Having a bunch of "free space" that's unused due the cap being too big brings a lot of problems and possible exploits. What centralizes Bitcoin is not being able to run a node in your computer, not the nonsense you are talking about. Bitcoin will never scale to worldwide levels by raising blocksize.
There is no free space. Damnit you guys really have no clue about the blocksize debate and comment on it.
The cap limit is not a default setting.
Blocks wont become 8mb size, and 7.5 mb will become empty.
Blocks will remain the same 0.5mb, but it just lets the maxium reach 8mb if it needs to.
Now i dont think 8mb should be increased immediately, but it should be written into the protocol to increase at some point.
I dont like the command & control aspect of the development. There should not be a need for development anymore. Add a final BIP into it, and then its over.
Developing it further just gives more power to the developers and it creates a centrally planned currency, that is nothing better than fiat.
I know there's no "default setting" and free space is just a way so say it, but yes having that "extra space" brings some problems.
In any case 8mb or 20 solves nothing in the long term. Right now those values are absurd, in the future, we'll see how the average internet connection/hardware deals with those, but the main point is, Bitcoin will never scale to worldwide levels without something like the Lightning Network.
At some point software solidifies, protocols need to stop being exposed to hard forks, and 1mb seem to be enough (+ sigwit) until we get LN. Only people clueless in scaling software of this nature would claim that doing hard forks periodically is a good idea. TCP/IP solidified and then we developed layers on top, the same will happen with Bitcoin. I don't care if it happens with 1mb or 2 or 8, in the long run we will encounter the same problem and only LN can save our asses if we want to go global, something big block guys don't seem to get (unless they are ok with the nightmare of huge datacenters running nodes + the aforementioned problem of periodic hard forks).