Months of our development time went into fixing stuff that should've been working out of the box or adding generic functionality that's not there only because the engine is a first person shooter engine and is built that way. Just two days ago I fixed a core Torque bug with memory management that's been there from the beginning of the century, and noone ever noticed it because there were no projects of AoD's (or DS) scale on build Torque. So, "knowing the engine" in my case is knowing how to fix that rusty old car on a daily basis. And even though I know that life is suffering, I'd prefer if there was less suffering and more game-related code, rather than crutches that keep the engine from falling apart.
"Sooner or later the switch is imminent. Torque is losing the battle, it desperately tries to catch up with heavy weight players, but it can't. Its heritage is dragging it down - starting from the above-mentioned number of bugs and compatibility, to the lack of integrated features: pathfinding library, physics, inverse kinematics, level-editing and world-crafting options, modern rendering engine. All those tools, conveniently integrated and ready to be at at our disposal, like SpeedTree or visual scripting. Then there is optimization: Torque is choking on our amount of resources and FPS count we get frequently dives below 20, which is nearly inappropriate for our nice, but modest graphics (Far Cry 4 gives me 40-60 on ultra settings, a game released just 3 months ago). But the main thing is that Torque essentially is just a code for a shooter game with editors slapped over it later. Unity or UE4 are software development kits, developed by a huge and successful company which makes a huge difference."