LZ4 is just such a nice little algorithm for compression. It's downright simple, and fast, and still works. It's so simple seems like it should have been invented long ago, but it's only 10 years old.
It is derived from the LZ77 family of compression algorithms, which have papers published in 1977 and 1978 and thus US patent expiration around the year 2000. It is understandable that it would take time for discourse, inclusion in educational curriculum, and programmers feeling free to tinker with the concepts to produce useful new art.
It's almost like patents are a bad thing for the progress of science and skilled arts.
PNG was explicitly designed to be unencumbered by patents, and it was released in 1996.
https://en.wikipedia.org/wiki/GIF#Unisys_and_LZW_patent_enfo...
PNG uses deflate, which is based on a combination of Huffman coding and LZSS, which is derived from LZ77. LZ77 encodes repetitions using back references into a sliding window. To my understanding, LZSS also uses a dictionary based approach, but only for finding such repetitions during encoding.
So this does not answer the question if LZ77 was patented or not.
The Wikipedia article about LZ77 does cite a patent (https://en.wikipedia.org/wiki/LZ77_and_LZ78#cite_note-3) and somebody in the talk page mentions that there's been a law suite (https://en.wikipedia.org/wiki/Talk:LZ77_and_LZ78#Patent_cont...). However, quickly scanning over the patent, it seems to discuss an improved algorithm based on LZ77, perhaps it relates to LZS? (Note: not LZSS from above; this is again is a different algorithm)
The LZS page mentions a law suite sounding similar to the one from the talk page, but contains no direct reference to the patent: https://en.wikipedia.org/wiki/Lempel%E2%80%93Ziv%E2%80%93Sta... It also states that the LZS patent was filed in 1993 and expired in 2007 (due to not paying their fees).
http://fastcompression.blogspot.com/2011/05/lz4-explained.ht...