Did you actually read the blog?
Yes, the Starship is suboptimal for Mars missions, but it technically can work.
The crucial thing is that Elon Musk already knows this.
Mars colonisation is just the marketing gimmick in the future.
The first versions of Starship will be focused on purely lowering LEO launch costs to $100/kg. That will be a game changer.
Afterwards, they'll likely develop different Starship versions or a new vehicle for Earth-Mars and/or Earth-Moon transits.
I cited that blog only because it was a source, not because I agree 100% with it. In reality from my own engineering judgement (which you may rightfully regard as suspect but if you look at my contribution to the semiconductor thread, it may be at least worth considering) I think the Starship has several huge flaws:
1. The bellyflop landing maneuver is bad because unlike every other reentry vehicle it exposes the maximum cross section to the heat load rather than the minimum cross section. This means greater total area heated and higher total chance of failure, given equal failure chance per heat shield element.
2. The bellyflop landing is also bad because of the complex aerodynamic forces with an oddly shaped cross section in the airstream that has very minimal control surfaces.
3. The landing is still powered which means need to keep fuel reserves. But the last unit of fuel is more payload deltaV than the first unit of fuel since it doesn't have to push other fuel.
4. The fuel tank design with shared fuel and oxidizer walls is a bad choice. The shared wall is weaker in the convex direction. 2 concave walls in a non shared configuration is stronger but wouldn't meet his requirements.
5. From previous launch footage, there are multiple engine restart failures, presence of green flames (showing metal vaporization), outright explosions, etc.
6. It has worked only once in a 10 km (10000 m) single stage landing attempt. This is not even close to a full reentry test which so far more complicated, and even a full reentry is far simpler than a full launch. Historical precedent is against it.
