I'm interested in this statistic, do you have a source?
Are you saying that 90% of all buy flows are abandoned at the payment stage while x% are abandoned at another stage and <10% are completed or are you saying that 90% of all abandoned buy flows were terminated at this stage? In case of the latter it seems like knowing what percent of all buy flows become abandoned would be relevant to your point.
Also, was this data collected from one vendor/industry or from multiple sources? I'm also curious if the study examined cases where the "pay now" screen already had the customer's payment information (e.g. for a pre-existing customer).
I bought some electronics today; I ended up almost placing an order at three different distributors until I found the one I wanted.
It varies quite a lot based on how your individual buy flow is set up. I personally prefer to be up-front about pricing with people; it skews your metrics if you make them get to the credit card entry page just to see a price. Regardless, if we're talking a SaaS buy flow the conversion rates are usually far worse: your best case scenario is that 3% of customers who sign up for a free trial will convert.
Honestly, this is a case where A/B testing helps. Build both buy flows, measure the outcomes you care about, and go with the one that produces the result you're looking for. This assumes you have sufficient volume for A/B testing: if you don't, you shouldn't be focusing on optimizing your buy flow, you should be focusing on traffic acquisition :)
Such scheme being HTTPS? Unless you are speaking about "stolen content" which basically boils down to the fact that copy costs nothing and nothing in the world will solve this.
Now, you can remove the friction by devising some automated system that pays sites some amount of money based on how long you spend there, but then you have to build an entire fraud-detection system to make sure people don't try to game the system and take more money than they are owed. This would add significant transaction costs -- and these costs would be high enough that very small transactions would no longer make sense to process. Keep going down this road and you'll eventually realize that you're just another credit card payment processor and have a similar cost structure. Bitcoin solves none of this, btw - you still want to make sure whatever micropayment scheme you use doesn't hand over your Bitcoin to a fraudster.
So now we're back at the user being the "filter" for good vs. bad transactions by making them explicitly authorize those transactions. And the friction problem you describe is very real: something like 90% of buy flows are abandoned at the credit card entry page. Micropayments are an intractable problem in my mind; they would solve a lot of business model problems but I don't know that they're very attractive to consumers.