[jofer]

No boreholes are perfectly straight and vertical. In most cases, you're twisting the drillstring at the top, so it naturally "wants" to corkscrew. Furthermore, anisotropy in the rock causes things to deflect (deviate in drilling terminology). Usually you don't know exactly where the well went until you run a directional survey afterwards (basically, you lower a gyroscope down the hole).

At the scale of a well, the steel drillstring is more like cooked spaghetti. You don't "push" down, you use the weight of fluid (drilling mud) for pressure. The drillstring is only meant to transmit torque. It's easy for it to bend over long distances. (You can even turn things entirely around and use fluid pressure to turn the bit, in which case the drillstring is basically a steel hose that's stored coiled up.)

[scrumper]

Thank you, really interesting.

Is the drilling mud pumped down to create that pressure then, or is it naturally present? Or one pumps water down to turn the grit into a slurry?

[jofer]

It is pumped, but the pumps are there to force circulation and keep things under control rather than to provide the pressure directly.

You need a steady gradient of pressure. The pressure you need at the bottom of the hole will fracture the rock in the shallow section. Pumps add a constant pressure, so there's no way to get the right pressures at each depth with a pump.

Instead, drilling exploits gravity to do the hard work. The drilling mud is very dense (almost as dense as the rock) so that the pressure at the bottom of the borehole from the weight of the mud is just a bit less than the pressure from the weight of the rocks.

[ramgorur]

I think even if they had no rock formation or the whole layer was uniform, they would still deflect. It's similar to walking in a straight line blindfolded.