The reMarkable 2 is an e-ink tablet that aims to replace traditional paper. It's primarily marketed towards students and doubles as a traditional e-reader. While the tablet and company have both had many shortcomings, I'm very impressed with the overall writing experience.
If you've owned a kindle or any other e-ink device, you know the displays are nothing like the common LCD displays we have today. The reMarkable 2's display is a step beyond the Kindle's — the typical e-ink display flicker is present, although is (from what I can tell) much less intrusive. The writing latency, while detectable, doesn't bother me. The screen is layered with actual glass and comes with a thin textured layer that gives the surface a paper-like feel. It doesn't fully emulate paper, but it feels far better than smooth glass.
The rM 2's hardware is great. It looks and feels like a piece of art. The writing experience is the best I've had on any device. The community is thriving and constantly producing alternative software packages.
I mentioned the snappy writing experience — well it seems they purposefully avoided filtering the writing input to avoid adding latency. This gives the stock writing experience slightly jagged lines (to varying degrees between units). The effect isn't very noticeable for short, quick strokes (normal writing). It is, however, very apparent for long and straight lines. The company's response to customer complaints on the matter has been disappointing to say the least. Thankfully there is an active community of open-source developers for both the rM 1 and rM 2. There's currently a simple fix that uses the LD_PRELOAD trick to intercept read calls to the input device. The fix adds a small amount of latency (depending on the N value provided), but it's hardly noticeable.
Another shortcoming is the default reading software included. It's slow to load large files and it's missing many features (e.g. bookmarks). Many users of the rM 1 use another reader called KOReader. Early support for KOReader on the rM 2 is being worked on. The rM 2's framebuffer doesn't use the E-paper Display Controller (EPDC) like the rM 1. Instead the waveform is written to the framebuffer through closed-source software. There is an ongoing project to establish a simple API for drawing to the rM 2's screen. This project uses the same LD_PRELOAD trick mentioned previously.
The rM 2 isn't for everyone. It's expensive (currently $450 with the standard stylus) and comes with mediocre reading software. Note that the screen doesn't include a backlight, so reading in the dark isn't practical. Despite these issues, I'm pleased with the device.