This is known as thrust lever stagger, and can be a PITA in some aircraft. In most new aircraft the engines have fully digital engine control, but in older ones there can be substantial mechanical linkage between the levers and the engine control system. No two engines are ever exactly the same. So, you have the difference between the engines, and various amounts of slop (for want of a better word) leading to different thrust lever positions for the same power setting.
You also need to realise that an autothrottle moves the thrust levers, but does not itself control the engine power. When the autothrottle wants say, TO/GA, it advances the levers until one engine hits the limit (or target power setting), and then it stops. One of the pilots will then push the lower engine‘s lever forward to match the power. Matching the power up used to be the domain of the flight engineers. On the 767 (other than OGV) up to 30 mm difference between the lever position at the same power setting, was common.
OGV (QF’s last new 767) was different because it had FADEC, a fully digital engine control. It was on all of the 747-400s but not the concurrent 767s. With that system engines would basically talk to each other, and the thrust lever system, and so remained matched as long as the levers were at least in proximity to each other. If you pulled one lever on a 747-400 back about an inch (in the cruise), the engine would initially roll back a bit, and the others would roll up to balance the total power. Then it would decide you were an idiot, and match all of the power settings, leaving that one lever an inch out of alignment. But, if you moved it a bit more it would fall out of the range, and the system would decide you might know what you were doing, and simply follow the throttle. In the 747-400, the engine levers were being moved by the system to match the power being demanded by the FADEC.
Airbus takes this a step further by decoupling thrust lever position with the power for 99% of the time, with the levers simply sitting at various gates, irrespective of the engines’ power settings. One extreme case comes up in alpha protection, in which the thrust levers could be at idle, with the autothrust disengaged, and yet the engines will automatically roll up to TO/GA whilst the levers still sit at the idle position (if it detects an extreme angle of attack, i.e. stall).
There are other differences between various implementations of FADEC. For example on the 747-400 as you rolled on to the runway, and selected your takeoff power, you could literally slam the levers forward, and the thrust would remain symmetrical. The FADECs would ensure that the slowest engine to spool up controlled all of their accelerations, and so ensure the power was symmetrical. The otherwise very automatic 380 did not do this, and quite a bit of care had to be exercised with the initial spool up on the runway (below about 70%), as the engines could spool up at very different rates and lots of differential power at low speeds can cause severe directional control issues.
