My first thought on looking at comp as a point system is that there is already a point system built into 40k so why do/should we need another?
We have the Force Org Chart that already tells us how many of a certain unit type we can have and each unit is worth X amount of points.
The problem here is GW has stated quite plainly many times that point costs are spitballed and are only considered based on that codex. So, for example, the Heldrake is cheap because the other units have the potential to be kitted out to be ~300pts/unit. You compare it to another book and it makes no sense, but within the context of the single book it is (somewhat) more logical. 250 points of Guardsmen =/= a Land Raider, you know?
In a larger game, point costs are not really any way of telling things apart. The Riptide is expensive so it is prohibitive to some degree to spam the thing. But we don't objectively know how much of the on-table effectiveness the unit has from just its point cost -- it's also how easy or hard it is to use them in a way to affect the game in your favor.
This is what a comp system should rate, IMO. I'd assign every unit in each codex a number 0 to 3. 0's are "those" units that we never see because they are so bad, 1's are units that are decent, 2's are units that are above average, and 3's are units that even having one of can drastically affect the game. I might even whip something along these lines up just to see how it works out. After that you just do what was done in this thread: run power lists through it and see how the numbers come out.
You end up with a bigger number, but it gives you a scale. So people can say "I have a Level [whatever] list" and it will be clear what they are expecting from their opponents.
This also saves some trouble publishing results -- you can see what level a guy's army is right in the results, it will plausibly show what armies have problems (i.e. if the top tier is all guys with maxed out army comp you know there's a power issue, and can correct the scores next time, or if it is a spread, you can show once and for all that the strength of a list isn't worth being concerned about).
You could even use it to sort who will face who.