Those aren't necessarily inherent flaws, they're bad execution. Ride-sharing _can_ be very profitable, but it's also supposed to be very low-overhead. SV startups don't do low-overhead.
Electric scooters are a little iffier, as they take significantly more capex and there isn't much data on market acceptance (Or, at least, none of which I know.)
Meal-in-a-box is not a bad idea either; new people are showing up every day whining about how, "Adult-ing is hard." The fact that people eventually "graduate" is not an issue so long as more people are being born. It's just not necessarily something Joe Six-Pack is going to buy.
If you've noticed a common theme here, it's that the idea of hyper-scaling can be flawed. The above all have significantly lower threshold for market saturation than they believe, if you ask me.
Ride-sharing can be successful, but may not take over the world. Electric scooters can be successful, but may not take over the world. Meal-in-a-box can be successful, but may not take over the world.
SV gets stuck in the pareto trap of scaling past economies of scale and to the point where that next 20% of your customer base takes 80% of your capital. You can push scaling, but the market only moves so fast.