The concept of lazy loading is an old one in the context of software programming. The idea is that you only do an expensive operation at the last possible moment, as to not do a bunch of expensive, slow, stuff unless it's absolutely necessary. As an example, you might only instanciate list objects when they become visible to a user as they scroll through a list of items. Creating every list item when the user might not ever scroll to 99% of them to see them, is a waste.
This is starting to apply in the user experience across many online services/products I use. The "expensive" operation is perceived as registration or the act of login in with a username/password, and those operations are being pushed down the line of the user experience with the product. Guess what; if I want to use the product, I don't mind logging in.
I'm seeing an unfortunate trend in the online user experience around "login." Services like Amazon, and eBay, are starting to let me do a lot of things on their sites without having to login. Things that have traditionally required a login. For example, I just added an item to a "watch list" on eBay, and it was added to a "Guest Watch List" even though I'm an eBay member. I had to login and convert the watch from a Guest Watch to my account. It was a confusing user experience.
Another way of putting the example is letting me get knee deep into functionality without logging in, then at the last possible moment, asking me to login.
There's a fine line here, and it's being crossed. We're also probably on the verge of some new
UX paradigms, so I may just be griping because change is coming.
Someone once said "if there's gold in the mountain, people will find it." I'm concerned products are becoming less and less useful, and are therefore driving their creators to make them easier and easier to use. They then blame the usability of the product for the lack of uptake. The problem generally lies in the underlying value folks. If you find yourself lazy loading the value of something, in order to become easier to use, you've probably got a problem.
As for the big companies I mentioned doing this, they're off in the weeds messing around with a paradigm that doesn't belong where they've applied it.