It's pretty ridiculous. It's very obtusely written and clearly cites work from others as the basis for their technique (primarily in description paragraphs 1-6). Then in paragraph 7 they claim "None of the previous research on word recognition has been applied to RSVP."
Squirt.io is another service that does the same thing as Spree, and I can't imagine that it won't soon be in Spritz's crosshairs as well -- especially since in its Acknowledgements section, the first to be credited is "Spritz Inc, the company whose patents are pending."
One thing I'm curious about is whether leaving the code publicly accessible, but in a "commented out" state, ostensibly rendering the program non-functional, is sufficient.
Like, suppose I wrote a small open-source text editor whose default text upon startup is the entirety of "Fifty Shades of Grey." And E.L. James' people contact me and say, "Hey, you can't do that. E.L. James holds the copyright to that text." If I merely comment out the text, rather than wipe it from the source code, have I really fulfilled my responsibility to not infringe on James' copyright?
The difference is copyright versus patents. For copyright, you definitely need to remove it. For patents, I think it was openssh that used to have patented algorithms in the source that you could compile into your binary if you had a license.
I don't like to see them stopping open source versions, but at the same time they have little else to protect their algorithm, which brought some significant advances in the speed reading (RSVP) space that hadn't changed for years.
As screens get smaller with devices like smartwatches, this type of technology becomes even more important and may even be a requirement for reading text longer than a few sentences on such a small screen.
Also, since Github is based in US you might want to look elsewhere for project hosting (assuming that it's an open source project).