success rate (as in, "made more money than a full time job would make in the same time") are so exceedingly rare, you can do it all your life and will never even meet a person who made it. Basically a myth.
Selection bias makes us perceive the startup thing as a lot more realistic than it is, just because those who failed in their overwhelming majority never speak out, out of shame.