Koniving, on 31 December 2012 - 09:50 AM, said:
I agree with the first three posts for a few reasons. None of which even need to be mentioned. However take a peek at what I've got below.
"The Guardian is typically used to shield allied units from such equipment by emitting a broad-band signal meant to confuse radar, infrared, ultraviolet, magscan and sonar sensors."
(↑ 2.0 2.1 Technical Readout: 3050, p. 197)
Okay so ECM shields nearby mechs with broad-band signals. It confuses sensors and advanced equipment. Therefore, if you're in the bubble the effect it has then expect to have a full force jamming you. That's fine. If you're outside the bubble then scanning into said bubble should be difficult. But not nil and void. That's what stealth armor is for to protect you from being scanned. In the mean time if you can see them then you should at least be able to get a confusing signal hit to team mates. One that flickers and is hard to target, but still visible.
After all it is stated to "Confuse" sensors. Not completely dismiss them.
"Sensors can sometimes override this jamming, though by that point the enemy unit is already within visual range and can track the opposition with their own eyes." (↑ 2.0 2.1 Technical Readout: 3050, p. 197)
I can visually see mechs moving at about 1,000 meters out or more. I'd say within 500 meters should be enough to cut through jamming with sensors on a direct line of sight. May not be able to lock missiles on them, but at the very least we should get a sensor node to track them with and share location info to other team mates. Tag may not have needed its boost if ECM actually confused sensors instead of dismissing them.
So, say 500 meters and the mech is in line of sight. The jamming should already be overwritten on sensors. Worst case scenario, I should be able to have at least 5 targets for 1 ECM-equipped mech, and have to sift through the false targets to fire upon the real thing with LRMs.
Now this is where tag would come in at 450 meters by helping to isolate the correct target so that team mates are not firing on dummy sensor targets created by the ECM-equipped mech.
Voila. Tag just got cooler without a buff, ECM just got reasonably more balanced, and we're all set.
Though I still agree with many of the other posters, there's absolutely no reason to carry NARC. I haven't found it even remotely useful as a scout or missile mech. Once an ECM comes along, no one can see my target. What's the point of using it?
------
Before that epiphany, however, this was my original idea for how to balance ECM. Changing which mech variants carry them to the ones with the LEAST missile slots.
http://mwomercs.com/...43#entry1679943
(And no ECM didn't drive me to Hawken, that's some whiner's thread. However my post does provide some good insight in simple changes to balance the game without changing anything about ECM.)
Whiner here.
In the BT books, many components have both their mechanics, and some descriptive text -- "fluff."
My version of TR:3050 reads differently than what you've quoted. That's not really surprising; that book was revised at least twice that I'm aware of, and probably more than that. That said, your quote reads like fluff, and not mechanics.
Likewise, Sarna pretty much copies the fluff. They don't always reproduce the mechanics. I can only assume this is for legal reasons.
In this case, and others before it, PGI seems to have missed out on the mechanics and just read the fluff. So they took the lines about "sensors" and applied it to all sensors, instead of just the add-ons that were explicitly listed in the TT rules.
Thing is, the TT rules on ECM were pretty good, and would probably carry over pretty well. It was balanced there, and would likely be balanced in a straight carry-over, here.
A sane beginning for ECM in MW:O would have been just to counter/disrupt/negate Artemis, Beagle and Narc. At the same time, they should have improved Beagle & Narc so that those systems were useful and desirable. If ECM did not feel useful enough as a counter to Artemis, Beagle and Narc, after those systems were brought up to spec, then they could have added some juice to ECM a little at a time. But, instead, we see that PGI favours a cluster-f*** at first, and perhaps - maybe - they'll tone it down to something sane in time.