However, strictly speaking, you're 'reasserting' control going down the scale. So, if on 23 the drone does nothing because you have remote control, then on 17 you release control, the drone still does nothing on 13 as it is assuming the new, lower control mode of autopilot. On 7, it's eligible to act but it does nothing, since it's no longer remote controlled and this is not its initiative. The drone will finally act on 3, so it doesn't seem like you can really overly abuse it to out-action a Sammy.
This is a bit confusing. Why isn't it eligible to act on 13 if the Decker used a free action to release it on 17? (I'm assuming both of these are IP#1) We already established that a drone is always in one of the control modes. Either it's under RC or it's Autopilot on 13, but it can't be neither.
Well the paralytic side effect only lasts until after the IP that you hijack control. If the opposing side re-asserts control (not using a manner higher in the control override hierarchy) before then, that's the 'lockdown'. You effectively have 2 people continuously struggling to control the drone; it makes perfect sense that it does nothing productive until the struggle is over. As for flying drones, I don't think there's really too much at stake (unless the struggle goes on for a ridiculous amount of time). Rotorcraft can autogyro down even in the event of engine failure, I don't see a barrage of confusing data coming in keeping the onboard computers from staying aloft. Noone is able to direct it to crash, after all, while the struggle is going on. If you assert remote control, then neglect to give it any instructions/control device actions, yeah, it probably will eventually crash. But that's what should happen.
That can't possibly be how it works. Since no enemy hacker is ever going to deliberately relinquish control and would have to to expend NO actions to maintain his claim to Remote Control on the drone, then by executing one Control Device action he's effectively and effortlessly locked out anyone without a Control Rig until a reboot.
It's worth pointing out that although we describe the competing commands as "struggling" for control, there's really no opposed tests here, so as soon as there is ANY control conflict at the RC level, the drone is effectively out of commission for 2+ turns.
Let me illustrate the logical consequences of this interpretation. (i.e. This is what I'd do as an enemy Hacker.)
Assumption: the Drone in question is not used by a Rigger.
End of Turn A - Enemy hacker uses control device to make the drone shoot your allies.
Sometime in turn B - Owner commands it to reboot (hoping the enemy hacker hasn't used Format Device on it)
End of Turn C - Drone comes back online with Autopilot in charge.
Result: The drone is out of action for 2 full rounds (B and C), I made the owner waste 1 Complex Action telling it to reboot, I may have made the Drone lose an action at the end of Turn A, and I got it to shoot one of his own allies. That's all nasty,
but here's the icing on the cake - while all that's happening, I can be causing other mayhem for 2 full rounds (B and C) without having to do anything else against that drone, and I can be assured that it's going to be paralyzed for at least 2 rounds.
On the other hand, if we say that Autopilot reengages automatically on the next IP, then I'd have to continuously spend RC actions each turn to keep the drone under my control, or at least not under it's owners control. That seems a lot more balanced and takes away the effortless lockdown trick.
Example:
IP#1
23 - Drone Autopilot autonomous action
17 - Enemy Hacker uses Control Device on Drone
IP#2
13 - Drone Autopilot superseded by Remote Control that was just asserted
7 - Enemy Hacker takes action other than Control Device, therefore RC is released, Autopilot in charge (but not its' turn yet)
IP#3
3 - Drone Autopilot takes autonomous actions again