If you have something that changes from 0 to 120000 gradually and you set it to cut off at 266, everything from 267 to 120000 would return 266, which in your formula is 0 due to linearConversion.

Not at all. In my case it will return values from 1 to 0. 1 for raw values from 3 to 0, and 0 for values from 150 and higher. There will be no jumps in return values at all. Take a closer look at the example above.
But it doesn't matter at all, it is just a use case. I guess I gave that as an example for nothing :)

max and min would create hard limit that would seem like a jump

No, it is not supposed there could be such jumps. I tested in a sequence of thousands iterations, and linearconversion works perfectly on both dedicated and local machines, the result is absolutely predictable for specified values.
But let's forget about the linearConversion. The main question - why the getLighting param jumps?

In T170544#2411210, @BIS_fnc_KK wrote:

as wiki shows second element can go up to 28526.2, if you limit it to 166 it is expected it will seem as instant jump

Usually I don't use laser-guided projectiles for AI, as bots are not good with it.
It can't attack even with fire-and-forget missiles, like Mavericks, which has a heat guidance system.

Hm... What LG armaments the AI use in this case? I tested all of the available options, with reveal command - and AI even can't open the bomb bay.

In T150733#2068492, @Wulf wrote:

Hello.

I need to know on which map you created the mission on.

Yes, SP test in editor