If I as an individual want to do something with a modern SBC that packaged in software doesn't do Pine64 is the only choice for hardware. Same in wearable tech space (pine watch). Good luck trying to get register level documentation for some parts of a raspberry pi's, or any other sbc's.
I've spent last few weeks working on pine64's hardware at a very low level (mostly rewriting sony's imx219 sensor driver to fit my needs and trying to squeeze out the latency out of the image processor). Specifically I use Soquartz and Quartz64-A (they're very similar).
I went from knowing nothing of the hw and the tool chain and finding every prebuilt image on the Internet as not booting to being able to rewrite the camera driver and tweak the rkISP drivers, generate my own os images etc in few weeks of spare time. This was possible only thanks to the rockchip documentation, the open source BSP kernel (that runs Linux 4.16) and someone at pine64 forums who published how to setup the tool chain. Without pine64 all I'm interested in would require significant reverse engineering effort. Also the mainline kernel support is progressing too.
Is the documentation really good? No, the English versions, save few crucial docs, are quite bad. The Chinese versions are better, but still they leave a lot of important stuff out. Is pine64 making only good decisions as a business? No way. Is the community vibrant? If you can call 2 guys doing all replying to questions on the forum than maybe.
Still, if we want truly hackable hardware to exist in future we should support their efforts. Let's explain what they do wrong, but let's not turn our backs on them completely.