Intuitively it seems like more would be better so that in the case of that engine failing you could use the other engines to push up and maybe light another engine.
Of course, if there's two engines and one fails, it's going to tilt. Same with 3, though to a lesser degree. 4 engines lessen it even more, etc. etc.
Or maybe they can detect a bad center engine on descent (it doesn't light or flame is detectably not the right) and then go to outer engines. Or drop it in the ocean. Hope there aren't people on that one.
The landing F9 first stage is so light even one engine at minimum throttle has a thrust-to-weight ratio greater than one; they can't throttle it down enough to hover. Three lit engines would send it shooting back up rapidly.
Could possibly this explain why there was so much lateral movement of the booster before it cleared the tower?
Launch clamps are released on command of the engine computer - i.e when it detects that sufficient engines are running correctly it commands a clamp release.
Note that Starship Booster's outer engines are actually started using feeds from the launch pad. They need power and pressurized nitrogen from ground equipment to start. The outer engines are not able to restart once off the ground.
So the clamps are required to keep the vehicle in place until all the engines that will start have had a chance to do so. They're also useful if there is an abort during engine start so the vehicle remains secure.
That isn't what it looked like to me when it made what, 3 go arounds?
However the engines were on the whole time, producing >1G of thrust along the direction of the booster. So that force is going to hold the StarShip and the booster together. But you would expect some lateral force from air friction to push the stack apart at some point.
I don't have the answer, but IMO cheap mass to Earth orbit is the biggie that enables a whole bunch of options.
Another idea (not starship specific) is to introduce cement (or similar) into the engine burn to build a landing pad from middair.