How Unix spell ran in 64 kB of RAM

# Summary In the 1970s, Douglas McIlroy solved the challenge of fitting a 250kB dictionary into a PDP-11 computer's 64kB RAM for Unix's spell checker by developing a custom compression algorithm that achieved 13.60 bits per word—within 0.03 bits of the theoretical compression limit. His solution combined linguistic stemming to reduce the dictionary to 25,000 words, a Bloom filter for fast lookups, hash compression techniques, and Golomb coding to exploit the geometric distribution of hash code differences. This engineering feat remains unbeaten and exemplifies how analyzing problems from first principles and leveraging mathematical insights can produce elegant solutions within strict resource constraints.