May I ask where you stand on the underlying issue of what Bitcoin's design goal should be? Do you have an opinion on whether it (on-chain transactions) should be an expensive-to-use settlement network or a cheap-to-use payment network? Personally I feel the latter case - direct mass adoption - would be better if it can be technically supported, which I'm not sure about. If mass adoption is not possible I'm left doubting a settlement network alone is viable. That's my current feel on the matter, I'd be interested to hear what you think.
Whatever path Bitcoin takes, it would be best for everyone if it remains a "censorship-proof" value transfer mechanism. Anything it can do on top of that is icing on the cake.
All transactions do not need to be censorship-proof in a world where censorship-proof transactions exist. The
possibility of censorship-proof transactions by itself will cut down on censorship because it will be seen as futile.
Bitcoin transactions, by their nature, are more valuable than VISA or PayPal transactions, so they
should cost more. Most "bitcoiners" (including myself) have been spoiled by the early days of high block subsidy, low cost transactions. Ultimately, you get what you pay for, and at some point we need to start paying for the utility that Bitcoin provides. I would prefer we pay directly so we know what we are getting.
If Bitcoin can scale while preserving it's ability to provide censorship-proof transactions, fantastic. Part of that promise is in the ability to easily share information between peers. It seems obvious to me that the more information shared, the more difficult it becomes. There are plenty of complaints about the size of the block chain already. I'm not sure if taking off the training wheels is appropriate until we better know what the repercussions will be.
We should be prepared for a future where information becomes more difficult to share. This is the future where Bitcoin's utility value will increase and we should make sure that it can reliably function in such a future.