When playing LACs in a multiplayer start, they will be disadvantaged by their opponent KNOWING that they need anti-LAC weaponry. In a regular start, it is very likely they will face opponents with Res 1 and Res 50 or Res 100 sensors, but not Res 20. And if the LACs have their fire controls at the right size, they will have a huge effective range and detection advantage.
I think this is one of the point I made above as well. Information about the enemy are probably more important than what type of ships that you have... and/or the ability to adapt your fleets.
For example...
Do you play a "regular" game using the fore knowledge that scripted AI and the special AI races exist as you build your space civilisation... or do you play as if you believe you are the only intelligent species around in this part of the universe until you encounter something?!?
This is an important question what type of limits you put on yourself.
I don't think using the scripted AI as a base of what is effective to use or not is a good base line either, you should rather think what someone like yourself would do in that situation and depending on that "in game" knowledge you do have about what is out there.
In all of my "regular" games I have always played it that I have NO CLUE what is out there and practically built nothing but patrol ships with some beam weapons to control civilian traffic, more as a police force rather than a real military, before I encountered any hostile species. I don't see why a unified Earth would waste resources on military vessels if they are not needed and there are no external threats and we seem to be alone as a species.
Once a hostile species is contacted my military ships would be built in direct response to what I'm facing and reconnaissance would be priority one and then ships would be built based on that knowledge.
The doctrines should basically always be in this order...
1. Know thy enemy.
2. Find the enemy before they find you.
3. Strike the enemy before they strike you.
If all you are interested in are best optimal fire power per invested resource and research points right now, then you should stick to as small a ship as you can. But larger ships will get you advanced abilities and other important benefits, but it will take you some time for those benefit to show when you compare the resources spent today. This is the same problem you get when you calculate how much resources you spend on military today versus tomorrow. Every resource you instead spend on expanding your economy is going to be worth considerably more in the future spending of military resources.
In short a 1000t ship in any one INSTANT moment are going to be much more effective per resources/science than a 10000t ship if you start at zero and all you are interested in is to gain 10000t worth of combat ships as fast as possible.
What you need to do is find a balance between what you need now and what potential you can get in the future and compare that with the threat environment that you exist in. In the best of worlds you can spend 100% of your economy on further expansion all the time. This is NEVER going to be an exact science since you are never going to have perfect information.
In general what I have found is that smaller ships 150-3000t perform most of the recon and offensive actions. Larger ships act more in a supportive and defensive role. This usually are what turns out to be the most effective use of different types of platforms. You need smaller ships to deliver the offensive long range punch and they also act as your sensor screen so you can attack without ever being detected in the first place. The larger ships hangs back and act as mobile bases for the smaller ships and defends them if it comes to that. You also need larger ship for JP attack and defence, small ships just are too squishy for that. Someone who mainly rely on small FAC are going to find it rather difficult to do JP assaults in a reliable manner against larger ships with large powerful guns and good armour and shields. JP assault and defence might not be a huge thing against AI but against a real live thinking enemy it will.