Make all the corrections to a new Bitcoin and use that one. Their are enough potential and known problems with Bitcoin to start over instead of continuing to patch on top of a patch of poorly written original code. If you read the entire article even Gavin said these changes won't be trivial. If you're going to quote something quote Satoshi saying the blocksize needs to be changed in this way. If I believe Gavin is relaying Satoshi's ideas correctly, then the problem was DDoSing the system. That's impossible with the 7tps rate but will that problem return when the blocksize increases?
No, it won't return. Your first post is wrong. The block size isn't something fundamental to Bitcoin while the number of coins and block generation time are.
The block size was changed previously, however it was lowered because there was too much headroom and people could try DDoSing because the network was unable to handle 30MB blocks now.
Gavin's initial calculations are flawed, hence the 20MB fork is FUD.
I've also learned this just recently: It should be 8MB (with corrected calculations), and DDoS isn't really possible at that block size. If I'm correct Luke (?) and Maxwell said that because of the increase they would not be able to run nodes anymore due to their internet. They've concluded from this that the majority won't be able to run nodes either. This is also wrong. There are plenty of places that offer sufficient internet speeds for 8MB blocks. Although their argument is also invalid due to the progressing block size. We might hit 8(originally 20)MB blocks in 2016, or in 2026/2xxx. Nobody can really know when.
I see a lot of people mentioning Satoshi. Please stop. He had ideas however his implementations weren't great (one could say that they weren't even good). He did not code Bitcoin properly as the initial code was quite a mess. Obviously he didn't foresee sidechains, neither did anyone if you back up 5 or more years (prior to Bitcoin). I'm more interested in the actual disadvantages of the lightning network. If there are none, why aren't we proceeding in that direction?
However I do not see much wrong in increasing the block size to let's say 4MB. That would be some sort of compromise for Gavin and the other developers which would also give the whole network more time for better solutions.