Let me be clear: Any player who refuses to recognize the value of a weapon system being deployed against them can and will likely be defeated, justifiably. Walking into the open with your Atlas like you can tank the whole team is a bad idea - regardless of what weapons are being used on you.
That said - the complaints with LRMs are as follows: Behavior is too inconsistent with the effect of the game in canon battletech for the weapon values to be reconciled. Spread, damage, and homing characteristics are nearly impossible to develop so that there is a relatively comparable performance across all target classes (light mechs will be next to invulnerable unless you turn missiles into assault mech coring nightmares).
The problem stems from LRM behavior.
LRMs do not behave in a manner comparable to real missiles. Real missiles do not "follow" the target. They intercept the target, detonating a claymore-like warhead to create a cone of shrapnel that the target flies into (in air-to-air missiles). This is reminiscent of the random clustering effect of the table-top game.
So - the best way to balance LRMs is to implement behavior that is a logical extension of real missile behavior into the battletech universe (well - assuming that missiles like the Javelin are ineffective at hitting a mech-like target and a clustering behavior is more ideal).
Thus, I present my humble attempts at an artistic engineering depiction with free-hand drawings and some supporting resources to further illustrate the point:

The key change is that missiles do two things - project a ballistic intercept trajectory ("shoot where the target will be instead of where it is"), and that they will have no guidance once the volley spreads into a cone.
LRMs would fly 'straight' for the first 35% of the distance to the target (as it was during launch) at a 45 degree angle before they are fed a ballistic intercept trajectory. This ballistic intercept is updated periodically before a terminal guidance phase that puts the missiles on roughly a 35 degree approach (relative to the map's XY plane). 45 meters from the target, accurate guidance to the target is assumed impossible and the missile swarm scatters across a cone of 10-15 degrees (which computes to a cone with a diameter of about 16 meters to 24 meters).
For an example of how real missile systems work:
The key point to carry across, here, is that we are replacing one missile with up to twenty. The reason for this is the presumed difficulty of achieving high PK (probability kill) with a single missile system. Since you are striking an armored target - fragmentation warheads would be useless (as compared to unarmored aircraft). Thus, a swarm of missiles are used, and each missile carries a HEAT warhead - the cluster scattering to improve the odds of hitting a maneuvering target.
Naturally, smaller mechs will be less likely to receive damage while larger mechs will be more likely to receive damage, but distributed. Effectiveness against maneuvering mechs would be improved while not creating a hopeless scenario for larger mechs. Cover would still be effective, but simply running crazy through the open is a strategy available to only the fastest of mechs.
There is another reason for this suggested change - the number of factors available for balance. The range at which the missiles scatter, the degree to which they scatter, the damage they do, their splash effects (if any), velocity, and numerous characteristics about their intercept path can be tweaked separately to a honed form that allows for good gameplay across a wide range of the community (some players will just not get the idea of cover).
It should be noted that I have not properly accounted for TAG and NARC effects - TAG should give roughly a 12 degree cone at 45 meters - not quite as effective as Artemis (the two should not stack for gameplay reasons), but transferable to other systems.
NARC should allow for "fire and forget" functionality once a target lock is acquired. Lock - fire - lend attention to something else. NARC beacons should also last indefinitely until the target has taken between 40 and 80 points of damage (with the probability going from 10% to 100% incrimentally). But that is a bit of a different debate.
Further, I have attempted to track down mathematical references for how to go about computing ballistic intercepts with source code. In the event developers (and/or community) see this as a good idea, they aren't left to try and figure it all out on their own (we can't expect the developers to simply know how to make any and all code that could ever exist for any process under the sun).
The best source is this: http://www.codeproje...-Guidance-Syste
Not only is that open source code, it's also a functional illustration of the mathematics involved.
For those whose interest has been piqued:
http://en.wikipedia....issile_guidance
http://www.fas.org/m.../fun/part15.htm
http://www.fas.org/m.../fun/part19.htm
Edit: ... For whatever reason - the poll didn't take like I originally structured it...
http://www.fas.org/n...sile/basics.htm - This last one is mostly in regards to ballistic missiles rather than interception missiles.
Edited by Aim64C, 29 March 2013 - 02:35 PM.
Changed poll answers to follow the rules