For what reason? You have no clue what transaction demand is going to be like then, and you seem to be suggesting that Bitcoin is going to hit your postulated final userbase of billions of users sometime between mid 2018 and 2019. What's wrong with you, is it only hubris!?
Here's a sensible way to look at it: wait. Wait until capacity is truly unanimously judged to be required, then expand capacity. You can't possibly expect anyone to take you seriously when your long-term engineering decisions are based on clairvoyance, or maybe you were talking to Satoshi through a spiritual medium, tell us lol

I agree with this line of thinking.
Global worldwide adoption by 2019 is an unreasonable expectation. Even the EU's switch to the Euro, which was well publicized and meticulously well planned (despite the obvious problems with their currency model), took some time for "the switch". Shadow EU was tracking/trading for at least a year prior to the switch and there were periods of overlap with the national currencies. Also, of course, that was for a subset of Europe, a decent sized population but not all of the planet.
I honestly don't believe that any current/proposed solutions would truly be able to scale in this period of time without a lot of disruptive "growing pains". Progress on scale and the best approaches to achievement of it should be incremental. Hopefully in advance of the need, based on reasonable projections, not pipe dreams of global domination virtually overnight.
A few years back, maybe 5-10 years, Visa or Mastercard (forget which) suffered an outage on Black Friday (which for those that are not familiar with the U.S. consumer market, is the day after Thanksgiving when virtually all retail businesses have begun big sales for Christmas -- it is the largest day of year for transaction volumes on payment cards here). That outage was a BIG DEAL and cost retailers many, many sales. Since then the card acquiring (merchant transaction) system has undergone significant upgrades including multiple global command centers that are highly secured and fortified installations. (Which, incidentally, highlights the advantages of distributed ledger blockchains over those centralized legacy payment systems.)
Incremental, careful, well-planned, and highly tested progress is the best way, in my view, to develop a payment system that can truly be resilient for global adoption as that occurs over time. Problems will happen, approaches that seem like good ideas will end up having blocking problems or unforeseen X-factors. Going too fast and getting it wrong will damage the reputation of Bitcoin and potentially cryptocurrencies in general, given that BTC is the dominant and incumbent.
Outages that ground our air traffic system, such as one last week, are damaging to commerce but do not grind it to a halt. They are manageable though perhaps alarming from a human safety perspective. Outages that affect commerce as a whole, due to either bugs or limitations that did not seem apparent on the drawing board, would kill Bitcoin or any other technology that was deemed to be the cause of such outage.
Slow and steady wins the race. Resilience requires deliberation and careful execution. Progress may be slower than many would like, but there's a very good reason to be careful when you are dealing with a store of value. There is little more in our societies that have to be trusted more than a means of exchange. Bitcoin enables "trustless" commerce, but only if Bitcoin itself is highly trusted, with the track record to truly earn and keep that trust.