I really hate to be dismissive, but I think you're thinking about this like a Battletech player, where lights had literally no ability to stand up to anything heavier than themselves and had no real 'on-table' role when taken as part of a BV-balanced force. That was a problem with Battletech itself, NOT with the idea of light units: the rules themselves simply don't favor units which can't take at least 10 points of damage in a single location, and the typical size of a Battletech game (4-6 mapsheets) just didn't allow the strength of lights to reveal themselves in a game of maneuver.
If you hate to be dismissive, then don't be. This game is still battletech, and there is no reason a unit that costs less than a third of another should be able to kill or nearly kill another unit by itself. The current point value of light mechs completely negates assaults and many heavies.
I want to make this clear, though: The main reason I'm being dismissive is that I spent some 6 months with a group of cutthroat players trying as hard as I could to break my own points system and found that speed HELPS, but doesn't WIN the game. It's an issue that I had myself so I examined, dissected, and tested the hell out of it in many different permutations with the help of some real bastards. The results were conclusive: if you pay too much for speed, it does nothing to help game balance. Speed is not armor, ARMOR is armor.
Armor is armor, and speed is an armor multiplier. It is not currently treated as such by the points calculator. If a unit is hit 40% less than the baseline then it's armor is worth 40% more per point than baseline.
Think about the roles that a light unit is supposed to play. On the tabletop, they SHOULD are skirmishers, screeners, and flankers - while they may not be able to fight a heavier unit head-on, if that heavier unit doesn't have screeners of their own then they should be punished for it by being pounded from out of range or by simply being outmaneuvered.
Light units absolutely can fight a heavier unit head on. Especially in the higher tiers of movement mod. With 26" of movement, you can literally keep yourself at long range until you win initiative and then close to short range BEHIND your target. The current inexpensiveness of lights means that there's no down side to taking them. The most game breaking army you could make would consist entirely of +4 and +5 to-hit mod units even if you were under points of your opponent.
Don't even try and tell me "well you can stand still and take a -2, or you could use arty, or you could use precision ammo." 1) Standing still means that you're an easier target as well, which if you're being swarmed, you're probably going to die. 2) Arty may not care about movement modifiers, but you still need to hit the spot you shot at, and any sort of arty that can be mounted on a mech and hit the same turn it's fired has a 2" diameter blast radius. That means that if you miss by even the minimum value, your shot will still scatter beyond damaging the "target" mech. 3) Precision ammo is not available to clan players at all.
At first, the tactic you're describing does seem to work out fine - the ability to peck at heavier 'Mechs from 30"+ away and never get hit by anything in return is impressive, for a given value of 'impressive' (I don't call taking 3-4 turns getting through an Atlas's armor for a single structure hit impressive). However, what's less impressive is when scout hunters or even other scouts close the distance, when artillery gets to within 31", or even a cluster bomb dropped from a quick overhead pass takes care of the issue.
It's not 'when' arty gets within 31", it's
if. Then you go on to say that I need to field assets other than mechs to counter light mechs? Why would I ever bring a heavy or assault if the only viable counter to light mechs is aerospace assets?
If my opponent were foolish enough to take NOTHING but slow line 'Mechs, then there's a problem (on his end, not mine). But the use of even a bit of combined arms and there's no issue whatsoever.
So if I were to take a star of 3 lights and 2 mediums, and a star of 1 medium, 3 heavies, and 1 assault against a company of 5 lights, 2 mediums, 3 heavies, and 2 assaults that's foolish? Or is the foolish part taking clan units whose light mechs are generally slower than their inner sphere counterparts? Speed is too cheap, it allows for viable min-maxing right now.
I think part of the problem is that you aren't thinking in terms of a battlefield, but of raw numbers. If a Skill 3 Atlas sat in woods and fired at a group of Fire Moth-Cs at long range in woods then yes, eventually the Atlas would come off worse unless it got really lucky. But why is it just sitting there?
I'm thinking in terms of having played out this scenario changing one variable at a time in a controlled manner. High to-hit modifiers are waaaaay under valued. Also, the Atlas is probably sitting in the woods because if it closes with the Fire Moth C's, it's going to be flanked and torn apart, while being in the woods at least gives it some cover bonus as it stands still to have a prayer at hitting the little buggers (which, hilariously enough, is exactly what people who favor the light armies keep telling heavy players to do as the counter).