The limit is there because resources are limited, and no limit would lead to massive node centralization. We have data that shows raising the block size now would not be a good idea. SegWit will buy us some time, then we can start considering a blocksize rise, always having in mind that scaling bitcoin worldwide without an additional layer is impossible without mass centralization on the nodes.
Exactly. The other guy does not know what he is talking about. 'The most influential' Core developers can't do anything on their own. This is very important and I don't think that some realize this. For example, if Maxwell was the only one against a feature and everyone else agrees then it would be implemented. The same happened with Gavin, almost everyone was against his proposal and thus it was rejected. Segwit should provide adequate amount of transaction space for now.
segwit increases full node requirements just as much as 2MB blocks.
if you believe 2MB blocks will lead to node centralization, then you should come to the same conclusion with segwit.
removing the ability to create a full node on a free AWS server should not be seen as a bad thing.